Students for Fair Admissions, Inc. v. President and Fellows of Harvard College et al

Filing 628

Response by President and Fellows of Harvard College to 620 Proposed Findings of Fact . (Waxman, Seth)

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 1 of 78

UNITED STATES DISTRICT COURT FOR
THE DISTRICT OF MASSACHUSETTS
BOSTON DIVISION
STUDENTS FOR FAIR ADMISSIONS, INC.,
Plaintiff,

Civil Action No. 1:14-cv-14176-ADB

v.
PRESIDENT AND FELLOWS OF HARVARD
COLLEGE (HARVARD CORPORATION),
Defendant.

HARVARD’S RESPONSE TO SFFA’S PROPOSED
FINDINGS OF FACT AND CONCLUSIONS OF LAW

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 2 of 78

TABLE OF CONTENTS

INTRODUCTION .......................................................................................................................... 1 
ARGUMENT .................................................................................................................................. 5 
I.  HARVARD DOES NOT DISCRIMINATE AGAINST ASIAN-AMERICAN APPLICANTS ....................... 6 
A.  SFFA Bears The Burden To Prove Intentional Discrimination .......................................... 6 
B.  SFFA Invokes An Inapplicable Legal Theory Because It Cannot Prevail Under The
Theory It Has Long Advanced ............................................................................................ 7 
i.  Teamsters does not apply .............................................................................................. 8 
ii.  SFFA’s evidence is just as inadequate under Teamsters ............................................ 10 
C.  SFFA’s Theory Of Discrimination Is Incoherent ............................................................. 15 
D.  The Statistical Evidence Does Not Show Intentional Discrimination .............................. 16 
i.  SFFA’s descriptive analysis relies on the incorrect view that grades and test scores
drive the admissions process ....................................................................................... 17 
ii.  SFFA’s regression analysis is deeply flawed.............................................................. 20 
a)  Dr. Arcidiacono’s regression analysis of the preliminary overall rating is not
probative of discrimination in admissions outcomes ............................................ 22 
b)  Dr. Arcidiacono was wrong to exclude the personal rating .................................. 23 
(1)  The personal rating is not affected by race ..................................................... 23 
(2)  Even if the personal rating were affected by race, the right approach would
not be to exclude it .......................................................................................... 27 
c)  Dr. Arcidiacono was wrong to exclude ALDC applicants ................................... 29 
d)  SFFA’s other statistical arguments fail ................................................................. 32 
E.  The Non-Statistical Evidence Does Not Show Discrimination ........................................ 35 
i.  Harvard takes pains to ensure that its consideration of race is not used to
discriminate ................................................................................................................. 37 
ii.  SFFA’s allegations of stereotyping are unsupported .................................................. 39 
iii.  The Class of 2023 reading procedures are not evidence of discrimination ................ 43 
-i-

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 3 of 78

iv.  OIR documents do not show intentional discrimination............................................. 43 
v.  Trends in the racial composition of the admitted class do not show intentional
discrimination ............................................................................................................. 49 
F.  Harvard’s Witnesses Were Credible ................................................................................. 50 
II.  HARVARD DOES NOT ENGAGE IN RACIAL BALANCING ........................................................... 54 
III.  HARVARD CONSIDERS RACE IN THE MANNER PERMITTED BY PRECEDENT ............................ 58 
A.  Harvard Has Articulated, And Is Pursuing, Exactly The Interests The Supreme Court
Has Recognized As Compelling ....................................................................................... 58 
B.  Harvard Permissibly Considers Race As A “Plus Factor”................................................ 61 
C.  Harvard Has Committed To Periodically Reviewing The Need To Consider Race......... 65 
IV. HARVARD HAS PROPERLY CONSIDERED WHETHER IT COULD ACHIEVE ITS EDUCATIONAL
INTERESTS WITHOUT CONSIDERING RACE AND HAS SHOWN IT COULD NOT.......................... 66 
CONCLUSION ............................................................................................................................. 70 

- ii -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 4 of 78

TABLE OF AUTHORITIES
Page(s)
CASES
Adams v. Ameritech Services, Inc.,
231 F.3d 414 (7th Cir. 2000) ...................................................................................................13
Adorno v. Port Authority,
258 F.R.D. 217 (S.D.N.Y. 2009) .............................................................................................49
Ahern v. Shinseki,
629 F.3d 49 (1st Cir. 2010) ........................................................................................................6
Aiken v. City of Memphis,
37 F.3d 1155 (6th Cir. 1994) ...................................................................................................65
Aliotta v. Bair,
614 F.3d 556 (D.C. Cir. 2010) .................................................................................................11
Anderson ex rel. Dowd v. City of Boston,
375 F.3d 71 (1st Cir. 2004) ....................................................................................................8, 9
Austin v. Unarco Industries, Inc.,
705 F.2d 1 (1st Cir. 1983) ..........................................................................................................8
Baker v. Goldman Sachs & Co.,
949 F. Supp. 2d 298 (D. Mass. 2013) ....................................................................................7, 9
Banerjee v. Board of Trustees of Smith College,
648 F.2d 61 (1st Cir. 1981) ........................................................................................................6
Barnstable County v. 3M Company,
2017 WL 6452245 (D. Mass. Dec. 18, 2017) ..........................................................................21
Bazemore v. Friday,
478 U.S. 385 (1986) .......................................................................................................9, 11, 12
Blunt v. Lower Merion School District,
767 F.3d 247 (3d Cir. 2014).....................................................................................................48
Chin v. Port Authority of New York & New Jersey,
685 F.3d 135 (2d Cir. 2012).................................................................................................9, 10
United States v. City of New York,
717 F.3d 72 (2d Cir. 2013).......................................................................................................14

- iii -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 5 of 78

Columbus Board of Education v. Penick,
443 U.S. 449 (1979) .................................................................................................................47
Cox v. American Cast Iron Pipe Co.,
784 F.2d 1546 (11th Cir. 1986) ...............................................................................................10
Cruz v. Dollar Tree Stores, Inc.,
2008 WL 2705087 (N.D. Cal. July 8, 2008) ............................................................................52
Cypress v. Newport News General & Nonsectarian Hospital Association,
375 F.2d 648 (4th Cir. 1967) .............................................................................................49, 50
Davis v. Califano,
613 F.2d 957 (D.C. Cir. 1979) .................................................................................................10
Davis v. Monroe County Board of Education,
526 U.S. 629 (1999) ...........................................................................................................48, 49
Desert Palace, Inc. v. Costa,
539 U.S. 90 (2003) .....................................................................................................................6
EEOC v. Sears, Roebuck & Co.,
839 F.2d 302 (7th Cir. 1988) ...................................................................................................13
EEOC v. Texas Roadhouse, Inc.,
215 F. Supp. 3d 140 (D. Mass. 2016) ......................................................................................12
Ensley Branch, NAACP v. Seibels,
31 F.3d 1548 (11th Cir. 1994) .................................................................................................56
Fisher v. University of Texas at Austin,
570 U.S. 297 (2013) ...........................................................................................................66, 67
Fisher v. University of Texas at Austin,
136 S. Ct. 2198 (2016) ..................................................................................................... passim
Goodman v. Bowdoin College,
380 F.3d 33 (1st Cir. 2004) ............................................................................................6, 16, 37
Grutter v. Bollinger,
539 U.S. 306 (2003) ......................................................................................................... passim
Hazelwood School District v. United States,
433 U.S. 299 (1977) .................................................................................................................11
Holcomb v. Iona College,
521 F.3d 130 (2d Cir. 2008).....................................................................................................49

- iv -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 6 of 78

International Brotherhood of Teamsters v. United States,
431 U.S. 324 (1977) ......................................................................................................... passim
Johnson v. Board of Regents of University of Georgia,
263 F.3d 1234 (11th Cir. 2001) ...............................................................................................64
Karp v. CIGNA Healthcare, Inc.,
882 F. Supp. 2d 199 (D. Mass. 2012) ......................................................................................10
Lam v. University of Hawaii,
40 F.3d 1551 (9th Cir. 1994) ...................................................................................................22
Latinos Unidos De Chelsea En Accion (Lucha) v. Secretary of Housing & Urban
Development, 799 F.2d 774 (1st Cir. 1986) .......................................................................12, 13
Maher v. Hyde,
272 F.3d 83 (1st. Cir. 2001) .....................................................................................................21
McGuire v. Reilly,
386 F.3d 45 (1st Cir. 2004) ......................................................................................................11
Metcalf v. Daley,
214 F.3d 1135 (9th Cir. 2000) ...................................................................................................6
Obrey v. Johnson,
400 F.3d 691 (9th Cir. 2005) ...................................................................................................13
Patterson v. Strippoli,
639 F. App’x 137 (3d Cir. 2016) .............................................................................................49
Payne v. Travenol Laboratories, Inc.,
673 F.2d 798 (5th Cir. 1982) ...................................................................................................50
Peightal v. Metropolitan Dade County,
26 F.3d 1545 (11th Cir. 1994) .................................................................................................56
Personnel Administrator of Massachusetts v. Feeney,
442 U.S. 256 (1979) .................................................................................................................47
Police Officers for Equal Rights v. City of Columbus,
644 F. Supp. 393 (S.D. Ohio 1985) ...........................................................................................9
Rathbun v. Autozone, Inc.,
361 F.3d 62 (1st Cir. 2004) ......................................................................................................12
Ravitch v. City of New York,
1992 WL 196735 (S.D.N.Y. Aug. 3, 1992) .............................................................................65

-v-

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 7 of 78

Schuler v. PricewaterhouseCoopers, LLP,
739 F. Supp. 2d 1 (D.D.C. 2010) .............................................................................................10
Segar v. Smith,
738 F.2d 1249 (D.C. Cir. 1984) ...............................................................................................14
Skains v. Lockler,
2009 WL 230037 (E.D. Cal. Jan. 20, 2009) ............................................................................52
Spath v. NCAA,
728 F.2d 25 (1st Cir. 1984) ......................................................................................................11
Students for Fair Admissions, Inc. v. President & Fellows of Harvard College,
__ F. Supp. 3d __, 2018 WL 4688308 (D. Mass. Sept. 28, 2018) ...................................6, 8, 59
Valentino v. United States Postal Service,
674 F.2d 56 (D.C. Cir. 1982) ...................................................................................................13
Village of Arlington Heights v. Metropolitan Housing Development Corp.,
429 U.S. 252 (1977) ......................................................................................................... passim
Zeno v. Pine Plains Central School District,
702 F.3d 655 (2d Cir. 2012).....................................................................................................48
DOCKETED CASES
Students for Fair Admissions, Inc. v. University of North Carolina,
No. 14-cv-954 (M.D.N.C.)...........................................................................................21, 26, 27
CONSTITUTIONAL PROVISIONS AND STATUTES
42 U.S.C. § 1983 ............................................................................................................................48
42 U.S.C. § 2000d et seq........................................................................................................ passim
42 U.S.C. § 2000e-6(a) ....................................................................................................................9
U.S. Const. amend. I .....................................................................................................................70

- vi -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 8 of 78

INTRODUCTION
Under established law, and the facts proven at trial, SFFA cannot prevail on any of its
claims. The evidence showed that Harvard does not intentionally discriminate against AsianAmerican applicants on the basis of their race and that Harvard considers race in the manner long
permitted by the Supreme Court. Harvard does not engage in racial balancing, does not consider
race as more than one among many factors in evaluating otherwise competitive candidates, and
could not achieve its educational objectives without considering race.
It is therefore no surprise that SFFA’s response is to rewrite the law and reimagine the
facts. Time and again, SFFA’s filing offers incorrect legal arguments it has never before
presented. And time and again, SFFA characterizes the evidence in ways irreconcilable with the
trial record.
First, on its intentional discrimination claim, SFFA argues for the first time that Harvard
bears the burden of persuasion, and it invokes a legal standard—the pattern-or-practice standard
set forth in International Brotherhood of Teamsters v. United States, 431 U.S. 324 (1977)—on
which it has never previously relied. But neither of SFFA’s novel arguments is correct. It is
well settled that plaintiffs bear the burden of proof under Title VI (as in civil litigation more
generally); the fact that strict scrutiny applies to Harvard’s consideration of race in pursuit of
diversity does not relieve SFFA of the burden to prove Harvard intentionally discriminates
between Asian-American and White applicants. And the Teamsters standard is inapplicable to
suits under Title VI and to suits brought by an individual or an association; rather, the governing
line of cases is the one SFFA has repeatedly invoked, beginning with Village of Arlington
Heights v. Metropolitan Housing Development Corp., 429 U.S. 252 (1977). SFFA presumably
invokes the Teamsters standard because it believes that standard to be more favorable to the use
of statistical evidence—an implicit concession that SFFA’s claim depends entirely on statistics.
-1-

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 9 of 78

But the reality is that courts applying the Teamsters standard have agreed with those applying the
Arlington Heights standard that statistical evidence—even where reliable—must be evaluated in
light of non-statistical evidence. SFFA has no non-statistical evidence to bolster its flawed
statistics.
SFFA equally mischaracterizes the facts relevant to its intentional discrimination claim.
One of the most glaring flaws in SFFA’s theory is its failure to explain why, if Harvard
disfavored Asian-American applicants on the basis of their race, it would exercise that supposed
animus against only some Asian-American applicants, and not the so-called ALDC applicants.
Rather than taking this final opportunity to offer an explanation, SFFA instead tries to pretend
the problem does not exist; it describes the evidence as showing that Asian-American ALDC
applicants are so highly qualified that their strength and their valued characteristics “simply
overwhelm” the supposed “penalty that racial stereotyping imposes.” SFFA ¶ 202.1 In fact, the
evidence shows not just that Asian-American ALDC applicants (like other ALDC applicants) are
strong candidates relative to the rest of the applicant pool but that they are more likely to be
admitted than White applicants of equal strength—and for legacy applicants, that effect is
statistically significant. Harvard ¶ 108.
SFFA seeks to elide the gaps in its evidence of intentional discrimination by accusing
Harvard’s admissions officers of unconsciously discriminating against Asian-American
applicants. But SFFA presented no actual evidence that anyone in the Admissions Office bore
unconscious bias. Instead, it relies on largely extra-record evidence of societal discrimination
against Asian Americans (¶ 189, citing a government report and a law review article)—and then

1

“SFFA” citations refer to SFFA’s proposed findings of fact and conclusions of law (Dkt.
620); “Harvard” citations refer to Harvard’s proposed findings of fact and conclusions of law
(Dkt. 619).
-2-

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 10 of 78

makes the extraordinary suggestion that Harvard must prove it “has uniquely escaped [the]
infiltration” of societal prejudice (id.). SFFA could hardly be more candid about its lack of
evidence.
Second, SFFA tries to salvage its racial balancing claim in the face of the unrebutted
evidence that the racial composition of Harvard’s admitted and matriculating classes has varied
meaningfully from year to year. Unable to respond to that evidence, SFFA ignores it. Instead,
SFFA bases its racial balancing arguments on the test-score thresholds used to send recruitment
letters and on reports, periodically generated for Admissions Office leaders, that summarize the
characteristics of the tentatively admitted class on numerous dimensions (not just race). Those
arguments simply highlight the weakness of SFFA’s case. Title VI does not limit the
consideration of race in recruiting in the way that it limits the consideration of race in
admissions—as SFFA’s Complaint and its own expert have recognized, in encouraging Harvard
to undertake more race-conscious recruiting—and, in any event, race-conscious recruiting hardly
suggests an effort to achieve a predetermined racial composition of the student body. And SFFA
mischaracterizes the evidence concerning the statistical reports and their use in estimating the
overall yield rate of the admitted class.
Third, SFFA’s challenge to the manner in which Harvard considers race once again
shows that its real disagreement is with the Supreme Court precedents that have held up the
Harvard process as a model of legality. SFFA faults Harvard for not articulating a quantitative
“level” of diversity that it believes is necessary, but although Supreme Court precedent allows
consideration of concepts like critical mass, it does not require an institution to articulate its
goals in that manner. See, e.g., Fisher v. University of Texas at Austin (Fisher II), 136 S. Ct.
2198, 2210 (2016) (“[S]ince the University is prohibited from seeking a particular number or

-3-

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 11 of 78

quota of minority students, it cannot be faulted for failing to specify the particular level of
minority enrollment at which it believes the educational benefits of diversity will be obtained.”).
SFFA points to subtle differences in how various admissions officers described their
consideration of race, but those variations are to be expected in Harvard’s highly individualized
process, which considers race as one of many contextual factors, rather than in the formulaic
manner the Supreme Court has condemned. And although SFFA faults Harvard for admitting a
greater number of African-American and Hispanic candidates as a result of its consideration of
race, that sort of meaningful increase in diversity—which immeasurably enriches Harvard’s
campus community and its educational environment—is exactly the purpose for which the
Supreme Court has allowed universities to consider race. Contrary to SFFA’s attacks, the trial
showed that Harvard considers race in the flexible, individualized manner permitted by the
Supreme Court; that race meaningfully affects the likelihood of admission only for candidates
who would be highly competitive no matter what their race; and that, for such candidates, race
does not have an effect disproportionate to that of other factors.
Finally, SFFA attempts to defend its claim that Harvard could achieve its educational
objectives without considering race. But after its own expert agreed that Harvard’s race-neutral
alternatives committee, comprising three of the most senior officials with responsibility for the
College, had examined the right set of race-neutral practices and evaluated them by the right
criteria, SFFA is left only with the ipse dixit that the four combinations of practices proposed by
its expert at trial would be viable for Harvard—even though all would produce a 30% decline in
the proportion of African-American students in the student body and would meaningfully
compromise Harvard’s standards of excellence, as the committee concluded after careful study.

-4-

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 12 of 78

Those are the sorts of sacrifices the Supreme Court has held are not required as the price of
achieving the educational benefits of diversity.
The trial thus showed that SFFA cannot prevail on its claims under established law. Two
footnotes buried in SFFA’s filing make clear its real ambition: to change the law, by asking the
Supreme Court to overrule its precedents and hold that Title VI forbids any consideration of race
“by any federally funded university.” SFFA 47 n.2; see id. at 48 n.3. If that day ever comes, it
will dramatically reduce diversity at Harvard and at colleges and universities across the country.
It will undercut the considered judgment of educators at Harvard and elsewhere that diversity
enhances the campus community and the learning that takes place in classrooms, around the
tables in dining halls, and on playing fields. It will leave generations of young Americans less
equipped to thrive in confronting the challenges of a complex world. And it will send the
message—and create the reality—that America’s universities are no longer its cradles of
opportunity and its beacons of social mobility. Congress cannot possibly have envisioned that
Title VI, a statute enacted to expand opportunity, would produce that result.
ARGUMENT
The sections that follow explain why SFFA’s proposed findings of fact (to the extent they
are inconsistent with Harvard’s) are not supported by the record evidence and why SFFA has
failed to establish Harvard’s liability on any of its remaining claims. To the extent any of
Harvard’s assertions about the record is not already made in Harvard’s proposed findings of fact,
Harvard asks that the Court consider the assertion as a supplemental proposed finding of fact.2
2

Two amicus briefs have been filed on SFFA’s behalf. See Dkts. 622, 624. Unlike some
of Harvard’s amici, who sought and were granted the right to submit evidence of their own, none
of SFFA’s amici has sought or been granted that right. Yet their briefs pervasively rely on facts
outside the record and offer arguments not supported by evidence presented at trial—even
though the evidence and record are closed. If (for example) SFFA’s amici economists had
evidence or opinions to offer, they should have done so at trial, and undergone cross-5-

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 13 of 78

I.

HARVARD DOES NOT DISCRIMINATE AGAINST ASIAN-AMERICAN APPLICANTS
A.

SFFA Bears The Burden To Prove Intentional Discrimination

SFFA argues that Harvard bears the burden of disproving intentional discrimination.
SFFA ¶¶ 164-165. That argument is incorrect, contrary to this Court’s prior ruling, and waived.
The First Circuit has squarely held that, to establish “claims of racial discrimination”
under Title VI, the “plaintiff must demonstrate” that the defendant acted with “racial animus.”
Goodman v. Bowdoin Coll., 380 F.3d 33, 43 (1st Cir. 2004); see also Ahern v. Shinseki, 629 F.3d
49, 54 (1st Cir. 2010) (“[T]he burden of proving intentional discrimination never shifts; it
remains with the plaintiff throughout[.]”). That is of course consistent with the “[c]onventional
rul[e] of civil litigation” that requires “a plaintiff to prove his case by a preponderance of the
evidence.” Desert Palace, Inc. v. Costa, 539 U.S. 90, 99 (2003) (alterations in original; internal
quotation marks omitted). The Court accordingly adopted that articulation of the law in its order
denying summary judgment. See Students for Fair Admissions, Inc. v. President & Fellows of
Harvard College, __ F. Supp. 3d __, 2018 WL 4688308, at *12 (D. Mass. Sept. 28, 2018)
(quoting Goodman, 380 F.3d at 43). SFFA does not even acknowledge the Court’s prior
holding, let alone argue the Court should reconsider it.
SFFA proposes, in effect, that it should be relieved of the ordinary burden placed on Title
VI plaintiffs because Harvard’s consideration of race in its admissions process must be evaluated
under strict scrutiny. But that argument ignores the distinction between Harvard’s consideration
of race in pursuit of a diverse student body—an element of the admissions process that Harvard

examination, just as other amici did. The Court should disregard their after-the-fact offers of
evidence and opinions. See, e.g., Metcalf v. Daley, 214 F.3d 1135, 1141 n.1 (9th Cir. 2000)
(striking extra-record material submitted with an amicus brief); Banerjee v. Bd. of Trustees of
Smith Coll., 648 F.2d 61, 65 n.9 (1st Cir. 1981) (“[T]he prime, if not sole, purpose of an amicus
curiae brief is what its name implies, namely, to assist the court on matters of law.”).
-6-

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 14 of 78

regards as fundamental to its institutional mission—and SFFA’s allegation that “Harvard
discriminates against Asian-American applicants vis-à-vis white applicants,” SFFA ¶ 164
(emphasis added), which Harvard vehemently denies and which the evidence disproves. SFFA
bears the burden to prove, in the face of Harvard’s denial and the strong evidence supporting it,
that Harvard actually does intentionally discriminate in that way. It cannot evade that burden
simply because it is undisputed that the admissions process considers race in other ways.
SFFA has consistently recognized as much in its prior filings. In its brief in support of
summary judgment, for example, SFFA referred to its “burden” of proving “intentional racial
discrimination.” Dkt. 413 at 5. In its summary judgment reply brief, SFFA again referred to the
plaintiff’s “burden … to prove discrimination by a preponderance of the evidence.” Dkt. 510 at
7 (internal quotation marks omitted). SFFA did not revisit that position in the joint pretrial
memorandum. See Dkt. 570. And in its closing argument at trial, SFFA again acknowledged
that it must “carry [its] burden and have proof of intentional discrimination.” 11/2 Tr. 43:4-5.
When a party adopts one position before and during trial and takes the opposite position in its
post-trial briefing, the argument is waived. See, e.g., Baker v. Goldman Sachs & Co., 949 F.
Supp. 2d 298, 315 (D. Mass. 2013) (rejecting “novel theories” not “presented or mentioned at or
before trial,” because “defendant had no notice of them prior to and during trial”), aff’d, 771 F.3d
37 (1st Cir. 2014).
B.

SFFA Invokes An Inapplicable Legal Theory Because It Cannot Prevail
Under The Theory It Has Long Advanced

SFFA makes another about-face in arguing that its allegations of discrimination should be
assessed under the Teamsters framework rather than the standard SFFA has consistently
advocated and the Court has already embraced—that of Arlington Heights. Again, SFFA’s
invocation of Teamsters is incorrect, contrary to the Court’s prior ruling, and waived. It is also

-7-

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 15 of 78

for naught: Although SFFA seems to think the Teamsters standard allows it to make its case
based solely on its flawed statistical evidence, that is not true.
i.

Teamsters does not apply

In its order denying summary judgment, the Court concluded that the “sensitive inquiry”
articulated in Arlington Heights, as reiterated by the First Circuit in Anderson ex rel. Dowd v.
City of Boston, 375 F.3d 71, 83 (1st Cir. 2004), provides the relevant framework for assessing
SFFA’s claim of intentional discrimination in Harvard’s “facially neutral policy.” 2018 WL
4688308, at *12. SFFA fails even to acknowledge that ruling, let alone explain why the Court
should reconsider it.
Much less does SFFA explain how it could have been erroneous for the Court to adopt a
position SFFA itself had urged. Cf. Austin v. Unarco Indus., Inc., 705 F.2d 1, 15 (1st Cir. 1983)
(under “the doctrine of ‘invited error[,]’ … a party may not appeal from an error to which he
contributed, either by failing to object or by affirmatively presenting to the court the wrong
law”). In its prior statements on the matter, SFFA has consistently taken the position that
Arlington Heights supplies the standard for determining whether Harvard intentionally
discriminates against Asian-American applicants. In its summary judgment brief, for example,
SFFA wrote that “[c]ourts apply the framework articulated in Village of Arlington Heights to
determine whether a facially neutral policy has a discriminatory purpose.” Dkt. 413 at 6
(internal quotation marks and citations omitted). SFFA again invoked Arlington Heights as the
relevant standard in its summary judgment reply brief, Dkt. 510 at 14, and in opposing Harvard’s
motions in limine, Dkt. 569 at 5-6. And it said nothing different in the pretrial memorandum.
See Dkt. 570. As noted above, parties are not allowed to adopt one position before and during
trial and change course in their post-trial briefing; considerations of fair notice preclude such

-8-

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 16 of 78

bait-and-switch tactics. Baker, 949 F. Supp. 2d at 315. SFFA’s invocation of Teamsters is
therefore waived just as its burden-allocation argument is waived.
There is a reason SFFA long invoked the Arlington Heights standard: It is the standard
that properly applies to claims, like SFFA’s, that a “facially neutral” policy (here, one that does
not favor White over Asian-American or Asian-American over White applicants) in fact reflects
racial animus. Anderson, 375 F.3d at 83. By contrast, the Teamsters pattern-or-practice
framework is inapposite here for two reasons: It applies neither to Title VI claims nor to suits
brought by private non-class plaintiffs.
First, the Teamsters framework—which originated in the Title VII context—reflects Title
VII’s express statutory authorization for “pattern or practice” suits by the United States.
Teamsters, 431 U.S. at 328 & n.1; see 42 U.S.C. § 2000e-6(a) (Title VII) (authorizing Attorney
General to bring civil actions based on “pattern or practice of resistance to the full enjoyment of
any of the rights secured by this subchapter”). Title VI has no such language. That is why not a
single one of the 45 cases SFFA cites on this issue (¶¶ 167-176) actually holds that the Teamsters
pattern-or-practice framework applies to Title VI claims. Indeed, only two of those 45 cases
feature any analysis of a Title VI claim—Bazemore v. Friday, 478 U.S. 385 (1986), which
involved Title VII’s express provision for pattern-or-practice suits, id. at 398; and Police Officers
for Equal Rights v. City of Columbus, 644 F. Supp. 393 (S.D. Ohio 1985), in which the Court’s
cursory Title VI analysis depended entirely on its Title VII analysis, see id. at 438.
Second, even under Title VII, pattern-or-practice claims are available only to the United
States and class plaintiffs. In Chin v. Port Authority of New York & New Jersey, 685 F.3d 135
(2d Cir. 2012), for example, the Second Circuit explained at length why “nonclass private
plaintiffs,” including an association suing on behalf of its members, could not proceed under the

-9-

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 17 of 78

Teamsters pattern-or-practice framework. Id. at 146-150. And the court explained that “all of
[its] sister circuits to consider the question have” similarly “held that the pattern-or-practice
method of proof is not available to private, nonclass plaintiffs.” Id. at 149-150 (citing, among
other cases, Davis v. Coca-Cola Bottling Co. Consol., 516 F.3d 955, 967-969 (11th Cir. 2008);
Bacon v. Honda of Am. Mfg., 370 F.3d 565, 575 (6th Cir. 2004); Celestine v. Petroleos de
Venezuella SA, 266 F.3d 343, 355-356 (5th Cir. 2001); Gilty v. Village of Oak Park, 919 F.2d
1247, 1252 (7th Cir. 1990); Lowery v. Circuit City Stores, Inc., 158 F.3d 742, 761 (4th Cir.
1998), vacated on other grounds, 527 U.S. 1031 (1999)). SFFA cites only a single out-of-circuit
district court decision to the contrary. SFFA 50 n.4 (citing Brotherhood of Maintenance Way
Employees v. Indiana Harbor Belt Railroad Co., 2014 WL 4987972 (N.D. Ind. Oct. 7, 2014)).3
Within this Circuit—as in other circuits—courts have certainly recognized that individual
plaintiffs may rely on statistical evidence of systematic discrimination, e.g., Karp v. CIGNA
Healthcare, Inc., 882 F. Supp. 2d 199, 212-213 (D. Mass. 2012), but they have declined to apply
the Teamsters burden-shifting framework in non-class suits, id. at 213.
ii.

SFFA’s evidence is just as inadequate under Teamsters

SFFA is presumably trying to change its position on the governing law in the misguided
view that the Teamsters standard is more favorable to statistical evidence than the Arlington
Heights standard—an implicit concession that SFFA’s intentional discrimination claim rests
almost entirely on statistical evidence and thus cannot prevail under Arlington Heights. It is well
recognized that, under the Arlington Heights standard, statistics cannot prove intent to

3

SFFA also cites Cox v. American Cast Iron Pipe Co., 784 F.2d 1546 (11th Cir. 1986), but
Cox was brought as a class action. And it cites Davis v. Califano, 613 F.2d 957 (D.C. Cir. 1979),
but that decision does not establish that an individual non-class plaintiff can pursue a pattern-orpractice claim, for the reasons explained in Schuler v. PricewaterhouseCoopers, LLP, 739 F.
Supp. 2d 1, 6-7 (D.D.C. 2010), aff’d, 421 F. App’x 1 (D.C. Cir. 2011).
- 10 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 18 of 78

discriminate except in the “rare” cases where they are so “stark” that no other inference is
possible. Arlington Heights, 429 U.S. at 266; see also McGuire v. Reilly, 386 F.3d 45, 63 (1st
Cir. 2004) (“[C]ourts have been loathe to infer intent from mere effect[.]”); Spath v. NCAA, 728
F.2d 25, 28 (1st Cir. 1984) (“[G]enerally courts must look to evidence other than statistical
impact to support a finding of discriminatory purpose.”). SFFA has acknowledged as much. SJ
Mem. (Dkt. 413) at 6 (only “[i]n rare cases” can “a stark statistical pattern … serve as the sole
proof of discriminatory intent” (internal quotation marks omitted)). SFFA is presumably trying
to shift to a new and inapplicable legal paradigm because it thinks it could prevail under the
Teamsters framework, unlike under Arlington Heights. But SFFA is wrong to think the
Teamsters framework would provide for a more favorable assessment of its evidence, for two
key reasons.
First, courts applying the Teamsters standard—like courts applying the Arlington Heights
standard—have recognized that there are limits to the value of statistical evidence in ascertaining
the existence of intentional discrimination. It is true that, where “gross statistical disparities can
be shown, they alone may in a proper case constitute prima facie proof of a pattern or practice of
discrimination.” Hazelwood School District v. United States, 433 U.S. 299, 307-308 (1977)
(emphasis added) (citing Teamsters, 431 U.S. at 339). But any statistical disparities shown in
this case are certainly not “gross” disparities, id. And under Teamsters or any other framework,
a statistical analysis (like Dr. Arcidiacono’s) that is “based on a flawed methodology” and does
“not control for … important explanatory variable[s]” is simply “not relevant to determining
whether” discrimination exists. Aliotta v. Bair, 614 F.3d 556, 251-252 (D.C. Cir. 2010).4 Even

4

SFFA cites Justice Brennan’s opinion in Bazemore v. Friday, 478 U.S. 385 (1986), for
the proposition that a plaintiff may prove discrimination through “‘a regression analysis that
includes less than all measurable variables[.]’” SFFA ¶ 181 (quoting 478 U.S. at 400; some
- 11 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 19 of 78

where statistics are deemed sufficient to establish a prima facie case of intentional
discrimination, moreover, that prima facie case may be rebutted by showing that such evidence is
“either inaccurate or insignificant.” Teamsters, 431 U.S. at 360.
In addition, contrary to SFFA’s wishful suggestion (¶ 173) that statistical evidence is
“sufficient ‘alone’” to demonstrate a pattern or practice of discrimination, both the Supreme
Court in Teamsters itself and other courts applying the Teamsters framework have rejected that
premise. The Supreme Court favorably observed in Teamsters that the government (as plaintiff)
had not “relied on ‘statistics alone’”; rather, “[t]he individuals who testified about their personal
experiences with the company,” and who “recounted over 40 specific instances of
discrimination,” “brought the cold numbers convincingly to life.” 431 U.S. at 338-339. The
Court also cautioned that “statistics are not irrefutable” and that “their usefulness depends on all
of the surrounding facts and circumstances.” Id. at 340. The First Circuit, too, has explained
that “statistics come in infinite variety and their usefulness or weight depend[s] on … the
existence of proper supportive facts and the absence of variables which would undermine the
reasonableness of the inference of discrimination which is drawn therefrom.” Latinos Unidos De
Chelsea En Accion (Lucha) v. Sec’y of Hous. & Urban Dev., 799 F.2d 774, 784 (1st Cir. 1986)
(internal quotation marks and citation omitted; alteration in original); see also, e.g., Rathbun v.
Autozone, Inc., 361 F.3d 62, 79-80 (1st Cir. 2004); EEOC v. Texas Roadhouse, Inc., 215 F. Supp.
3d 140, 169 (D. Mass. 2016). Indeed, the First Circuit, in applying Teamsters, has cited—as the

internal quotation marks omitted). But Justice Brennan’s opinion was not the endorsement of
incomplete regression models SFFA makes it out to be. It concluded only that an analysis that
includes “less than ‘all measurable variables’” may be “admissib[le]” as long as it “accounts for
the major factors” and does not suffer from any “other infirmity.” 478 U.S. at 400. SFFA’s
analysis excludes “major factors,” see Harvard ¶¶ 121-145, and suffers from “other
infirmit[ies],” see id. ¶¶ 106-120, so it is not probative evidence of intentional discrimination.
- 12 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 20 of 78

sorts of cases “in which impact alone provides sufficient evidence of invidious intent”—exactly
the extreme examples cited in Arlington Heights, 429 U.S. at 266. Latinos Unidad, 799 F.2d at
785 (citing Gomillion v. Lightfoot, 364 U.S. 339, 341 (1960); Yick Wo v. Hopkins, 118 U.S. 356,
374 (1886)); see also Harvard ¶ 266 (discussing Gomillion and Yick Wo). And one of the very
decisions on which SFFA relies (¶ 173)—Adams v. Ameritech Servs., Inc., 231 F.3d 414 (7th
Cir. 2000)—recognizes that although “statistical evidence can be very useful to prove
discrimination” in both disparate-treatment and pattern-or-practice cases, “it will likely not be
sufficient in itself.” Id. at 423.
In this case, both experts testified that their models did not account for all the factors
considered in Harvard’s admissions process, in part because of the important role of qualitative
factors. 10/25 Tr. 80:13-15 (Arcidiacono); 10/30 Tr. 123:1-17 (Card); Harvard ¶ 148. The
factors in Dr. Card’s model explained only 64% of the variation in admissions outcomes, and the
factors in Dr. Arcidiacono’s model explained only 56%; for Dr. Arcidiacono’s model of the
personal rating, the figure was only 29%. DD10.61; DD10.86; 10/31 Tr. 40:14-43:25 (Card); id.
at 93:22-94:12. “[W]hen the statistical evidence does not adequately account for the diverse and
specialized qualifications necessary for (the positions in question), strong evidence of individual
instances of discrimination becomes vital to the plaintiff’s case.” Valentino v. United States
Postal Serv., 674 F.2d 56, 69 (D.C. Cir. 1982) (internal quotation marks omitted); see also, e.g.,
Obrey v. Johnson, 400 F.3d 691, 698 (9th Cir. 2005) (citing Valentino); EEOC v. Sears, Roebuck
& Co., 839 F.2d 302, 311 (7th Cir. 1988) (same). But SFFA did not offer a single witness to
bring its “cold numbers convincingly to life,” Teamsters, 431 U.S. at 339. Cf. Sears, 839 F.2d at
311 (“The EEOC’s reasons for not presenting such individual testimony are not satisfying.”).

- 13 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 21 of 78

Nor did SFFA introduce into evidence or offer testimony regarding the application files of any of
its standing members, despite having access to those files. Harvard ¶ 190.
Second, although SFFA places great emphasis on the burden-shifting component of the
Teamsters framework, it is essentially irrelevant here. When a plaintiff establishes a prima facie
case of discrimination under Teamsters, the burden that shifts to the defendant is a burden of
production, not of persuasion. The defendant’s task is simply to produce enough countervailing
evidence to “cast sufficient doubt on the plaintiffs’ proof to permit the trier of fact legitimately to
decline to draw an inference of discrimination from that proof.” Segar v. Smith, 738 F.2d 1249,
1269 (D.C. Cir. 1984). “[A] defendant’s burden of production ‘can involve no credibility
assessment,’ and ‘necessarily precedes the credibility-assessment stage[.]’” United States v. City
of New York, 717 F.3d 72, 87 (2d Cir. 2013) (citation omitted) (quoting St. Mary’s Honor Ctr. v.
Hicks, 509 U.S. 502, 509 (1993)). If the defendant meets that burden of production, then “the
presumption arising from the plaintiff's prima facie case ‘drops out,’ and the trier of fact must
then determine, after a full trial, whether the plaintiff has sustained its burden of proving by a
preponderance of the evidence the ultimate fact at issue.” Id. (quoting Hicks, 509 U.S. at 510511). Here, SFFA could not seriously dispute that the statistical analysis presented by Dr. Card
at the very least satisfies any burden of production Harvard might incur by way of the Teamsters
burden-shifting framework. The shifting of burdens is thus irrelevant, and all that remains is for
the Court to decide the same ultimate question of fact it would decide under the Arlington
Heights framework: whether SFFA has proven that Harvard intentionally discriminated against
Asian-American applicants.
In short, SFFA is wrong to argue that the Teamsters standard applies here—and also
wrong to think that standard would save its deeply flawed case.

- 14 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 22 of 78

C.

SFFA’s Theory Of Discrimination Is Incoherent

Beyond the multifarious flaws of SFFA’s statistical and non-statistical evidence—
discussed below in greater detail—SFFA’s theory of discrimination fails for a broader, more
basic reason: SFFA yet again fails to explain why, if Harvard intended to discriminate against
Asian-American applicants on the basis of their race, it would exercise that intent only against
some Asian-American applicants and not against Asian-American ALDC applicants.
SFFA and its statistical expert do not even claim that Harvard’s supposed scheme of
discrimination extends to Asian-American ALDC applicants. Harvard ¶¶ 107-108, 274. ALDC
applicants are a significant group; they comprise nearly 30% of admitted students. Id. ¶ 107.
Yet SFFA would have the Court believe that Harvard targets non-ALDC Asian-American vis-àvis non-ALDC White applicants—and then turns around and favors ALDC Asian-American
applicants relative to otherwise similar White applicants. That scenario is implausible absent an
explanation, yet SFFA makes no attempt to supply one.
To the contrary, SFFA pretends the problem does not exist. SFFA tries to suggest that
what is happening is that Asian-American ALDC applicants are of such surpassing quality that
their strength (and their ALDC status) “simply overwhelm the penalty that racial stereotyping
imposes.” SFFA ¶ 202; see also id. ¶ 203 (suggesting that Asian-American ALDC applicants
“overcame discrimination”). But that either misunderstands or mischaracterizes the evidence.
Dr. Arcidiacono’s analysis showed not only that Asian-American ALDC applicants were
admitted at a higher rate than White applicants, but indeed that they are more likely to be
admitted than otherwise identical White ALDC applicants—in other words, applicants who are
exactly as strong in every respect and who also benefit from the consideration of their ALDC
status. Harvard ¶ 108. For legacy applicants, that difference is statistically significant. Id. Such
a pattern would be truly implausible if Harvard bore racial animus against Asian-American
- 15 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 23 of 78

applicants—as SFFA is required to show to prove its case, Goodman, 380 F.3d at 43. And it
would be even less plausible if, as SFFA now suggests, Harvard’s admissions officers bore
unconscious bias on the basis of race—since an unconscious bias would presumably be even
harder to turn on (for non-ALDC applicants) and off (for ALDC applicants) at will.
To be clear, Harvard’s position is not (as SFFA wrongly suggests, ¶ 204) that an
intentional discrimination claim must fail unless every member of a protected class suffers
discrimination. Harvard’s point, rather, is that it is implausible to think Harvard’s admissions
officers intend to discriminate on the basis of race against any Asian-American applicants
where—by SFFA’s own admission—the supposed discrimination is manifested only against
some Asian-American applicants, absent an explanation supported by the evidence for why that
would happen.
D.

The Statistical Evidence Does Not Show Intentional Discrimination

SFFA’s statistical case is pervasively marred by its failure to model how the admissions
process actually works. Dr. Arcidiacono’s descriptive analysis relies on the incorrect view that
academic factors drive the admissions process. And his regression analysis is outcome-driven;
he excludes large groups of applicants and factors that are important in the admissions process in
an evident attempt to reach the result SFFA wanted him to find.5
5

To distract from the flaws of its own expert’s analysis, SFFA tries to diminish the
qualifications of Harvard’s expert, Dr. Card, accusing him (¶ 23) of having never “previously
created models of this size and scope.” Dr. Card is a leading expert in the fields of study that
informed his work in this case: econometrics, which is “the application of statistical methods to
problems” in the real world, and labor economics, which is the study of “individual decisionmaking,” including in the context of “discrimination.” 10/30 Tr. 73:16-74:4. He has “written a
number of articles on both gender discrimination and race discrimination issues,” including a
study of the effects of race-conscious admissions. Id. at 74:7-14. SFFA’s expert did not criticize
Dr. Card’s expertise; to the contrary, he testified that he respected Dr. Card “[v]ery much” and
described one of Dr. Card’s awards as “[a]rguably more important” than the Nobel Prize. 10/25
Tr. 178:23-179:4. The fact that SFFA apparently feels the need to resort to attacking Dr. Card’s
credentials—when its own expert lauded them—speaks volumes.
- 16 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 24 of 78

i.

SFFA’s descriptive analysis relies on the incorrect view that grades and
test scores drive the admissions process

Unlike a regression analysis, which seeks to discern the effect of a given factor on an
outcome holding all other factors equal, descriptive statistics (as the name suggests) simply
describe aggregate data in simplistic terms. See, e.g., 10/19 Tr. 121:15-21 (Driver-Linn); 10/25
Tr. 21:7-17, 185:21-186:2 (Arcidiacono). SFFA leans heavily (¶¶ 24-37) on Dr. Arcidiacono’s
descriptive statistics, which compare the ratings and admissions outcomes of applicants across
the entire pool and within deciles of the “Academic Index”—a mechanical formula, combining
applicants’ high-school GPAs and SAT scores, that the Admissions Committee does not use in
making admissions decisions but maintains for Ivy League athletic reporting purposes (see
Harvard ¶¶ 162, 212). For example, SFFA argues (¶¶ 31, 32) that race must affect the personal
rating because “African Americans and Hispanics in the top four academic index deciles have a
substantially higher probability of receiving a personal rating of 1 or 2 than whites and Asian
Americans,” and because the distribution of the personal rating by race and academic decile
resembles that of the preliminary overall rating (which does consider race as one among many
factors).
That argument reflects SFFA’s fundamental failure to acknowledge that Harvard’s
admissions process is driven largely by non-academic factors, as well as by academic factors that
go well beyond the two components measured in the Academic Index.6 Comparing applicants’
At any rate, SFFA mischaracterizes the testimony on which it relies. Dr. Card was asked
whether he had created “discrete choice models with over 200 variables,” not whether he had
created any statistical models of similar size and scope. 11/1 Tr. 10:19-11:1.
6
The non-academic factors that matter in the admissions process include not only those
reflected by the extracurricular and personal ratings but also, importantly, those reflected by the
athletic rating. SFFA continues to insist (¶ 28) that the athletic rating does not matter except for
recruited athletes. It is understandable that SFFA would wish for that to be true, since White
applicants on average have higher athletic ratings than Asian-American applicants—which partly
explains why they are admitted at a slightly higher rate than Asian-American applicants.
- 17 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 25 of 78

ratings and admissions outcomes within a given decile of the Academic Index is in effect like
running a regression that controls only for GPA and SAT scores; it compares applicants who are
similar on those two factors but who may vary in many other ways for which the analysis does
not account. Cf. 10/30 Tr. 158:15-18 (Card) (controlling for legacy and recruited-athlete status is
comparable to examining descriptive statistics for non-legacy, non-athlete students).
But although grades and test scores are certainly important to the admissions process,
excellence on those measures is more abundant in the applicant pool than almost any other
measure, and even when academic strength is more broadly defined, it is more abundant than
other dimensions of applicant strength. Harvard ¶¶ 33-34. The overwhelming majority of
admitted applicants are strong not just in academics but in multiple dimensions. Harvard ¶ 34;
DD10.8; DX672. And contrary to SFFA’s assertion that “[t]here is a strong positive relationship
between the academic index” and all of the profile ratings (¶ 33), there is in fact very little
correlation between academic and non-academic strengths as measured by the profile ratings.
DD10.65 (showing almost no correlation between the academic rating and the personal rating);
10/31 Tr. 49:4-52:9 (Card). Thus, SFFA’s failure to recognize the importance of non-academic
factors means that its descriptive statistics are far less consequential than SFFA claims. SFFA
interprets Dr. Arcidiacono’s descriptive statistics as evidence of discrimination only by

DD10.10; DX692. But the reality is that athletic strength does influence admissions outcomes
for applicants who are not recruited athletes. Harvard ¶ 45. For example, among applicants with
one other strength (i.e., profile rating of 1 or 2), applicants with an athletic rating of 2 have an
admission rate of 8%, while applicants with an athletic rating of 3 or worse have an admission
rate of just 2%. DD10.11. Among applicants with two other strengths, those figures are 32%
and 13%; among applicants with three other strengths, the figures are 68% and 50%. Id. In each
scenario, the difference in outcomes between applicants with an athletic rating of 2 and those
with a rating of 3 or worse is quite substantial.
- 18 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 26 of 78

overlooking the reality that high-school grades and SAT scores are very far from the only
qualifications that matter.
For example, SFFA crows (¶ 31)—under the apparent impression that it is damning—that
there are racial differences in the overall and personal ratings within deciles of the Academic
Index, with African-American and Hispanic applicants having somewhat higher ratings on
average than White and Asian-American applicants in the same Academic Index decile. But
those differences fail to control for the many dimensions of strength—both academic and nonacademic—that are not captured by controlling only for grades and SAT scores. And crossracial comparisons across deciles of the Academic Index are misleading for another reason as
well: There is wide variation by race in the number and proportion of applicants who fall within
a given decile, reflecting (for example) the fact that African-American and Hispanic applicants
disproportionately face challenges of educational opportunity that may limit the degree to which
grades and test scores reflect their true academic potential. 10/17 Tr. 157:23-158:6
(Fitzsimmons).
As Dr. Arcidiacono’s slide PD38.8 illustrates, for instance, the top Academic Index
decile includes roughly 18% of Asian-American applicants, 9% of White applicants, 2% of
Hispanic applicants, and 1% of African-American applicants (or 7,225 Asian-American
applicants, 4,963 White applicants, 380 Hispanic applicants, and 132 African-American
applicants, PD38.7). So when SFFA says (for example) that Asian-American applicants in the
top decile have, on average, lower personal or preliminary overall ratings than African-American
applicants in that decile (PD38.16, PD38.18), it is comparing the top 1% of African-American
candidates by grades and test scores to the top 18% of Asian-American candidates by the same
measure. That is simply not an illuminating comparison; SFFA has not explained why one

- 19 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 27 of 78

would necessarily expect the 1% of African-American candidates with the highest grades and
test scores to have non-academic strengths identical to those of the 18% of Asian-American
candidates with the highest grades and test scores.
SFFA also points (¶ 36) to the fact that White applicants are admitted at slightly higher
rates than Asian-American applicants with comparable grades and SAT scores. That, too, does
not show intentional discrimination. Although SFFA claims that fact “suggests that Asian
Americans are being penalized vis-à-vis whites with similar qualifications,” id., all it reflects in
reality is that applications submitted by White applicants tend to show greater non-academic
strengths than those submitted by Asian-American applicants with similar grades and SAT
scores. That is not evidence of discrimination. To the contrary, the evidence showed that,
among applicants with identical academic ratings, Asian-American applicants tend on average to
have weaker support than White applicants from their teachers, guidance counselors, and alumni
interviewers. Harvard ¶ 139; DX692; DD10.68-75. That support often plays an important role
in illuminating a candidate’s non-academic strengths.
In short, Dr. Arcidiacono’s descriptive statistics focus relentlessly on grades and
standardized test scores, and SFFA interprets those statistics as evidence of discrimination only
by overlooking the importance of non-academic factors in the Harvard admissions process.
SFFA might well prefer an admissions process in which grades and test scores were the
overriding considerations. But Harvard has long valued qualities in its applicants that go well
beyond those quantitative measures of prior academic success, and the law does not require it to
adopt a different approach.
ii.

SFFA’s regression analysis is deeply flawed

Dr. Arcidiacono’s regression analysis is no more persuasive evidence of intentional
discrimination than his descriptive statistics. On one methodological issue after another, Dr.
- 20 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 28 of 78

Arcidiacono chose to exclude applicants considered as part of the same admissions process and
to exclude factors considered in that process, with no sound basis for doing so. His choices
appear to have been driven by the desire to reach the result that SFFA wanted to find. As a
result, his analysis does not reflect an appropriate attempt to model Harvard’s admissions process
and thus has little probative value.7
The more comprehensive and reliable model constructed by Harvard’s expert, Dr. Card,
found no statistically significant effect of Asian-American ethnicity on the likelihood of
admission to Harvard, either in any given admissions cycle or on average across all six cycles the
experts analyzed. DX685 & DD10.30; 10/30 Tr. 132:24-134:11 (Card). And Dr. Card found a
slightly positive (though still statistically insignificant) effect of Asian-American ethnicity on the
likelihood of admission for two groups of Asian-American applicants—female applicants and
applicants from California—who together comprise nearly two thirds of Asian-American
applicants. DX686.2-4 & DD10.31-32; 10/30 Tr. 138:19-139:12 (Card).
SFFA’s efforts to rehabilitate Dr. Arcidiacono’s analysis, and denigrate Dr. Card’s, are
unpersuasive.

7

The outcome-driven nature of Dr. Arcidiacono’s analysis is highlighted by the fact that,
in the analysis he conducted for SFFA’s case against the University of North Carolina (UNC), he
insisted that it was essential to include UNC’s equivalent of the personal rating in the model even
though he believed that rating to be affected by race—which is exactly the reason he gave for
excluding the personal rating from his model in this case. See Rebuttal Expert Report of Peter S.
Arcidiacono 3, 8 n.3 (Apr. 6, 2018) (“Arcidiacono UNC Rebuttal”), Students for Fair
Admissions, Inc. v. University of North Carolina, No. 14-cv-954 (M.D.N.C.) (Dkt. 154-26); see
also, e.g., Maher v. Hyde, 272 F.3d 83, 86 n.3 (1st Cir. 2001) (“‘It is well-accepted that federal
courts may take judicial notice of proceedings in other courts if those proceedings have relevance
to the matters at hand.’”); Barnstable County v. 3M Co., 2017 WL 6452245, at *4 (D. Mass.
Dec. 18, 2017) (filings in another case are judicially noticeable as to the claims and arguments
made by the parties).
- 21 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 29 of 78

a)

Dr. Arcidiacono’s regression analysis of the preliminary overall
rating is not probative of discrimination in admissions outcomes

SFFA begins (¶¶ 44-48) by discussing Dr. Arcidiacono’s regression analysis of the
preliminary overall rating, which purports to show that Asian-American ethnicity negatively
affects applicants’ ratings. But that analysis does not advance SFFA’s claim for two reasons.
First, as SFFA emphasized in its closing argument, the claim in this case is that AsianAmerican applicants suffered discrimination in admissions outcomes. See 11/2 Tr. 117:16-18
(“[W]e don’t see discrimination[] in admissions outcomes [for Asian-American ALDC
applicants]. That’s what discrimination means. You were denied admission.”). The statistical
means of examining whether Asian-American applicants suffered discrimination in admissions
outcomes is a regression analysis of admissions outcomes—not a regression analysis of the
preliminary overall rating.8
Second, and relatedly, the preliminary overall rating is simply a preliminary assessment
of the overall strength of the application—in other words, how likely it is that the applicant will
be admitted—based on whatever materials are available at the time the rating is assigned.
Harvard ¶¶ 39-40, 48. It is effectively superseded by the admissions decision actually rendered
at the conclusion of the process, which reflects the full Admissions Committee’s examination of
the entire application file, including materials received after the preliminary overall rating was
8

SFFA quotes Lam v. University of Hawaii, 40 F.3d 1551, 1560-1561 (9th Cir. 1994), for
the proposition that a plaintiff “‘need not prove intentional discrimination at every stage of the
decisionmaking process; impermissible bias at any point may be sufficient to sustain liability.’”
SFFA ¶ 183. That is true but irrelevant; neither Lam nor any other case means that a
discrimination plaintiff can prevail without providing evidence of discrimination in the ultimate
outcome. In Lam, for example, the Ninth Circuit explained that the reason it mattered whether
two participants in a faculty hiring process were biased was that, given the nature of the process
in question, “even a single person’s biases [could] be relatively influential” in the ultimate hiring
decision. 40 F.3d at 1560. Moreover, the preliminary overall rating is not a “stage” of the
admissions process; it is simply a preliminary estimate of the overall strength of the application.
See infra pp. 22-23, 28-29.
- 22 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 30 of 78

assigned. Harvard ¶¶ 39-40, 48, 52-55. It therefore makes no sense to analyze the preliminary
overall rating, a prediction of the likelihood of admission, as distinct from the applicant’s actual
admissions outcome. As Dr. Card explained, it would be like trying to assess the effect of a
given factor on the likelihood of retirement by analyzing a supervisor’s prediction of whether an
employee would retire in the next year, rather than analyzing whether the employee actually did
retire. 11/1 Tr. 70:22-25.
b)

Dr. Arcidiacono was wrong to exclude the personal rating

SFFA next argues (¶¶ 49-58, 63-68) that the personal rating is influenced by race and that
Dr. Arcidiacono was therefore correct to exclude it from his analysis of admissions outcomes.
That argument fails for two reasons: (1) the evidence did not show that the personal rating is
affected by an applicant’s race, and (2) even if it were, the proper approach would not be to
exclude it.
(1)

The personal rating is not affected by race

Admissions officers testified credibly and uniformly that they do not consider the fact of
an applicant’s race in assigning the personal rating.9 Harvard ¶ 65. That longstanding practice is
now formally codified in the Admissions Office’s reading procedures. Harvard ¶¶ 72-73. And
the veracity of the officers’ testimony is bolstered by the fact that the average difference in
personal ratings between White and Asian-American applicants persists among ALDC
applicants—a group for whom SFFA and Dr. Arcidiacono make no claim of discrimination.
Harvard ¶ 142. SFFA again fails to explain why, if the average difference in personal ratings

9

Dr. Card based his decision to include the personal rating in his model largely on those
admissions officers’ similar statements in their depositions—not, as SFFA wrongly asserts,
simply on a confirmatory phone call with Dean Fitzsimmons. 11/1 Tr. 101:14-102:2.
- 23 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 31 of 78

between White and Asian-American applicants actually reflected discrimination on the basis of
race, that difference would persist among ALDC applicants.
SFFA argues (¶ 49) that, contrary to the testimony at trial, the personal rating must be
influenced by the applicant’s race because Dr. Arcidiacono’s regression analysis of the rating
shows a statistically significant negative effect of Asian-American ethnicity. In effect, SFFA
argues, Dr. Arcidiacono’s results are so clear that the many admissions officers who testified
they do not consider race in the personal rating must all have been either untruthful or unaware
of their intent to discriminate on the basis of race. That would be a serious charge to level even
if the statistical analysis were clear.
But in fact, as Dr. Card explained, Dr. Arcidiacono’s regression does not prove race
affects the personal rating. Harvard ¶¶ 134-141. Dr. Arcidiacono arrives at the contrary
conclusion largely on the basis of his regression model of the personal rating, which estimates a
negative effect of Asian-American ethnicity. But both experts agree that regression estimates
cannot always be interpreted as reflecting the actual effect of the factor in question; they can be
interpreted that way only if one can rule out the possibility of omitted variable bias—i.e., the
absence from the model of factors that both correlate with the variable of interest and affect the
outcome. 10/25 Tr. 102:23-103:8 (Arcidiacono); 10/30 Tr. 113:1-14 (Card). That is why, when
Dr. Arcidiacono found a positive effect of Asian-American ethnicity in his regression models of
the academic and extracurricular ratings, he did not conclude that Harvard was biased in favor of
Asian-American applicants; rather, he reasoned that the positive estimate could be explained by
factors outside the model. Harvard ¶ 135. Yet he arrived at the opposite interpretation when his
regression model of the personal rating estimated a negative effect of Asian-American ethnicity.

- 24 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 32 of 78

Dr. Arcidiacono tried to justify his inconsistent interpretations of the ratings regressions
by arguing that Asian-American applicants are stronger than White applicants on factors in the
data that inform the personal rating—which, SFFA contends, would make it unlikely for factors
outside the model to explain the estimated negative effect. See Harvard ¶¶ 136-137. SFFA now
repeats that assertion (¶ 52). But it is simply not true that Asian-American applicants are
stronger than White applicants on factors in the data that inform the personal rating. Among
applicants with the same academic rating, Asian-American applicants were on average weaker
than White applicants on ratings assigned by alumni interviewers for personal qualities and
overall strength, as well as ratings assigned by admissions officers to recommendation letters
written by high-school teachers and guidance counselors, all of which inform the personal rating.
Harvard ¶ 139; DX692; DD10.68-75.
SFFA bases its contrary assertion (¶ 52) on one of Dr. Arcidiacono’s slides, PD38.33,
which shows that Asian-American applicants were slightly stronger than White applicants across
the full range of factors included in Dr. Arcidiacono’s personal rating model. But that analysis
includes academic factors, even though the personal rating is not based on academic factors
(Harvard ¶ 46) but rather is more or less uncorrelated with grades and test scores (DD10.65;
10/31 Tr. 49:4-52:9 (Card)). And Dr. Arcidiacono’s slide is limited to non-ALDC applicants,
see PD38.33 (relying on Dr. Arcidiacono’s “Baseline Dataset”), which has the effect of
comparing Asian-American applicants to a subset of White applicants that excludes many of the
strongest such applicants. 10/31 Tr. 71:24-72:7, 73:19-25 (Card). Dr. Card showed that
applications submitted by Asian-American applicants were less strong than those submitted by
White applicants across the full range of non-academic factors in the model, even when the
personal rating and ALDC characteristics were excluded from the list of non-academic factors.

- 25 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 33 of 78

10/31 Tr. 70:9-71:20; DD10.77-78. Again, Dr. Arcidiacono’s analysis on this point reflects
methodological choices calculated to reach a desired result, rather than an evenhanded attempt to
model the process.
SFFA goes on to argue (¶ 55) that the higher school support ratings of White applicants
cannot explain the average differences in personal ratings between White and Asian-American
applicants, because school support ratings are included as control variables in Dr. Arcidiacono’s
regression model. But neither Harvard nor Dr. Card has argued that the disparity in personal
ratings is fully explained by factors in the data.10 The relevant question is whether the portion of
the disparity that is not explained by factors in the data can be explained only by racial bias (as
Dr. Arcidiacono argues) or, instead, whether it is likely attributable to factors outside the data (as
Harvard argues). The reason it matters that White applicants were stronger than Asian-American
applicants on factors in the data that affect the personal rating is that it suggests they are also
stronger on factors outside the data that affect the personal rating, such that their strength on
those unobserved factors likely explain the results of Dr. Arcidiacono’s personal rating
regression. See Harvard ¶ 138; SFFA ¶ 45 (“A general principle of economics is that if a group
is strong on observable characteristics (criteria the dataset captures), they tend to be strong on
unobservable characteristics (criteria the dataset does not capture).”).
It is simply not plausible that admissions officers rating the same applicant’s academic,
extracurricular, and personal strengths discriminate in favor of Asian-American applicants on the
first two ratings only to discriminate against them on the third.11 Yet that is what Dr.

10

Indeed, factors in the data explained only 29% of the variation in personal ratings under
Dr. Arcidiacono’s model. DD10.61; 10/31 Tr. 40:14-43:25 (Card).
11
Again, Dr. Arcidiacono himself recognized this point in the UNC case: He wrote that
“[w]e would not expect UNC to, for example, penalize African-American applicants in one of
the ratings and then give them a bonus later for admission.” Reply Expert Report of Peter S.
- 26 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 34 of 78

Arcidiacono’s regressions would suggest if the estimated effects of Asian-American ethnicity
were interpreted as reflecting actual effects of race, as opposed to factors not captured by the
model. Rather, the explanation Dr. Arcidiacono gave for the academic and extracurricular rating
regressions—that the estimated effect of race actually reflected factors not captured in the data—
explains the personal rating regression as well.
(2)

Even if the personal rating were affected by race, the right
approach would not be to exclude it

Even if the personal rating were affected by the race of the applicant, the right approach
would not be to remove the rating entirely from the model but rather to adjust it to remove the
estimated effect of race. See Harvard ¶ 143. That allows the model to consider the considerable
amount of information captured in the personal rating that is otherwise unavailable in the data
and does not reflect any effect of race. See id.; 10/31 Tr. 78:10-22 (Card); 11/1 Tr. 100:16-101:2
(Card).
SFFA repeatedly insists that Dr. Card agreed it would be appropriate to exclude the
personal rating if it were affected by race. See SFFA ¶¶ 43, 58, 63. That too is simply not true.
To the contrary, when SFFA asked Dr. Card on cross-examination to agree that a rating “should
not be included in” the model if it “may be influenced by an applicant’s race,” Dr. Card rejected
that simplistic characterization of his views. 11/1 Tr. 78:5-9. He explained that it is proper to
exclude only those ratings (like the preliminary overall rating) that can directly reflect the
applicant’s “race per se,” as opposed to ratings (like the personal rating, as admissions officers

Arcidiacono 27, Students for Fair Admissions, Inc. v. University of North Carolina, No. 14-cv954 (M.D.N.C. Jan. 18, 2019) (Dkt. 160-3). That is why he tries in this case to rationalize his
inconsistent interpretations of the three ratings regressions—because interpreting the regressions
consistently would mean either that admissions officers act in the implausible way he rejected in
the UNC case, or that they do not actually consider race in any of the academic, extracurricular,
and personal ratings and thus that the personal rating is properly included in the model.
- 27 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 35 of 78

testified) that do not consider the applicant’s race as such but may reflect the applicant’s life
experiences associated with his or her race. Id. at 78:22-79:14; see also 10/31 Tr. 83:1-21
(drawing the same distinction); 10/17 Tr. 226:24-227:15 (Fitzsimmons); 10/19 Tr. 254:22255:24 (McGrath). As discussed above, Dr. Card explained that the right approach for ratings
that might indirectly reflect the applicant’s race is not to exclude the ratings but rather to adjust
them to eliminate the estimated effects of race. When Dr. Card adjusted the academic,
extracurricular, and personal ratings for the effects that Dr. Arcidiacono’s models attributed to
race, and then ran his regression of admissions outcomes using the adjusted ratings in place of
the actual ratings, he continued to find no statistically significant effect of Asian-American
ethnicity on the likelihood of admission. Harvard ¶ 144.12 Dr. Arcidiacono performed no such
analysis; he was eager to throw out the personal rating entirely, while maintaining the ratings that
(according to his own analysis) favored Asian-American applicants.
SFFA objects (¶ 66) that Dr. Card did not adjust the preliminary overall rating in the
same way as he adjusted the academic, extracurricular, and personal ratings. But Dr. Card and
Dr. Arcidiacono agreed that the preliminary overall rating should be excluded from the
admissions model because it may directly reflect the applicant’s self-identified race or ethnicity,
see 10/26 Tr. 77:14-16 (Arcidiacono), so there was nothing in the model to adjust. And whether
or not Dr. Card calculated a race-adjusted preliminary overall rating, it would have been
improper to include such a rating in a model of admissions outcomes, because as noted above,
the preliminary overall rating is simply an evaluation of the applicant’s overall strength—in

12

One of the amicus briefs filed on SFFA’s behalf twice states that Dr. Card adjusted the
ratings for “only Asian-American applicants.” Dkt. 624 at 12; see id. at 13-14. That is not
correct. Dr. Card adjusted the ratings for applicants of all races. See 10/31 Tr. 80:13-21; DX694
& DD10.83.
- 28 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 36 of 78

essence, a preliminary assessment of how likely it is that the applicant will be admitted. See
supra pp. 22-23; Harvard ¶ 48.
SFFA also objects (¶ 66) that Dr. Card “does not believe that the academic and
extracurricular ratings are influenced by race.” But neither does he believe the personal rating is
influenced by race; rather, his view—supported by the uniform testimonial evidence—is that
none of the profile ratings is actually affected by race. What Dr. Card explained is that if one
interprets the negative effect of Asian-American ethnicity in Dr. Arcidiacono’s personal rating
regression as reflecting as a causal effect of race, then the positive effects estimated by his
models of the academic and extracurricular ratings should be interpreted in the same way—and
in that case, all three ratings should be adjusted to remove the estimated race effects. 10/31 Tr.
78:10-79:2, 80:15-21; DX694 & DD10.83. The point is that it is inconsistent and statistically
unsound to interpret only the personal rating regression as reflecting a genuine effect of race.
Finally, SFFA objects (¶ 66) that Dr. Card “did not calculate a hypothetical race-free
rating for any of the school support ratings.” But even Dr. Arcidiacono did not claim the school
support ratings were tainted by race in a way that precluded including them in the model, and
indeed he chose to include the ratings in his model of the admissions process. 10/25 Tr. 101:23102:10.
c)

Dr. Arcidiacono was wrong to exclude ALDC applicants

Dr. Arcidiacono was equally incorrect to exclude ALDC applicants from his model.
ALDC applicants participate in the same admissions process as do all other domestic
applicants.13 10/17 Tr. 205:5-212:17, 214:16-19 (Fitzsimmons); 10/18 Tr. 73:17-24, 110:2513

SFFA is incorrect that the Dean’s and Director’s interest lists are reserved for “children
of donors or potential donors” (SFFA ¶ 14); those lists include promising students who come to
Dean Fitzsimmons’s or Director McGrath’s attention in any context, such as those they may
meet during recruiting events around the country. 10/17 Tr. 66:12-15 (Fitzsimmons); 10/18 Tr.
- 29 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 37 of 78

111:6 (Fitzsimmons). The proper approach is therefore to include them in a model of the
admissions process and to control for their ALDC attributes, like all their other characteristics.
10/30 Tr. 153:6-154:8, 158:10-14 (Card). A model that excludes ALDC applicants cannot
account for a significant part of the admissions process or for 30% of the admitted class, and it is
therefore unreliable. Harvard ¶¶ 110-116.
SFFA argues (¶ 73) that including ALDC applicants in the model distorts the model’s
estimates of the effects of certain factors, such as the academic rating. But there is no basis to
infer, as SFFA suggests, that the inclusion of ALDC applicants in Dr. Card’s model masked any
evidence of discrimination against Asian-American applicants. To the contrary, when Dr. Card
excluded from his model applicants who were recruited athletes—the group of ALDC applicants
who most differ from the overall average in their rate of admission, PD38.2, and for whom
academic factors are supposedly least relevant—he continued to find no statistically significant
effect of Asian-American ethnicity, either in any admissions cycle or on average across the six
cycles for which data were produced. DD10.42; see 10/30 Tr. 164:23-166:18 (Card). SFFA has
no response to that analysis.
Nor does SFFA acknowledge Dr. Arcidiacono’s prior statement that it would be proper to
deal with the effects of ALDC characteristics in the admissions process simply by allowing the
effects of those attributes to vary by race, see 10/30 Tr. 161:22-164:22 (Card) (addressing Dr.

72:25-73:16 (Fitzsimmons). Nor has SFFA ever tied admissions consideration given to children
of donors to discrimination against Asian-American applicants. See 10/18 Tr. 75:9-12
(Fitzsimmons) (“Q. Has anybody ever suggested to you that considering donors or whether folks
have been donors has somehow disadvantaged Asian-Americans? A. No.”); 10/25 Tr. 190:16-22
(Arcidiacono) (“Q. Let me add one more group that has come up during the course of the
evidence. You have ma[d]e no claim that Harvard discriminates against Asian-Americans who
are the children of donors, correct? A. Well, to the extent that they’re on the dean’s or director
list, true.”).
- 30 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 38 of 78

Arcidiacono’s statement that “[i]t is … essential to either … remove [ALDC] applicants from the
analysis; or … allow for the possibility that the effect of race is different for these applicants,
i.e., interacting these variables with race” (emphases added)). When Dr. Card took Dr.
Arcidiacono’s suggestion and ran a version of his model in which he allowed the effect of ALDC
attributes to vary by race, he continued to find no evidence of discrimination. Id.; see DD10.41.
Again, SFFA has no response.
SFFA argues (¶ 76) that it is proper for Dr. Arcidiacono to exclude ALDC applicants, on
the theory that the Department of Education’s Office for Civil Rights (OCR), Harvard’s Office
of Institutional Research (OIR), and the Admissions Office had all done so. Again, SFFA
misstates or misunderstands the facts.
OIR’s regression analysis included ALDC applicants—which is why, for example, the
results shown for that regression indicate the estimated effects of legacy status or an athletic
rating of 1 (assigned to recruited athletes). P26 at HARV00023550. And OCR ran two
regression analyses—one that included legacy applicants and recruited athletes, and one that
excluded them—to discern the effect of legacy and athletic consideration in the admissions
process; OCR did not simply exclude legacy applicants and recruited athletes as Dr. Arcidiacono
did. P555 at 33-35.
Separately, OCR, OIR, and the Admissions Office all calculated certain descriptive
statistics (as opposed to regressions) for the “non-legacy, non-athlete” portion of the applicant
pool. P9 at 5 (OIR); P555 at 36-38 (OCR); DX42.4-9 (Admissions Office); 10/17 Tr. 71:1-12
(Fitzsimmons); 10/18 Tr. 101:7-102:18 (Fitzsimmons); 10/19 Tr. 121:5-21 (Driver-Linn). But as
Dr. Card explained, that is not remotely akin to removing legacy applicants and recruited athletes
from a regression model. 10/30 Tr. 157:17-158:18. It is simply a crude way of controlling for

- 31 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 39 of 78

those attributes by looking at outcomes for applicants who possess the attributes (i.e., legacy
applicants or recruited athletes) and separately for those who do not. Id. That is essentially the
same as including legacy and recruited-athlete applicants in a regression model and controlling
for their legacy or recruited-athlete attributes, as Dr. Card did. Id.
In any event, all agree that the ultimate consequence of Dr. Arcidiacono’s choice to
exclude ALDC applicants is that his model simply does not examine the presence or absence of
discrimination across Harvard’s entire applicant pool; rather, he limits his inquiry to the nonALDC portion of the applicant pool. For the reasons discussed above (at 15-16), that makes it
very difficult to infer from his findings that Harvard intended to discriminate against AsianAmerican applicants on the basis of their race, absent an explanation—which SFFA has never
offered—of why, if Harvard intended to discriminate on the basis of race, the supposed
discrimination would affect some Asian-American applicants but not others.
d)

SFFA’s other statistical arguments fail

SFFA offers a smattering of other statistical arguments: that Dr. Card was wrong to
analyze the data on a year-by-year basis; that he was wrong to include the intended career,
parental occupation, and staff interview variables; that he should have interacted race with
disadvantaged status; and that his results for Californian and female applicants are meaningless.
Those arguments uniformly misunderstand or mischaracterize the evidence.
First, Dr. Arcidiacono was incorrect to conduct his regression analysis on a pooled set of
data from six admissions cycles, rather than modeling each cycle separately and averaging the
results (as Dr. Card did). As Dr. Card explained, it is proper to model each cycle separately
because the admissions process is in fact conducted separately each year. See Harvard ¶¶ 117120. SFFA halfheartedly tries to defend Dr. Arcidiacono’s approach on the theory that “[a]
pooled model ensures a larger sample size, which increases the model’s statistical power.”
- 32 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 40 of 78

SFFA ¶ 79. But SFFA has no response to Dr. Card’s finding that the average of the effects from
his year-by-year models produced an overall estimate of the effect of race with slightly greater
power than Dr. Arcidiacono’s pooled approach. Harvard ¶ 119.
Second, Dr. Arcidiacono was also incorrect to exclude from his model the intended
career, parental occupation, and staff interview variables. Because those variables are correlated
with race and affect admissions outcomes, their omission creates omitted variable bias. See
Harvard ¶¶ 121-124, 127. Indeed, it appears that the only reason Dr. Arcidiacono excluded this
seemingly random list of variables is that the exclusion of all three—like Dr. Arcidiacono’s other
methodological choices—allowed his model to find a more negative effect of Asian-American
ethnicity on the likelihood of admission. See Harvard ¶¶ 121, 123.14
SFFA argues (¶¶ 81, 85) that the parental occupation and intended career variables are
“unreliable” because the data vary in “unexplainable ways” from year to year. But as Dr. Card
testified, the variations are not at all “unexplainable”; they are in fact commonplace in data of
this sort. They simply reflect the fact that the options available to applicants completing
standardized application forms can vary from year to year. Harvard ¶ 125. The ability to include
variables that vary in that manner from year to year is a virtue of Dr. Card’s choice to run his
model separately for each admissions cycle. Id.

14

Dr. Arcidiacono took a markedly different position on the issue of excluding variables in
the UNC case. There, he determined that including five disputed ratings variables would cause
his model to estimate a larger statistical effect of race (i.e., the result favorable to SFFA).
Arcidiacono UNC Rebuttal 3, 7-8 & n.3. Advocating for the inclusion of those variables, he
argued that “[i]t is important to account for all the relevant factors when determining the role of
race in admissions,” and warned that excluding the ratings variables would “improperly attribute
the effect of those ratings to race.” Id. at 7. He included the five rating variables despite finding
that African-American and Hispanic applicants received lower scores on four of the five ratings
(which he attributed to those applicants being “weaker” on those dimensions) and higher scores
on one of the ratings (which he attributed to their receiving “racial preferences”). Id. at 8 & n.4.
- 33 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 41 of 78

SFFA also suggests that “[i]ncluding too many control variables” in a regression model
“runs the risk of ‘overfitting’ the model, which can reduce its explanatory power.” SFFA ¶ 81;
see id. ¶ 85. Tellingly, though, SFFA does not argue that Dr. Card’s inclusion of the intended
career and parental occupation variables caused his model to be “overfitted,” only that some
models might hypothetically suffer from that problem. Nor does any evidence in the record
support the conclusion that Dr. Card’s model is overfitted. To the contrary, Dr. Card explained
that overfitting is a concern only when a model includes “spurious” variables, those that are not
“really part of the process” being modeled. 10/30 Tr. 179:11-23. That is not true of intended
career and parental occupation, which are considered in Harvard’s admissions process, as Dr.
Arcidiacono agreed. Harvard ¶ 124.
SFFA argues (¶ 88) that it is necessary to exclude the staff interview variable because
ALDC applicants are more likely to participate in staff interviews than other applicants. But all
applicants to Harvard College may request and obtain staff interviews, Harvard ¶ 126, and even
if ALDC applicants are more likely to participate in such interviews because of their ALDC
status, that would be no reason to exclude the variable. To the contrary, including and
controlling for both ALDC characteristics and the staff interview indicator would allow the
model to determine the marginal effect of participating in a staff interview, holding ALDC
characteristics equal. In other words, it would allow the model to determine whether an
applicant who participates in a staff interview is more or less likely to be admitted than an
otherwise identical (ALDC or non-ALDC) applicant who does not. Harvard ¶ 130. Excluding
the staff interview variable because ALDC applicants are more likely to participate in staff
interviews makes as little sense as excluding SAT scores because ALDC applicants are more
likely to have high scores.

- 34 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 42 of 78

Third, SFFA argues (¶ 90) that “Professor Card should have interacted race and
socioeconomically disadvantaged status.” But Dr. Card did run a version of his model that
interacted race with disadvantaged status—and when he did so, he continued to find no
statistically significant effect of Asian-American ethnicity, either in any admissions cycle or on
average across all six cycles in the analysis. DX699.
Finally, SFFA argues (¶ 93) that Dr. Card’s finding of positive effects of Asian-American
ethnicity for Californian and female applicants are dependent on his other modeling choices—for
example, his inclusion of ALDC applicants and the personal rating. That is true but ignores the
point of the analysis. As Dr. Card explained, the reason researchers often analyze subgroups of
the data is that doing so can help them “understand what’s driving the results.” 10/30 Tr. 136:819. Dr. Card therefore found it “reassuring” that he continued to find no evidence of
discrimination—indeed, the results were more favorable for Asian-American applicants—when
he examined these “two large subsets of students where there’s a significant Asian population,”
subgroups that together comprise 64% of Asian-American applicants to Harvard. 10/30 Tr.
136:8-139:12. Were Harvard truly discriminating against Asian-American applicants on the
basis of their race, those results would be exceptionally surprising.
E.

The Non-Statistical Evidence Does Not Show Discrimination

SFFA’s proposed findings on the non-statistical evidence supposedly supporting its
intentional discrimination claim highlight the dearth of such evidence. Notably, SFFA failed to
introduce testimony from any applicant who claims to have suffered discrimination, or to
introduce even a single application file—from SFFA’s members or anyone else—that contains
any indication that an applicant was denied admission even partly on the basis of his or her race.
Lacking evidence that Harvard discriminated against Asian-American applicants—which
is, of course, the relevant issue—SFFA instead expounds at length (¶¶ 94-95) on the history of
- 35 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 43 of 78

stereotypes and societal discrimination against Asian Americans. The historical account is both
true and deeply regrettable, but SFFA has no evidence linking Harvard’s admissions process to
that troubled history. Indeed, SFFA’s description of the supposed bias against Asian-American
applicants in college admissions (¶ 95) takes unpardonable liberties with the record. SFFA
repeatedly quotes—in the present tense—from a report co-authored by amicus witness Margaret
Chin, which describes concerns about the treatment of Asian-American applicants in college
admissions, without mentioning that the report was written in 1983. Nor does SFFA mention
Ms. Chin’s testimony that the report had to be viewed “in the context of the historical period” in
which it was written, 10/29 Tr. 51:10-11, or her testimony that the statements in question “may
not” have been “in particular about Harvard,” id. at 54:12, or her testimony “that AsianAmerican admissions has improved since 1983,” id. at 55:12-13, or—perhaps most glaringly—
her testimony that Harvard has in fact taken many steps to help ensure Asian-American
applicants do not suffer discrimination, id. at 57:19-58:3; 58:21-22.15
SFFA’s basic theory is that because Asian Americans have suffered discrimination in our
society, Harvard must prove it was an exception to that practice. But much as SFFA would like
for Harvard to be considered guilty until proven innocent, that is not the way our legal system
works.

15

SFFA’s references to Harvard’s history of discrimination against Jewish applicants also
cannot support its current allegations of discrimination against Asian Americans. Contrary to
SFFA’s claims (¶ 96), there is no evidence that “Harvard’s current admissions process”—
including the committee process in which applications are read and discussed by numerous
admissions officers at multiple levels of review, and with extensive training—bears any
resemblance to the process employed a century ago.
- 36 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 44 of 78

i.

Harvard takes pains to ensure that its consideration of race is not used to
discriminate

SFFA argues (¶¶ 95-98) that Harvard’s admissions process is a “subjective” one,
involving many factors not susceptible to rigid or formulaic application, and asks the Court to
draw the conclusion that because a subjective process could be used to discriminate, Harvard
must prove its process was not used to discriminate. As SFFA would have it (¶ 189), because
Asian Americans have been subjected to stereotyping and discrimination in our society—a point
SFFA makes on the basis of a law review article and a government report, both dating back more
than two decades—“[t]he issue is whether Harvard has uniquely escaped [the] infiltration” of
prejudice. But as discussed above, the law does not permit a Title VI plaintiff to prevail by
placing on the defendant the burden of showing it did not engage in forms of bias allegedly
prevalent in society at large. Rather, the “plaintiff must demonstrate” that the defendant itself
acted with “racial animus.” Goodman, 380 F.3d at 43. That is not, as SFFA appallingly suggests
(¶ 201), a “rule[] of litigation from the Jim Crow era”; it simply reflects the ordinary burden for
plaintiffs to prove the elements of their claims.
Not only does SFFA fail to offer evidence that Harvard’s admissions officers themselves
acted with “racial animus”; it discounts the extensive evidence that Harvard acts to ensure that its
process is not used to discriminate. That includes evidence of the Admissions Office’s extensive
training procedures, including a weeks-long orientation that includes guidance on considering
race in the admissions process. Harvard ¶¶ 66-67. It also includes evidence of the numerous
checks and balances in the admissions process, which ensure that multiple admissions officers
have the opportunity to review all applications and that all applicants are treated fairly. Harvard
¶¶ 51-55, 74. And it includes the consistent and credible testimony of the individuals involved in

- 37 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 45 of 78

the process that they have neither participated in discrimination nor witnessed it among their
colleagues. Harvard ¶¶ 199-201.
SFFA’s account of the way admissions officers consider race is flatly at odds with the
testimony elicited at trial. SFFA suggests (¶ 100) that admissions officers never received
training regarding the use of race, but the record shows the opposite: Admissions officers
receive training regarding the consideration of race during their orientation and throughout their
employment. Harvard ¶¶ 66-73. Similarly, SFFA’s suggestion (¶ 14) that Asian-American
applicants never receive a “tip” on the basis of their racial identification is false. For example,
Dean Fitzsimons explained that the compelling story of Thang Diep’s Asian-American heritage
contributed to the preliminary overall rating assigned to his application and ultimately to his
admission. 10/18 Tr. 60:12-61:17; SA-2.2; see also 10/23 Tr. 220:11-22 (Yong); DX527.0002.
SFFA claims that Harvard’s training has led to “inconsistencies in how Harvard’s
admissions officers use race.” ¶¶ 100-103. But the testimony revealed no such inconsistencies.
Admissions officers testified consistently and credibly that they consider race as one factor in the
individualized review of each applicant. Harvard ¶¶ 59-63.16 They testified consistently and
credibly that they do not consider the fact of an applicant’s race itself when assigning the
personal rating.17 Harvard ¶ 65. And they explained that an applicant’s race itself does not
inform the assignment of any of the profile ratings, id., but that experiences applicants may have

16

SFFA cites admissions officer Christopher Looby’s deposition testimony regarding the
use of race in the admissions process (¶ 102) but fails to note Mr. Looby’s explanation at trial
that he was attempting to convey that he “use[s] race as one factor of many within the
application with assessing an applicant’s overall rating.” 10/18 Tr. 184:1-185:23 (emphasis
added).
17
SFFA is also incorrect to assert (¶ 98) that “[a]dmissions officers received no instruction
at all” on whether to consider race in the personal rating. See 10/22 Tr. 164:15-24, 179:16-180:7
(McGrath); DX24.
- 38 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 46 of 78

had as a consequence of their race or ethnicity may be taken into account. See 10/17 Tr. 227:715 (Fitzsimmons); 10/19 Tr. 50:17-51:10 (Bever).
What SFFA identifies as inconsistent testimony instead reflects the flexible, nonmechanistic nature of Harvard’s admissions process—exactly the kind of process required by the
Supreme Court, Grutter v. Bollinger, 539 U.S. 306, 334 (2003). It is little surprise, given the
absence of any rigid formulas or rubrics in Harvard’s process, that admissions officers articulated
how they take race into account in slightly different ways, particularly because race can play a
different role in the consideration of different applicants. 10/18 Tr. 207:14-19 (Bever) (“Race
can play a different role in a different applicant’s case.”); 10/18 Tr. 169:11-13, 22:24 (Looby);
10/16 Tr. 25:23-26:8, 26:25-27:10 (Fitzsimmons); 10/24 Tr. 223:10-14 (Ray). That is hardly
improper, let alone evidence of intentional discrimination.
ii.

SFFA’s allegations of stereotyping are unsupported

SFFA argues (¶¶ 105-112) that Harvard’s Admissions Office stereotypes AsianAmerican applicants. SFFA grossly mischaracterizes the evidence on which it relies, however.
And the thinness of that evidence is itself an indictment of SFFA’s case. SFFA presented no
expert on stereotyping or unconscious bias, barely questioned admissions officers about those
subjects, and pointed to nothing in the hundreds of application files produced in this case that
would suggest that admissions officers actually stereotyped Asian-American applicants or
exhibited unconscious biases. Indeed, SFFA appears to have adopted this theory only well into
the trial, after realizing its evidence would not suffice to prove Harvard’s admissions officers
acted with racial animus. SFFA fails in its belated attempts to stretch the evidence to fill the
gaps in its case.
First, contrary to SFFA’s allegations (¶ 104), the Harvard Admissions Office has “a
longstanding practice” of reminding its staff to avoid implicit biases and racial stereotyping.
- 39 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 47 of 78

11/1 Tr. 145:24-146:1 (McGrath). Bias has been “a subject of training and a subject of staff
discussion” for decades (id. at 164:10-16), and admissions officers have received trainings
“designed to give context more generally about the experiences of communities of color and
students of color in the United States at large” (10/24 Tr. 227:14-22 (Ray); DX19; DX36). One
such training specifically “highlight[ed] the fact that there’s a lot of diversity within [the AsianAmerican] population” and emphasized the need for admissions officers to be wary “of
aggregated data for Asian American student populations” and to “[h]onor the nuance of both
identity and context.” 10/24 Tr. 229:9-18 (Ray); DX19.28.
Second, SFFA accuses Dean Fitzsimmons of “stereotyp[ing] Asian Americans as
perpetual foreigners” in his comments about sparse country states. SFFA ¶ 106. But that
accusation is groundless. SFFA bases it on the premise that Dean Fitzsimmons testified
“Harvard is more interested in recruiting students who ‘have lived [in sparse country states] for
their entire lives’ than those who ‘have only lived in the Sparse Country state for a year or two.’”
Id. But Dean Fitzsimmons did not say the Admissions Office was “more interested” in recruiting
certain types of students than others; all he said in the quoted passage of his testimony was that
the Admissions Office did not regard residence in a sparse country state as an automatic basis for
a tip, given the individualized differences among candidates. 10/15 Tr. 149:1-10 (“The fact that
a person applied from Sparse Country might be one of, again, many, many factors that could
possibly go into a decision. But there are people who, let’s say, for example, have only lived in
the Sparse Country state for a year or two. Let’s say that can happen. And then on the other
hand there are people who have lived there for their entire lives under very different settings. So
what we’re trying to make sure we do, in an even-handed way, is to reach out to what lots of

- 40 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 48 of 78

people would say is the heartland of America.”). Nor did his statement have anything even
arguably to do with Asian-American candidates.
Third, SFFA misstates the record when it claims (¶ 107) that Harvard has been “racially
stereotyping Asian Americans for decades.” SFFA invokes OCR’s 1990 Statement of Findings
as evidence of stereotyping, but ignores the findings that “[i]n some cases the [allegedly
stereotypical] comments actually originated from the interviewers, teacher or counselor
recommendations, or self-descriptions given by the applicant”—in other words, not from
admissions officers—and that the comments “could not be shown to have negatively impacted
the ratings given to these applicants.” P555 at 25, 26; see Harvard ¶ 155. OCR also found some
“evidence of readers’ sensitivity to the obstacles facing Asian Americans.” P555 at 26.
Following OCR’s investigation, the Admissions Office reviewed the Statement of Findings to
ensure the Admissions Committee did not engage in racial stereotyping and that it reviewed each
application individually. 10/16 Tr. 62:5-63:10 (Fitzsimmons). Dean Fitzsimmons testified
credibly and forcefully that the Admissions Office “abhor[s] stereotypical comments.” 10/16 Tr.
63:16-21.
Fourth, SFFA points (¶ 108) to a handful of comments on a docket binder—statements
that particular applicants were, for example, “very quiet” or “quiet and strong”—that SFFA tries
to paint as stereotypical. That docket binder included information on thousands of applicants,
from which SFFA drew a tiny handful of comments. It is unclear what, exactly, the isolated,
handwritten comments even mean, including whether they refer to the writer’s own assessment
of the applicant as opposed to, for example, the writer’s notation of comments actually made by
the applicant’s recommenders or interviewer, or even the writer’s notation of a self-description
offered by the applicant himself or herself. In contrast to OCR’s painstaking review, SFFA

- 41 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 49 of 78

made no effort to show that even those few comments actually reflected stereotypes as opposed
to the fair assessment of individual applicants. And there is no evidence to suggest that the
statements were a product of racial stereotyping or unconscious bias. Dean Fitzsimmons
testified that he has described candidates of all racial backgrounds as “quiet” and has described
Asian-American candidates as “outgoing.” 10/18 Tr. 129:8-21.
Fifth, SFFA argues (¶¶ 110-111) that Asian-American applicants were more likely to be
labeled “standard strong” than White applicants. But as Dr. Card explained, the White and
Asian-American applicants labeled “standard strong” were in fact “virtually identical” in their
overall strength, as measured by the sum of their profile ratings, as opposed to just the narrow
academic qualifications Dr. Arcidiacono considered. 10/31 Tr. 94:15-97:17; DX709. In other
words, Asian-American applicants labeled “standard strong” were not stronger than White
applicants described that way.
Sixth, SFFA argues that the Admissions Office must be stereotyping because of
Harvard’s arguments at trial that Asian-American applicants were less multidimensional than
White applicants. ¶ 112. What Harvard and its expert were referring to is the unquestioned
evidence that Asian-American applicants were less likely than White applicants, across the six
admissions cycles for which data were produced, to have three or more profile ratings (academic,
extracurricular, athletic, or personal) of 2 or better—a characteristic associated with considerably
higher rates of admission. See DD10.9 (applicants with three or more strengths by race);
DD10.6 (significant difference in admission rate for applicants with three or four strengths
relative to applicants with one or two strengths); 10/30 Tr. 90:18-92:12, 95:14-96:14 (Card).
Those observations were not based on stereotyping; they reflected careful data analysis.

- 42 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 50 of 78

iii.

The Class of 2023 reading procedures are not evidence of discrimination

SFFA next relies on the updated reading procedures that the Admissions Office issued for
the Class of 2023 admissions cycle, consistent with its regular practice of updating the
procedures for each annual cycle. But SFFA never explains why the reading procedures
establish liability—perhaps because they are plainly evidence in Harvard’s favor.
According to SFFA (¶ 114), the revised reading procedures were meant to address a
number of alleged problems with the admissions process. But Harvard witnesses testified
without contradiction that the reading procedures reflected not a change in policy but rather a
codification of existing policy. 10/24 Tr. 206:25-207:6 (Ray); 11/1 Tr. 176:25-177:23
(McGrath). SFFA’s argument (¶ 114) that the revised reading procedures cured problems in the
admissions process is therefore self-defeating: In fact, the revised procedures—combined with
the testimony that they merely codified existing policy—establish that the supposed problems
never existed in the first place. SFFA also claims (¶ 113) that its motion for summary judgment
caused admissions officers to suggest changes to the reading procedures. But the Admissions
Office’s annual staff retreat to which SFFA refers took place during the first week of June,
before SFFA filed its motion for summary judgment on June 15. 11/1 Tr. 129:23-130:5
(McGrath); P696.
iv.

OIR documents do not show intentional discrimination

SFFA turns next to documents prepared by OIR. After the evidence at trial made clear
that SFFA had entirely misunderstood both the nature of the OIR documents and the chronology
by which they were prepared and some of them shared with others at Harvard, SFFA tries to
salvage its arguments by reciting the version of the facts it wishes it had elicited at trial. But that
imaginative reconstruction cannot be reconciled with the evidence, which showed that neither
the OIR documents themselves nor the response to them showed any intent to discriminate on
- 43 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 51 of 78

the basis of race. Harvard ¶¶ 160-188. And SFFA’s legal arguments concerning the relevance
of the OIR documents would be unpersuasive even if its factual arguments were more accurate.
First, SFFA mischaracterizes the record in arguing (¶ 117) that “Harvard instructed OIR”
to investigate whether the admissions system disadvantages Asian-American applicants or that
OIR performed its work “with the assistance of the Admission Office.” The OIR documents
introduced at trial were not part of an investigation into discrimination against Asian-American
applicants—in fact, the documents did not even seek to assess that question—and did not reflect
input from the Admissions Office. 10/19 Tr. 123:9-16, 129:10-14, 136:13-137:3, 145:15-147:7
(Driver-Linn); 10/24 Tr. 48:7-12, 54:24-55:1, 64:1-6 (Hansen); see also P14 (emails from OIR
Director Erin-Driver Linn to Admissions Office leaders, offering to share analyses, and from
Admissions Office leaders accepting that offer). OIR conducted a separate, limited analysis
relating to the discrimination claims made by Ron Unz’s magazine article, but that analysis was
done at the direction of the Office of the General Counsel, and the resulting documents were
accordingly privileged and not introduced at trial. Order on Mot. to Compel (Dkt. 383) at 2.
Second, SFFA mischaracterizes both the substance and the process of OIR’s analysis. In
January 2013, OIR employee Mark Hansen created a set of his working notes relating to
admissions data. P9; Harvard ¶¶ 174-177. Mr. Hansen’s working notes were not a final work
product of OIR—let alone a report—and were not intended to be given as a presentation.
Harvard ¶ 176. And SFFA’s statement that “OIR gave this report to Fitzsimmons” (¶ 118) is
flatly contradicted by the record. SFFA made this same claim in its opening statement, 10/15 Tr.
29:17-31:6, but the evidence at trial was to the contrary: OIR did not show Mr. Hansen’s
working notes to Dean Fitzsimmons, 10/19 Tr. 113:14-114:19 (Driver-Linn); Harvard ¶ 177. In
fact, nobody other than Mr. Hansen saw those notes before this litigation. Harvard ¶ 177.

- 44 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 52 of 78

SFFA next asserts that the supposed “report”—Mr. Hansen’s working notes never seen
by anybody else—“found ‘evidence that Asians are disadvantaged in the admissions process.’”
SFFA ¶ 118. That, too, mischaracterizes the record. As Mr. Hansen explained, his working
notes are “not necessarily a comprehensive look at all the factors considered in admission” and
do not “establish causal relationships between” any admissions factor and “admissions
outcome[s].” 10/24 Tr. 46:19-47:11.
SFFA also mischaracterizes the four-model analysis created by OIR that was presented to
Dean Fitzsimmons. The four models neither sought to simulate the actual admissions process
nor estimated the effect of any particular factor on an applicant’s chances of admission. Harvard
¶¶ 161-168. And Model 4—on which SFFA has relied to suggest that Asian-American ethnicity
was estimated to have a negative effect on applicants’ likelihood of admission—is circular: It
uses the racial composition of the actual admitted class as a basis to predict the racial
composition of the simulated admitted class, which is why the predicted composition aligns so
well with the actual composition. Harvard ¶ 164.
SFFA ignores the record evidence in suggesting (¶ 120) that Dean Fitzsimmons
understood the four-model analysis to show that Asian-American applicants were penalized in
the admissions process. Dean Fitzsimmons did not understand the four-model analysis to show
bias or discrimination against Asian-American applicants (Harvard ¶¶ 172-173), and no witness
suggested he should have interpreted it that way. Rather, he understood the OIR analysis to be
consistent with his understanding of the admissions process in the sense that as the models
included additional factors considered in the process, the racial composition of the simulated
class more closely reflected that of the actual class. Harvard ¶ 172. Dean Fitzsimmons’s
understanding that the analysis did not show discrimination is consistent with the testimony of

- 45 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 53 of 78

every OIR witness. 10/19 Tr. 18:8-15 (Bever), 136:17-20, 146:25-147:3, 154:18-155:8 (DriverLinn); 10/24 Tr. 65:17-66:4 (Hansen). Dean Fitzsimmons therefore did not believe there was
any basis for him to act in response to the analysis; he did not understand the analysis to show
anything that should have caused him concern, nor did those more familiar with the analysis
suggest there was a cause for concern.
SFFA also makes much (¶ 123) of internal OIR drafts of a memorandum addressing the
statistical effect of an admissions tip to low-income applicants. But neither Dean Fitzsimmons
nor anyone else outside OIR ever saw those drafts. 10/19 Tr. 40:18-20 (Bever), 152:1-3 (DriverLinn). And although SFFA places great emphasis on an email exchange during the drafting
process between Erin Driver-Linn of OIR and Christine Heenan of Harvard’s Public Affairs and
Communications Office, that exchange hardly has the sinister meaning SFFA ascribes to it. Ms.
Driver-Linn contacted Ms. Heenan simply because she believed Dean Fitzsimmons might
publicize OIR’s findings and wanted to ensure that the findings were not taken out of context—a
concern that proved reasonable given how SFFA has mischaracterized OIR’s work. 10/19 Tr.
150:21-151:1 (Driver-Linn).
Dean Fitzsimmons received a later version of the memo on May 1, 2013. Harvard ¶ 181.
Contrary to SFFA’s claim that Dean Fitzsimmons did not request follow-up, he in fact did
request additional analysis on whether the low-income tip was being applied “in an evenhanded
manner for students from all ethnic backgrounds” and, specifically, whether low-income AsianAmerican applicants received a tip. Harvard ¶ 183. SFFA misleadingly suggests (¶ 126) that
Dean Fitzsimmons understood OIR’s follow-up analysis to show a negative effect of AsianAmerican ethnicity on the likelihood of admission to Harvard. In fact, he did not. Dean
Fitzsimmons explained that he understood from a graph in the follow-up report “that everybody

- 46 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 54 of 78

gets some kind of a tip for being low income versus not” based on the fact that across all racial
groups the admission rate for students who were low-income was higher than the rate for
students who were not. 10/17 Tr. 143:21-144:10; P28 at 6.18 When asked about the tables of
coefficients that appear later in the report, Dean Fitzsimmons explained that he did not “fully
understand” those statistics and that he was “not sure how to interpret” that analysis. 10/17 Tr.
143:12-20.19 Dean Fitzsimmons did not understand the follow-up analysis—or any of OIR’s
analysis—to show bias or discrimination against Asian-American applicants; there was thus no
reason for him to request still further analysis or to take any other action. Harvard ¶¶ 182-184.
Finally, SFFA’s legal arguments concerning the relevance of the OIR documents
(¶¶ 195-196) would be unpersuasive even if SFFA’s factual arguments were accurate. SFFA
first asserts (¶ 195) that Dean Fitzsimmons’s supposedly inadequate response to the OIR
documents “is powerful evidence of intentional discrimination” on the theory that it is
impermissible to “‘[a]dher[e] to a … policy or practice’ … ‘with full knowledge of the
predictable effects of such adherence.’” For that proposition, SFFA quotes Columbus Board of
Education v. Penick, 443 U.S. 449, 465 (1979). But SFFA fails to mention Penick’s recognition
that “disparate impact and foreseeable consequences, without more, do not establish a
constitutional violation.” Id. at 464. Nor does it mention the Supreme Court’s holding in
Personnel Administrator of Massachusetts v. Feeney, 442 U.S. 256 (1979), the very same year,
that “‘[d]iscriminatory purpose’ … implies more than intent as volition or intent as awareness of

18

One of the amicus briefs submitted on SFFA’s behalf twice interprets the OIR analysis to
mean that low-income Asian-American applicants do not receive any benefit in the admissions
process from the fact that they are low-income. Dkt. 622 at 7, 8. In fact (as discussed above) the
OIR analysis shows the opposite.
19
Dean Fitzsimmons’s confusion is understandable. Even Ms. Bever—a former OIR
employee—could not be sure from the face of the document how the relevant coefficients should
be interpreted. 10/18 Tr. 234:6-235:18.
- 47 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 55 of 78

consequences”; rather, “[i]t implies that the decisionmaker … selected or reaffirmed a particular
course of action at least in part ‘because of,’ not merely ‘in spite of,’ its adverse effects upon an
identifiable group.” Id. at 279 (emphases added); see Harvard ¶ 282. Even if, contrary to all the
evidence, Dean Fitzsimmons had understood the OIR documents to show that Harvard’s
admissions process was disadvantaging Asian-American applicants, there would certainly be no
basis to conclude that he reacted as he did because he was glad to see Asian-American applicants
disadvantaged in that way.
SFFA next argues (¶ 196) that the deliberate-indifference standard applicable to
municipal-liability claims under 42 U.S.C. § 1983 should be imported into the Title VI context—
a proposition for which it cites exactly one out-of-circuit district court decision. If SFFA had
undertaken a more careful review of the (relatively few) cases applying a theory of deliberate
indifference in the Title VI context, it would have understood that courts have interpreted that
standard to require that the defendant actually knew about ongoing discrimination and chose not
to respond to it. See, e.g., Blunt v. Lower Merion Sch. Dist., 767 F.3d 247, 273 (3d Cir. 2014)
(“Constructive knowledge is not sufficient; ‘only actual knowledge is a predicate to liability.’”);
Zeno v. Pine Plains Cent. Sch. Dist., 702 F.3d 655, 666 (2d Cir. 2012) (similar). That follows
from the Supreme Court’s recognition of the same principle in the Title IX context. See Davis v.
Monroe Cty. Bd. of Educ., 526 U.S. 629, 642 (1999) (“[W]e declined the invitation to impose
liability under what amounted to a negligence standard—holding the district liable for its failure
to react to teacher-student harassment of which it knew or should have known. Rather, we
concluded that the district could be liable for damages only where the district itself intentionally
acted in clear violation of Title IX by remaining deliberately indifferent to acts of teacher-student
harassment of which it had actual knowledge.” (citation omitted) (citing Gebser v. Lago Vista

- 48 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 56 of 78

Indep. Sch. Dist., 524 U.S. 274, 283, 290 (1998))). The evidence in this case came nowhere near
showing that Dean Fitzsimmons or others actually knew about discrimination in the Admissions
Office and chose not to respond. SFFA interprets the OIR documents to mean at most that Dean
Fitzsimmons “should have known” about the possibility of discrimination—but even if that
interpretation were supported by the evidence, which it is not, it is not enough to establish
intentional discrimination. Davis, 526 U.S. at 642.
v.

Trends in the racial composition of the admitted class do not show
intentional discrimination

Finally, SFFA is wrong to argue (¶ 197) that recent trends in the racial composition of the
class show Harvard intended to discriminate in the past. As an initial matter, the growth in the
Asian-American proportion of the student body in recent years simply continues the long-term
trend of steady growth in the number and proportion of Asian-American applicants in the
admitted student pool. Harvard ¶¶ 79, 82-83; DX713.2. For example, the Asian-American share
of admitted students grew from 3.4% to 10.2% between the Classes of 1980 to 1990; from 10.2%
to 16.4% between the Classes of 1990 and 2000; from 16.4% to 17.6% between the Classes of
2000 and 2010; and from 17.6% to 20.6% between the Classes of 2010 and 2019 (the most
recent year analyzed by the experts). DX713.
Even if that growth trend were somehow evidence of remedial measures—and it is not—
the cases SFFA cites do not support its suggestion that the trend shows that Harvard intended to
discriminate in the past. Three of the decisions simply hold that summary judgment is
inappropriate; they do not say that the defendants’ actions were discriminatory. See Patterson v.
Strippoli, 639 F. App’x 137, 143 (3d Cir. 2016); Holcomb v. Iona Coll., 521 F.3d 130, 143-144
(2d Cir. 2008); Adorno v. Port Auth., 258 F.R.D. 217, 233 (S.D.N.Y. 2009). Another addresses
mootness, not the merits of a discrimination claim. Cypress v. Newport News Gen. &

- 49 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 57 of 78

Nonsectarian Hosp. Ass’n, 375 F.2d 648, 658 (4th Cir. 1967). And the last simply canvasses
evidence on which the district court had relied in finding discrimination, without explaining the
implications of the evidence. Payne v. Travenol Labs., Inc., 673 F.2d 798, 818-819 (5th Cir.
1982). None of those decisions, and none of the evidence in this case, supports SFFA’s
argument that the growth over time in the Asian-American portion of the undergraduate student
body is evidence of past discrimination.
F.

Harvard’s Witnesses Were Credible

SFFA called no fact witnesses of its own to support its claims, relying instead on the
testimony of 13 current and former Harvard employees whom it examined adversely. Faced
with the fact that those witnesses’ testimony supports Harvard’s position, not SFFA’s, SFFA
now attacks their credibility in broad-brush terms, largely on the theory that the testimony was
implausible. But as discussed above, it is SFFA—not Harvard’s witnesses—that has taken
liberties with logic and with the evidence. It is not the fault of Harvard’s witnesses that their
testimony does not fit the fictional narrative on which SFFA built its case.
First, although SFFA complains (¶ 130) that there could be “no coherent explanation” for
Dean Fitzsimmons’s response to the OIR analyses, Dean Fitzsimmons repeatedly and
consistently explained how the actions he took followed reasonably from his understanding of
OIR’s work. See supra pp. 45-47; Harvard ¶¶ 172-173, 182, 184; 10/17 Tr. 128:8-129:1, 137:38, 138:10-24, 145:5-8, 186:4-13 (Fitzsimmons). SFFA argues (¶ 130) that Dean Fitzsimmons
could not simultaneously have trusted OIR’s general abilities to “do good research” and produce
“robust and reliable work,” while also recognizing the limitations of specific OIR analyses. But
there is no inconsistency at all. The limitations of OIR’s work reflected the gaps in OIR’s
understanding of the admissions process (of which Dean Fitzsimmons had plainly superior
knowledge) and in the available data. Those limitations were acknowledged on the face of the
- 50 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 58 of 78

documents and by the OIR witnesses themselves. Harvard ¶¶ 161, 166, 171, 179, 180; 10/19 Tr.
148:15-149:11 (Driver-Linn); 10/24 Tr. 28:7-29:3, 33:6-18 (Hansen).
Nor are SFFA’s other scattershot accusations (¶ 131) any more persuasive. For example,
SFFA claims that Harvard has “has no credible explanation” for “why it needs to give
preferences to legacies and the children of donors and faculty,” but Dean Fitzsimmons, Dean
Smith, Harvard’s expert, President Simmons, and the committee on race-neutral alternatives
explained the institutional objectives served by considering an applicant’s relationship to alumni,
donors, or faculty. 10/17 Tr. 25:15-20, 209:19-211:8, 211:18-212:14 (Fitzsimmons); 10/23 Tr.
100:12-25 (Smith); 10/30 Tr. 36:8-40:11 (Simmons); P316 at 16-17.
Similarly, although SFFA claims (¶ 131) that “multiple witnesses” testified (in a manner
inconsistent with Dean Fitzsimmons’s testimony) that the personal rating “measure[s]” “an
applicant’s ‘personality,’” the record shows no such inconsistency. No witness testified that the
personal rating is a “personality rating,” only that the rating may consider “among many other[]”
things “what the applicant could add to the community … [i]n terms of personality and potential
impact.” 10/24 Tr. at 205:13-17 (Ray); see also id. at 190:19-22 (same); 10/18 Tr. at 180:9181:1 (Looby) (“a positive personality” “[c]ould be” among the characteristics valued in addition
to an ability to work well with others and create meaningful relationships).
SFFA next claims (¶ 132) that “[n]early all of Harvard’s witnesses had their testimony
impeached.” It is of course for the Court, not SFFA, to determine whether SFFA’s attempts to
impeach witnesses actually succeeded in showing inconsistencies between witnesses’ trial
testimony and their prior statements, let alone whether any such inconsistencies demonstrated a
lack of credibility. As the Court recognized during trial, there was often no inconsistency at all
between a witness’s trial testimony and the prior statement SFFA was allegedly using for

- 51 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 59 of 78

impeachment. See, e.g., 10/16 Tr. 23:22-24:17 (“I’m not sure it’s proper impeachment[.]”);
10/17 Tr. 57:4-10 (“[T]hat’s not inconsistent.”); 10/18 Tr. 117:13-21 (“I don’t think it’s strictly
inconsistent.”), 207:14-208:5 (“I don’t think it’s directly contradicting[.]”). To the extent the
examples SFFA now cites reflect any inconsistencies at all, they were insignificant,20 or
attributable either to a change in circumstance between the prior statement and the trial or to
witnesses’ having had time to reflect since their depositions.21 Such “minor inconsistencies …
are not uncommon when one person testifies at two different times, months apart, about the same
events.” Cruz v. Dollar Tree Stores, Inc., 2008 WL 2705087, at *2 (N.D. Cal. July 8, 2008); see
also, e.g., Skains v. Lockler, 2009 WL 230037, at *8 (E.D. Cal. Jan. 20, 2009) (“Inconsistencies
are not uncommon when witnesses testify at trial. Minor inconsistencies do not equal
untruthfulness.”), adopted, 2009 WL 806829 (E.D. Cal. Mar. 26, 2009), aff’d sub nom. Skains v.
California, 386 F. App’x 620 (9th Cir. 2010).
SFFA focuses in particular on admissions officer Christopher Looby’s testimony. Again,
however, SFFA mischaracterizes the record. As Mr. Looby himself observed, his testimony at
trial that he “do[es]n’t recall” whether the Class of 2019 reading procedures were “the first place
that [he] learned” to consider race in the admissions process was not inconsistent with his prior
testimony that there is such guidance “on the first page” of those reading procedures. 10/18 Tr.
170:1-171:7. SFFA also cites Mr. Looby’s deposition testimony—in which he was asked

20

See, e.g., 10/19 Tr. 91:22-93:1 (Driver-Linn) (testimony regarding author of document);
10/18 Tr. 205:21-206:25 (Bever) (testimony regarding practice in completing assignments).
21
See, e.g., 10/23 Tr. 206:24-207:2 (Yong) (“I’ve been thinking about it more carefully,
recalling my training. Things are coming back to me from the start of my career that I hadn’t
thought about in 26 years.”); 11/1 Tr. 229:1-5, 232:1-233:13 (Faust) (explaining that deposition
testimony reflected unfamiliarity with specific details of Harvard’s decades-past admissions
process, not disagreement that discrimination had historically occurred); 10/24 Tr. 15:25-17:1
(Hansen) (explaining that language describing OIR document adopted “the premise of the
[deposition] question, and it really narrowed” what he viewed as the scope of the document).
- 52 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 60 of 78

whether he would “take a student’s race into account when assigning him … any” of the
application ratings, and he answered that race was “one of [the] factors” he considers, id. at
177:8-19—as evidence that Mr. Looby considers race in the personal rating. But SFFA ignores
Mr. Looby’s explanation at trial that what he was trying to convey in his deposition was that he
considers “race as one factor of many within the application when assessing an applicant’s
overall rating.” 10/18 Tr. 177:20-178:12, 178:23-179:6, 184:1-185:23 (Looby). He was
unequivocal that he does not consider the fact of an applicant’s race when assigning an
applicant’s personal rating. Id. at 184:6-8 (“Q. Mr. Looby, do you consider race when assigning
an applicant a personal rating? A. No.”), 186:4-11 (“Q. Now, Mr. Looby, at the time that you
were deposed 14 months ago, did you, as an admissions officer, consider race when assigning an
applicant a personal rating? A. No.”). And to the extent Mr. Looby’s deposition testimony were
read to suggest that he considers race in assessing the personal rating, it would equally mean that
he considers race in the academic, extracurricular, and athletic ratings, see id. at 177:9-15, which
no one has claimed.
Finally, SFFA levels the grave accusation (¶ 133) that several witnesses “gave false and
misleading testimony” regarding the 2018 revisions to the reading procedures. Those allegations
were put to rest during trial, when SFFA recalled Director McGrath specifically to inquire about
the testimony it claims was “false.” Director McGrath explained that she understood the
question initially posed to her about whether there were written instructions regarding the
consideration of race in the personal rating to be focused on the discovery period in the litigation
and the materials that existed during that period. 11/1 Tr. 124:12-21, 125:5-11, 181:10-182:8.
She “had in mind [the] earlier reading instructions” (11/1 Tr. 125:5-11), not the most recent
reading procedures, which simply memorialize the “practice” that has existed “throughout [her]

- 53 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 61 of 78

tenure.” 11/1 Tr. 157:1-3, 175:20-177:15 (McGrath); Harvard ¶¶ 71-73. Director McGrath’s
recall testimony so clearly showed that there was no “false and misleading testimony” that SFFA
did not even bother to recall Dean Fitzsimmons (who, at SFFA’s insistence, was waiting in the
hall outside the courtroom) or Ms. Kim—the very witnesses it now baselessly alleges were
untruthful. See 10/25 Tr. 16:3-4 (Sealed); 11/1 Tr. 197:17-18 (Sealed). SFFA evidently would
prefer simply to accuse Dean Fitzsimmons and Ms. Kim of giving “false” testimony than to have
them explain—as Director McGrath did—why there was no inconsistency at all. It is serious
business to accuse a witness of giving false testimony under oath, and counsel have an obligation
not to levy such accusations without substantiation. SFFA has disregarded that obligation.
*

*

*

In sum, SFFA’s filing—far from showing why SFFA should prevail on its intentional
discrimination claim—simply highlights the insufficiency of SFFA’s evidence. SFFA tries to
rewrite both the law and the facts in its effort to fill yawning evidentiary gaps in its presentation
at trial. That is not how a plaintiff confident in its evidence behaves. The actual evidence
persuasively shows that Harvard does not discriminate against Asian-American applicants.
Harvard should therefore be granted judgment on that count.
II.

HARVARD DOES NOT ENGAGE IN RACIAL BALANCING
SFFA’s racial balancing claim is no better supported by the evidence at trial. To the

contrary, the evidence refutes SFFA’s claim (¶¶ 207-214) that Harvard targets a particular racial
composition of the admitted class: In the most recent decade analyzed by the experts, the AsianAmerican share of admitted students grew by 17%, from 17.6% to 20.6%; the African-American
share grew by more than 10%, from 10.5% to 11.6%; and the Hispanic share grew by more than
40%, from 8.2% to 11.6%. DX713.2. Moreover, the evidence showed without contradiction
that the proportion of students of all races in both the admitted and matriculating classes has

- 54 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 62 of 78

varied meaningfully from year to year (Harvard ¶¶ 80-81) and that the racial composition of the
admitted class fluctuates to a greater degree from year to year than the racial composition of the
applicant pool—which, as Dr. Card explained, is exactly the opposite of what one would expect
were Harvard deliberately pursuing a predetermined racial mix (Harvard ¶ 82). Those numbers
belie SFFA’s charge that “Harvard’s objective is to keep each racial group with[in] a certain
range year over year, and it does just that.” SFFA ¶ 214.
SFFA’s expert does not challenge this basic math (Harvard ¶ 84), nor could he. Indeed,
Dr. Arcidiacono declined to offer any testimony in support of SFFA’s racial balancing claim.
Harvard ¶ 84; 10/25 Tr. 202:9-203:1 (Arcidiacono). Instead, SFFA accuses Harvard of racial
balancing based on one of its recruitment strategies and its efforts to avoid admitting more
freshmen than it can house. Neither argument is availing.
First, SFFA criticizes Harvard for sending recruitment letters to high-achieving high
school students on the basis of standardized test score cutoffs that vary by racial group. SFFA
¶ 134. But Title VI does not limit the consideration of race in recruiting in the way that it limits
the consideration of race in admissions; to the contrary, courts regularly identify recruitment of
minority candidates as a “race-neutral” practice. Harvard ¶ 280 (citing authorities). Nor does
Harvard’s use of differential test-score cutoffs in any way imply an effort to achieve a particular
racial composition of the admitted or matriculating class. By sending recruiting letters to
African-American and Hispanic students with lower test scores than White and Asian-American
students, Harvard is simply recognizing the challenges of educational opportunity that sometimes
lead African-American and Hispanic candidates with outstanding academic potential not to score
as highly as they otherwise might. That is precisely the kind of recruitment effort that SFFA, its
expert, and federal courts have endorsed as a permissible means to increase diversity—indeed,

- 55 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 63 of 78

the sort of approach universities must exhaust before considering race in the admissions process
itself. See, e.g., Ensley Branch, NAACP v. Seibels, 31 F.3d 1548, 1571 (11th Cir. 1994);
Peightal v. Metro. Dade Cty., 26 F.3d 1545, 1557-1558 (11th Cir. 1994); Complaint (Dkt. 1)
¶ 340 (“Harvard could achieve its student body diversity without the use of racial preferences by
improving its recruitment of … high-achieving minorities[.]”); 10/22 Tr. 23:22-24:6
(Kahlenberg) (“I think it’s perfectly acceptable for a university to recruit students in a raceconscious fashion.”); Harvard ¶¶ 233, 280.
SFFA’s charge that Asian-American students from “sparse country” must have higher
standardized test scores than White students to be recruited by Harvard is also incorrect. SFFA
cherry-picks one example from a particular standardized-test search in a single year, but there
was no consistent or systematic difference in the test-score thresholds for White and AsianAmerican students to receive recruitment letters. Harvard ¶¶ 194-198. Indeed, during the same
years in which letters were mailed to White students in “sparse country” states with slightly
lower PSAT scores than Asian-American students in the same states, the ACT-score thresholds
for White and Asian-American students in those states—where the ACT is more popular than the
SAT—were identical. Harvard ¶ 197. And in other years, Harvard sent letters to AsianAmerican students with lower standardized test scores than White students. Harvard ¶ 195.
Second, SFFA mischaracterizes the evidence concerning statistical reports that leaders of
the Admissions Office consult from time to time as the process progresses. SFFA attempts to
twist the consideration of those reports into an effort by the Admissions Office to achieve a
targeted racial composition. But one key reason Admissions Office leaders consult the reports—
which include statistics on many characteristics of the class, not just race—is that they have
learned that admitted students with different characteristics (racial and otherwise) tend to accept

- 56 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 64 of 78

their offers of admission at different rates. Harvard ¶¶ 88-89. As a consequence, if more
students with particular characteristics (racial or otherwise) are admitted in a given year than in a
prior year, the class’s overall yield rate may differ from what the Admissions Office initially
estimated it would be, and the Office may therefore need to revise its projection of the overall
number of offers that will yield a number of matriculating students that does not exceed the
number of available beds. Harvard ¶¶ 88-89.
Not only do those facts fail to establish racial balancing; they weigh against the allegation
of racial balancing. If the Admissions Office were indeed targeting a class with a predetermined
racial composition, consistent from year to year, then it would know from the start how the
targeted racial composition of the admitted class would affect the overall yield rate of the class.
But Admissions Office leaders need to consult statistics as the process progresses precisely
because they do not know ahead of time what the final composition of the class (racial or
otherwise) will be. They have to keep track of how the class is shaping up because it can and
does vary from what past experience might have led anyone to predict.
SFFA’s account of how the Admissions Office leaders use the statistical reports is
misleading in other ways as well. Information Dean Fitzsimmons occasionally shares with the
full Admissions Committee is not limited to “the racial makeup of the tentatively admitted
class,” as SFFA claims (¶ 136); Dean Fitzsimmons provides a broader “overview of the class” on
a range of dimensions that include gender, race, ethnicity, geography, and financial aid status.
10/24 Tr. 83:2-16 (Banks); 10/19 Tr. 198:2-9 (McGrath); 10/24 Tr. 129:12-25 (Kim); Harvard
¶ 90. And although the Admissions Committee may carefully consider candidates from any
group that has notably declined as a proportion of the tentatively admitted class, to ensure that it
does not overlook excellent candidates “by inadvertence or lack of care,” 10/19 Tr. 201:21-202:3

- 57 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 65 of 78

(McGrath), that hardly means the committee is targeting a particular racial composition of the
class. The Supreme Court has explicitly held that paying “[s]ome attention to numbers … does
not transform a flexible admissions system into a rigid quota,” Grutter, 539 U.S. at 336 (internal
quotation marks omitted), and there is no evidence that Harvard “insulat[es]” any category of
applicants “from competition with all other applicants,” id. at 334 (internal quotation marks
omitted).
Thus, the admissions process and the results of that process provide compelling proof that
Harvard does not engage in racial balancing. There are no racial targets, and because the racial
composition of the tentatively admitted class evolves as the admissions cycle unfolds,
Admissions Office leadership must continually estimate the projected yield rate in order to
determine the total number of offers that can be extended.22 The demographic makeup of
Harvard College has changed dramatically during the past several decades, and it changes
meaningfully from year to year. The evidence thus refutes SFFA’s claim of racial balancing, and
Harvard should be granted judgment on that claim as well.
III.

HARVARD CONSIDERS RACE IN THE MANNER PERMITTED BY PRECEDENT
A.

Harvard Has Articulated, And Is Pursuing, Exactly The Interests The
Supreme Court Has Recognized As Compelling

The Supreme Court has recognized the pursuit of the educational benefits flowing from
all forms of diversity—including but not limited to racial diversity—as a compelling interest.
E.g., Grutter, 539 U.S. at 327-333; Harvard ¶¶ 296-298. That is the interest Harvard is pursuing,
and Harvard has articulated its diversity-related educational objectives in precisely the reasoned,

22

The suggestion that race plays a disproportionate role during the “lop process” (SFFA
¶¶ 138, 213) is also unsupported by the evidence; as multiple witnesses testified, the lop process
is “no different” from the regular full committee process, and the full committee “discusse[s] the
whole candidate.” 11/1 Tr. 253:25-254:9 (Walsh); see also 10/18 Tr. 19:5-20:7 (Fitzsimmons);
10/24 Tr. 198:12-23 (Ray); 11/1 Tr. 245:5-14 (Cheng), 251:5-21 (Weaver); DX56.
- 58 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 66 of 78

principled manner required by the Supreme Court. Harvard’s most recent articulation of those
objectives—a report authored by the Committee to Study the Importance of Student Body
Diversity and unanimously endorsed by the Faculty of Arts and Sciences (which includes the
College) (P302; Harvard ¶¶ 299-302)—echoes goals the Supreme Court expressly approved as
sufficiently “concrete and precise” in Fisher II: “the destruction of stereotypes, the promotion of
cross-racial understanding, the preparation of a student body for an increasingly diverse
workforce and society, and the cultivation of a set of leaders with legitimacy in the eyes of the
citizenry,” 136 S. Ct. at 2211 (brackets and internal quotation marks omitted); see also 10/23 Tr.
61:12-22 (Khurana).
SFFA faults Harvard for “not articulat[ing] ‘what level of diversity [it] would like to
achieve in order to get the educational benefits of diversity.’” SFFA ¶ 150 (emphasis added).
But no Supreme Court precedent requires Harvard to articulate its educational objectives in
terms of a numerical target for the racial composition of its student body.23 SFFA ignores the
Supreme Court’s explanation that “the compelling interest that justifies consideration of race in
college admissions is not an interest in enrolling a certain number of minority students,” but
rather an interest in “obtaining the educational benefits that flow from student body diversity,”
Fisher II, 136 S. Ct. at 2210 (internal quotation marks omitted)—exactly the interest Harvard is
pursuing.
SFFA argues (¶ 216) that Harvard’s commitment to diversity is not serious, or that
Harvard is not pursuing the type of diversity-related educational benefits the Supreme Court

23

SFFA also faults Harvard (¶ 151) for not pursuing “a ‘critical mass’ of underrepresented
minorities in its student body.” But as the Court recognized in denying summary judgment, the
pursuit of a “critical mass” is not the only interest the Supreme Court regards as compelling.
2018 WL 4688308, at *16; see also Dkt. 435 (Harvard SJ Opp.) at 35-36.
- 59 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 67 of 78

regards as compelling, because Harvard is making insufficient efforts to pursue types of diversity
other than racial diversity, including socioeconomic, geographical, and religious diversity. That
manifestly distorts the record—including evidence submitted by SFFA itself.
Harvard makes extensive efforts to pursue socioeconomic diversity. SFFA’s own expert,
Mr. Kahlenberg, noted that Harvard was highly ranked by the New York Times’s College Access
Index (10/22 Tr. 98:3-5) and that Harvard has “made a concerted effort to bring in low-income
students of all races” (id. at 89:14-90:2). Harvard offers one of the most generous need-based
financial aid programs in the country, with the aim of ensuring that ability to pay will not prevent
any admitted student from attending. Harvard ¶¶ 25-27. It makes extensive efforts to recruit
high-school students from lower-income families, Harvard ¶¶ 23, 26, and it considers, in the
admissions process itself, the extent to which students will contribute to socioeconomic diversity
on campus, Harvard ¶ 30. Harvard similarly pursues geographic diversity—providing a tip to
applicants from geographically underrepresented areas, DX5.11; 10/24 Tr. 208:4-10 (Ray), and
seeking to admit students from as many states as possible given applicants’ qualifications, 10/24
Tr. 176:6-177:20 (Kim); see also 11/1 Tr. 192:17-25 (Faust). And Harvard also values religious
diversity, as Dean Fitzsimmons and Director McGrath both testified. 10/17 Tr. 193:1-11
(Fitzsimmons); 10/19 Tr. 187:21-24 (McGrath). SFFA is incorrect to suggest that Harvard’s
consideration of race is somehow improper because the Admissions Office (after consulting with
counsel) does not receive a particular data field where applicants can indicate their religion
(10/19 Tr. 188:10-15 (McGrath)); in any event, the record makes clear that where an applicant
indicates his or her religious identity in other ways, the Admissions Office “may well consider”
that information and “often find[s] it helpful.” Id. at 193:18-22, 194:6-9.

- 60 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 68 of 78

Contrary to SFFA’s misleading statistics (¶ 143) and unsupported claims (¶ 144),
Harvard’s efforts to achieve diversity on many dimensions have paid off; the socioeconomic and
geographic diversity of the student body have increased over time. 10/24 Tr. 105:1-21 (Banks);
11/1 Tr. 193:11-194:13 (Faust); Harvard ¶ 78. Some 20% of Harvard undergraduates now fall
below the threshold for zero parental contribution (household income of $65,000), and roughly
15% are first-generation college students. Harvard ¶¶ 24, 26-27.
In sum, Harvard has done exactly what the Supreme Court requires of universities that
consider race in admissions: It has articulated its diversity-related educational objectives in a
reasoned way and pursued those objectives with respect to diversity of all forms.
B.

Harvard Permissibly Considers Race As A “Plus Factor”

SFFA argues that Harvard considers race as more than the “‘plus’ factor” allowed by
Supreme Court precedent, on the theory that Harvard’s consideration of race has a “substantial”
average marginal effect on the likelihood of admission for African-American and Hispanic
applicants. SFFA ¶¶ 141, 145-152, 217-220. But the evidence shows that Harvard’s admissions
process complies with the standard set by the Supreme Court: It is “flexible enough to ensure
that each applicant is evaluated as an individual and not in a way that makes an applicant’s race
or ethnicity the defining feature of his or her application,” Grutter, 539 U.S. at 337.
First, the evidence shows that admissions officers do not consider race as the defining
feature of an applicant’s candidacy; they consider many different factors in a wide-ranging effort
to learn as much as possible about the applicant as an individual. Harvard ¶¶ 28-34, 38-50.
Admissions officers receive extensive training about how to consider race as one of many
relevant factors, including through a weeks-long orientation (Harvard ¶ 66), training from
Harvard’s Office of the General Counsel (Harvard ¶ 67), feedback from a senior admissions
officer on a new admissions officer’s review of his or her first 50 to 100 applications (Harvard
- 61 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 69 of 78

¶ 68), annual staff retreats (Harvard ¶ 70), and written reading procedures, which are updated
annually (Harvard ¶¶ 71-73). And admissions officers testified credibly and consistently that
they consider race only as one factor among many that may be relevant in reviewing an
application. Harvard ¶ 93.
The statistical evidence confirms this testimonial and documentary evidence. Dr. Card’s
analysis showed that, knowing only an applicant’s race, one would know almost nothing about
the likelihood of the applicant’s being admitted—far less than one would be able to predict if one
knew only the applicant’s intended concentration, intended career, high-school and
neighborhood characteristics, or profile ratings. Harvard ¶¶ 94; DX715 & DD10.93; 10/31 Tr.
100:9-102:10 (Card).
Second, and relatedly, the evidence shows that race matters not at all for most applicants;
it is potentially relevant to the admissions decision only for those applicants who are otherwise
highly competitive for admission. Admissions officers testified credibly and consistently that
race can make a difference only for applicants who would be highly competitive regardless of
their race. Harvard ¶ 93. And Dr. Card’s analysis confirmed that race has a meaningful effect
only for candidates who are already highly qualified. Harvard ¶ 94. Dr. Arcidiacono agreed that
“a large number of applicants to Harvard will be rejected without race ever becoming a factor,”
and that race makes a difference only for a “competitive pool” of applicants that is “defined by a
variety of variables and factors” other than race. 10/25 Tr. 200:1-17. Thus, race alone is never
the reason an applicant is admitted or denied admission; to the extent race plays a role at all in
the consideration of an individual applicant’s candidacy, it is at most one of many reasons a
highly competitive applicant is admitted.

- 62 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 70 of 78

Third, the evidence shows that for those highly competitive applicants for whom race can
matter in the admissions process, the effect of race is not disproportionate to that of other factors.
Harvard ¶ 95. SFFA argues that race “can,” for the most competitive African-American
applicants, “be as valuable in securing admissions as receiving a coveted ‘1’ on the academic,
extracurricular, or personal rating.” SFFA ¶ 220 (citing SFFA ¶ 149). But even taken at face
value, that argument establishes at most that race can be one of the factors that may be important
for certain applicants, not that it is “the defining feature” of an application or “the predominant
factor” in Harvard’s admissions process, Grutter, 537 U.S. at 320, 337 (emphasis added).24 That
simply underscores that Harvard’s admissions process is “flexible enough to consider all
pertinent elements of diversity in light of the particular qualifications of each applicant,” as
Grutter requires, 539 U.S. at 334. And the magnitude of the estimated effect of race, for
applicants who would be strong no matter what their race, simply reflects the highly competitive
nature of Harvard’s admissions process. Harvard ¶ 95. As Dr. Card explained in the context of
his retirement hypothetical, “any factor has … a multiplied or a much more powerful effect” for
individuals who are “on the bubble” of a decision being modeled. 10/30 Tr. 118:25-119:23
(emphasis added). Or, in more basic terms: When exceptionally qualified applicants are
competing so intensely for so few spots, any additional factor may have a considerable effect on
whether an otherwise competitive applicant is admitted.
Thus, when race is considered as a “plus factor” for highly competitive applicants, some
applicants who might otherwise not have been admitted will be admitted. But that does not

24

Even an academic rating of 1 is not dispositive in the admissions process; Dr. Card
showed, for example, that applicants with an academic rating of 1 are more likely than not to be
denied admission unless they also have an extracurricular, athletic, or personal rating of 1 or 2.
DX672; DD10.7.
- 63 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 71 of 78

render the consideration of race unlawful. Indeed, in Fisher II, the Supreme Court approved an
admissions process under which race could “make a difference to whether an application is
accepted or rejected.” 136 S. Ct. at 2207; see also Grutter, 539 U.S. at 339 (even if race was
“‘likely outcome determinative’” in the admissions process “‘for many members of minority
groups’ who do not fall within the upper range of” grades and test scores, “the same could be
said of the Harvard plan discussed approvingly by Justice Powell in Bakke, and indeed of any
plan that uses race as one of many factors”).
Nor is it problematic that consideration of race affects enough admissions decisions, in
aggregate, to increase meaningfully the number of African-American and Hispanic students at
Harvard. SFFA ¶ 147. The Supreme Court has recognized that one of the interests supporting
the consideration of race in college admissions is to increase the admission of minority students
who otherwise “might not be represented in [the] student body in meaningful numbers.” Grutter,
539 U.S. at 316; see also Fisher II, 136 S. Ct. at 2212 (explaining that the 54% increase in
Hispanic candidates and 94% increase in African-American candidates “enrolled through holistic
review … show[ed] that consideration of race has had a meaningful, if still limited, effect on the
diversity of the University’s freshman class”).
SFFA’s proposed conclusions of law focus, tellingly, on the straw-man argument that a
university may not pursue racial diversity through rigid point-based systems. SFFA ¶ 218
(describing process in Gratz, where “[b]eing an underrepresented racial minority was worth 20
points”); SFFA ¶ 219 (describing process in Johnson v. Board of Regents of University of
Georgia, 263 F.3d 1234 (11th Cir. 2001), where “race was worth 0.5 points” for non-White
applicants). No one disagrees with that proposition. But Harvard’s admissions process bears not
the slightest resemblance to those in SFFA’s cases, where applicants were automatically

- 64 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 72 of 78

assigned a fixed number of points solely for their race, without regard to their individual
characteristics. SFFA fails to grapple with the reality that Harvard’s system, like the one
approved in Grutter and “[u]nlike the program at issue in Gratz … , awards no mechanical,
predetermined diversity ‘bonuses’ based on race or ethnicity,” Grutter, 539 U.S. at 337. In
effect, SFFA reasons that if statistical analysis reveals that the consideration of race has a
quantifiable effect on the likelihood of admission for certain applicants—even if race is
considered as one among many factors in an individualized process—then that estimated effect is
equivalent to a rigid allocation of points. But that argument would collapse the distinction
articulated in Grutter and Gratz and would effectively prohibit any consideration of race. That is
not the law, though SFFA would no doubt like it to be.
C.

Harvard Has Committed To Periodically Reviewing The Need To Consider
Race

Finally, SFFA accuses Harvard of defying the Supreme Court’s requirement that the
consideration of race in an admissions process be “limited in time,” Grutter, 539 U.S. at 342.
SFFA ¶¶ 224-225. SFFA apparently thinks that a race-conscious policy is unlawful unless it
contains a predetermined “‘sunset provision[]’” to enforce the limited-in-time requirement. Id.
But SFFA points to no precedent holding as much, because none exists.25 Indeed, the admissions
policy upheld in Grutter itself did not contain a definite sunset date. Id. at 343 (“We take the

25

Neither of the cases on which SFFA relies—Aiken v. City of Memphis, 37 F.3d 1155 (6th
Cir. 1994), and Ravitch v. City of New York, 1992 WL 196735 (S.D.N.Y. Aug. 3, 1992)—is to
the contrary. In Aiken, the Court’s concern was that the defendant had “made no effort to limit
the duration of” its race-conscious practices even after having promised, fifteen years earlier, that
it expected race-neutral practices to be available in two to three years. 37 F.3d at 1164. In
Ravitch, the problem was that the race-conscious provision was encoded in the City Charter and
so, by its terms, would remain in force indefinitely. 1992 WL 196735, at *7. Neither case
suggests that a sunset provision is necessary where, as here, the defendant credibly asserts its
intention to review the continuing need for race-conscious measures.
- 65 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 73 of 78

Law School at its word that it would ‘like nothing better than to find a race-neutral admissions
formula’ and will terminate its race-conscious admission program as soon as practicable.”).
The committee that considered race-neutral alternatives for Harvard College admissions
expressly concluded that “it will be important to reassess, periodically, the necessity of
considering race and ethnicity in the admissions process.” P316 at 18-19. To that end, the
committee recommended that Harvard “re-evaluate its consideration of race-neutral alternatives
five years from now” (which is now just over four years away). P316 at 19. There is no basis to
doubt that Harvard will follow through on that recommendation, which was echoed at trial by
Dean Smith, Dean Khurana, and President Faust. 10/23 Tr. 60:11-21 (Khurana); 10/23 Tr.
134:15-19 (Smith); 11/1 Tr. 222:25-223:5 (Faust).
The evidence thus does not support SFFA’s claim that Harvard considers race in a
manner not permitted by Supreme Court precedent, and the Court should enter judgment for
Harvard on that claim.
IV.

HARVARD HAS PROPERLY CONSIDERED WHETHER IT COULD ACHIEVE ITS
EDUCATIONAL INTERESTS WITHOUT CONSIDERING RACE AND HAS SHOWN IT COULD
NOT
In an effort to support its claim that Harvard could achieve its educational objectives

without considering race, SFFA attacks (¶¶ 153-158, 226-233) both the process Harvard
followed in assessing race-neutral alternatives and the substance of the conclusion it reached.
But the evidence showed that Harvard’s consideration of race-neutral alternatives was a “serious,
good faith” review, Fisher v. Univ. of Texas at Austin (Fisher I), 570 U.S. 297, 312 (2013), and
that “no workable race-neutral alternatives would produce the educational benefits of diversity,”
id.
First, the evidence showed that Harvard’s process of reviewing race-neutral alternatives
was rigorous. Over several months, a committee of three deans with decades of experience in
- 66 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 74 of 78

higher education administration and undergraduate admissions carefully evaluated every single
race-neutral alternative that SFFA suggested, and more. Harvard ¶¶ 206-253; 10/22 Tr. 142:22143:2 (Kahlenberg). With the benefit of expert simulations, academic literature, and their
professional judgment, the members of that committee concluded that no combination of raceneutral practices would allow Harvard to achieve its educational objectives at this time. Harvard
¶¶ 206-253.
SFFA argues (¶ 227) that the committee’s consideration of race-neutral alternatives is
inadequate because, it says, in the past Harvard did not formally study race-neutral alternatives
before it began taking race into account as one factor in its admissions process. But Harvard has
long complied with the obligation to undertake “serious, good faith consideration of” raceneutral means of pursuing diversity, Grutter, 539 U.S. at 339; it has for decades made significant
efforts, beyond the consideration of race, to achieve diversity in its applicant pool and admitted
class. Harvard ¶¶ 19-27, 75-78. The Supreme Court has made clear that “a university’s
experience and expertise in adopting or rejecting certain admissions processes” is relevant to the
determination whether the university could achieve its educational objectives without
considering race. Fisher I, 570 U.S. at 311. And to the extent the Supreme Court’s precedents
require a formal analysis of what would happen if the university stopped considering race in its
admissions process, the only relevant question—in this case seeking exclusively forward-looking
relief—is whether Harvard has now undertaken the requisite analysis, not whether it had done so
by some indeterminate time in the past. Harvard ¶ 308. It has done so.
SFFA denigrates the committee’s analysis as “not serious” and “comical[].” SFFA ¶ 230.
That name-calling is unjustified. The committee undertook its work conscientiously, meeting
seven times over an eight-month period and reviewing and discussing both the academic

- 67 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 75 of 78

literature on race-neutral alternatives and the expert reports produced in this case. Harvard
¶ 208.
SFFA also never explains why the fact that the committee included “only three” members
(¶¶ 155, 230) is problematic. The three members of the committee were chosen for their specific
expertise. Harvard ¶ 207; 11/1 Tr. 220:18-221:5 (Faust). And although they were fewer in
number than the members of the University-wide committee originally formed to examine
diversity and race-neutral alternatives (Harvard ¶ 204), there were evident differences between
the two committees’ charges that explain why the Committee to Study Race-Neutral Alternatives
in Harvard College Admissions needed fewer members. The committee at issue here examined
only Harvard College’s admissions practices, whereas the prior committee was charged with
studying the thirteen schools making up the University as a whole; the College committee
examined only race-neutral alternatives, whereas the University-wide committee was also
charged with articulating the University’s diversity-related educational interests; and the College
committee, unlike the University-wide committee, had the benefit of the extensive statistical
analysis of race-neutral alternatives produced in the context of this litigation, which limited the
additional work the committee needed to conduct. Harvard ¶¶ 204-206, 214.
Second, SFFA’s challenge to the committee’s substantive conclusions is equally
meritless. SFFA blithely characterizes as “viable” (¶ 157) four sets of alternative practices that
Mr. Kahlenberg presented. But although SFFA claims all four sets of practices would “maintain
or increase racial diversity,” each would in fact result in a nearly 30% drop in the number of
African-American students admitted to Harvard. Harvard ¶ 249. SFFA apparently believes that
an increase in the proportion of students of other ethnicities would compensate for that 30%
decline, leaving total “diversity” unchanged or increased. But it has no answer to the many

- 68 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 76 of 78

witnesses, including current and former Harvard students as well as members of the committee
on race-neutral alternatives, who explained why a decline of that magnitude in the representation
of African-American students would inhibit Harvard’s educational objectives. 10/23 Tr. 155:15156:7 (Smith); 10/29 Tr. 21:5-22:3 (Vasquez-Rodriguez); 10/29 Tr. 78:12-79:13 (Cole); 10/29
Tr. 154:18-155:21 (Diep); 10/29 Tr. 177:6-22 (Trice).
Rather than actually addressing that testimony, SFFA belittles the very notion that an
approach might be “unworkable because the admission of African Americans might decrease
slightly”—as if a 30% drop could be characterized as “slight[].” SFFA ¶ 232. SFFA equates the
testimony about the need to avoid such a precipitous decline in the representation of AfricanAmerican students with the pursuit of “a quota.” Id. But there is no such equivalence—“‘[s]ome
attention to numbers,’ without more, does not transform a flexible admissions system into a rigid
quota,” Grutter, 539 U.S. at 336—and the record certainly does not support SFFA’s accusation
that Harvard is attempting to “preserv[e] a precise number of spots for any racial group,” SFFA
¶ 232. Dean Smith’s committee did not set a target for the racial composition of the class,
Harvard ¶ 211, but reasonably concluded that Harvard’s diversity-related goals would be
undermined by a profound decline in African-American representation. Harvard ¶ 249.
SFFA is also wrong to insist that Harvard could adopt race-neutral alternatives without
compromising the academic excellence of the admitted class. In making that argument, SFFA
focuses exclusively on high-school grade point averages and SAT scores, SFFA ¶ 158, but
Harvard regards those metrics as having only limited value in distinguishing truly stellar students
from others, Harvard ¶ 212. The alternatives SFFA embraces would significantly reduce the
proportion of admitted students receiving an academic rating of 1 or 2, Harvard ¶ 250—the
metric Harvard regards as far more meaningful, Harvard ¶ 212. And the law does not “require a

- 69 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 77 of 78

university to choose between maintaining a reputation for excellence [and] fulfilling a
commitment to provide educational opportunities to members of all racial groups.” Fisher II,
136 S. Ct. at 2208 (alteration in original; internal quotation marks omitted). To the contrary,
both diversity and academic excellence are so fundamental to Harvard’s educational mission that
forcing Harvard to sacrifice one to achieve the other would strike at the heart of the academic
freedom protected by the First Amendment, see Harvard 66 n.2.
In the final analysis, Harvard studied the race-neutral alternatives SFFA suggested and
others SFFA chose not to suggest, considered the factors SFFA’s expert deemed appropriate, and
reached a reasonable conclusion supported by testimony and empirical analysis. SFFA’s quarrel
with the composition of Harvard’s committee and its insistence that race-neutral alternatives
would work at Harvard (notwithstanding Harvard’s actual experience deploying many of SFFA’s
preferred alternatives) are unpersuasive. Harvard engaged in a serious, good-faith examination
and demonstrated that workable race-neutral alternatives would not allow it to achieve its
educational objectives. It is therefore entitled to judgment on this count.
CONCLUSION
The Court should enter judgment in Harvard’s favor on all remaining claims.

- 70 -

Case 1:14-cv-14176-ADB Document 628 Filed 01/23/19 Page 78 of 78

Respectfully submitted,

/s/ Seth P. Waxman
Seth P. Waxman (pro hac vice)
Paul R.Q. Wolfson (pro hac vice)
Danielle Conley (pro hac vice)
Brittany Amadi (pro hac vice)
Daniel Winik (pro hac vice)
WILMER CUTLER PICKERING
HALE AND DORR LLP
1875 Pennsylvania Ave. NW
Washington, D.C. 20006
Tel: [phone]
Fax: [phone]
[e-mail]
[e-mail]
[e-mail]
[e-mail]
[e-mail]

William F. Lee (BBO #291960)
Felicia H. Ellsworth (BBO #665232)
Andrew S. Dulberg (BBO #675405)
Elizabeth Mooney (BBO #679522)
Sarah R. Frazier (BBO #681656)
WILMER CUTLER PICKERING
HALE AND DORR LLP
60 State Street
Boston, MA 02109
Tel: [phone]
Fax: [phone]
[e-mail]
[e-mail]
[e-mail]
[e-mail]
[e-mail]

Debo P. Adegbile (pro hac vice)
WILMER CUTLER PICKERING
HALE AND DORR LLP
7 World Trade Center
250 Greenwich Street
New York, NY 10007
Tel: [phone]
Fax: [phone]
[e-mail]
Counsel for Defendant President and
Fellows of Harvard College

Dated: January 23, 2019

CERTIFICATE OF SERVICE
I hereby certify that this document filed through the CM/ECF system will be sent
electronically to the registered participants as identified on the Notice of Electronic Filing.
/s/ Seth P. Waxman
Seth P. Waxman

- 71 -