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

Filing 435

MEMORANDUM in Opposition re 412 MOTION for Summary Judgment filed by President and Fellows of Harvard College. (Waxman, Seth) (Additional attachment(s) added on 7/30/2018: # 1 SEALED Oppostion to Motion for Summary Judgment) (McDonagh, Christina).

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 1 of 51

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
Motion to File Under Seal Granted
July 24, 2018 [Dkt. 431]

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

DEFENDANT’S MEMORANDUM IN OPPOSITION TO PLAINTIFF’S
MOTION FOR SUMMARY JUDGMENT

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 2 of 51

TABLE OF CONTENTS
INTRODUCTION ...........................................................................................................................1
ARGUMENT ...................................................................................................................................4
I.

HARVARD DOES NOT DISCRIMINATE AGAINST ASIAN-AMERICAN APPLICANTS ..................4
A.

The Statistical Evidence Does Not Entitle SFFA To Summary Judgment ..............5
1.

The Deficiencies Of Dr. Arcidiacono’s Regression Analysis Preclude
Summary Judgment In SFFA’s Favor .........................................................6

2.

Dr. Arcidiacono’s Crude Comparisons Equally Fail To Entitle SFFA To
Summary Judgment ...................................................................................15

B.

The OIR Documents Are Not Evidence—Let Alone Undisputed Evidence—Of
Discrimination........................................................................................................19

C.

SFFA’s Supposed “Corroborating Evidence” Is Cherry-Picked And Far From
Sufficient To Support Summary Judgment............................................................23

D.

As The Court Has Already Recognized, Harvard’s Treatment Of Jewish
Applicants In The 1920s Is Irrelevant To This Case .............................................28

II.

HARVARD DOES NOT ENGAGE IN RACIAL BALANCING ......................................................29

III.

HARVARD CONSIDERS RACE FLEXIBLY, ALONG WITH MANY OTHER FACTORS, TO
ACHIEVE THE EDUCATIONAL BENEFITS OF DIVERSITY ......................................................35

IV.

HARVARD COULD NOT ACHIEVE ITS EDUCATIONAL OBJECTIVES WITHOUT CONSIDERING
RACE IN THE ADMISSIONS PROCESS ...................................................................................39

CONCLUSION ..............................................................................................................................45
CERTIFICATE OF SERVICE ......................................................................................................47

-i-

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 3 of 51

TABLE OF AUTHORITIES
Cases
Abbott v. Perez,
138 S. Ct. 2305 (2018) .............................................................................................................28
Blunt v. Lower Merion School District,
767 F.3d 247 (3d Cir. 2014).......................................................................................................5
Estrada v. Progressive Direct Insurance Co.,
53 F. Supp. 3d 484 (D. Mass. 2014) ........................................................................................35
Fisher v. University of Texas at Austin (Fisher I),
570 U.S. 297 (2013) ...........................................................................................................40, 41
Fisher v. University of Texas at Austin (Fisher II),
136 S. Ct. 2198 (2016) .................................................................................2, 30, 31, 35, 36, 44
Gonzalez v. El Dia, Inc.,
304 F.3d 63 (1st Cir. 2002) ......................................................................................................27
Gratz v. Bollinger,
539 U.S. 244 (2003) .................................................................................................................38
Grutter v. Bollinger,
539 U.S. 306 (2003) ...................................................................................30, 34, 35, 36, 39, 44
Johnson v. Dismore,
2012 WL 1855031 (S.D. Ill. Apr. 24, 2012) ............................................................................22
McGuire v. Reilly,
386 F.3d 45 (1st Cir. 2004) ........................................................................................................5
Morse v. Hanks,
172 F.3d 983 (7th Cir. 1999) ...................................................................................................22
Perry v. Roy,
782 F.3d 73 (1st Cir. 2015) ........................................................................................................4
Regents of the University of California v. Bakke,
438 U.S. 265 (1978) .....................................................................................................28, 36, 38
Spath v. NCAA,
728 F.2d 25 (1st Cir. 1984) ........................................................................................................5
Steir v. Girl Scouts of the USA,
383 F.3d 7 (1st Cir. 2004) .......................................................................................................41

- ii -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 4 of 51

Straughn v. Delta Air Lines, Inc.,
250 F.3d 23 (1st Cir. 2001) ......................................................................................................27
United States v. Oregon State Medical Society,
343 U.S. 326 (1952) .................................................................................................................41
Village of Arlington Heights v. Metropolitan Housing Development Corp.,
429 U.S. 252 (1977) .................................................................................................................28
Rules
Fed. R. Civ. P. 56(c) ......................................................................................................................28
Fed. R. Evid. 402 ...........................................................................................................................28
Other Authorities
Daniel L. Rubinfeld, Reference Guide on Multiple Regression, in Reference
Manual on Scientific Evidence 305 (3d ed. 2011) .....................................................................8

- iii -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 5 of 51

INTRODUCTION
SFFA’s summary judgment filing is not a serious effort to convince the Court that this
case can be resolved in SFFA’s favor without a trial. It is a 45-page press release, devoted to
presenting a misleading narrative that is manifestly subject to genuine dispute. And SFFA barely
even tries to argue that its 900 supposedly undisputed material facts are actually undisputed;
SFFA well knows that Harvard disputes those assertions, their materiality, or both.
Given the record that has been compiled in this case, SFFA’s contention that it is entitled
to summary judgment is not remotely plausible. Harvard is fully prepared to make its case at
trial, but as Harvard’s summary judgment papers explain, no reasonable factfinder could rule for
SFFA on this record. It is even clearer that SFFA cannot prevail without a trial. Each of SFFA’s
remaining counts rests on allegations that have been conclusively refuted or, at a minimum,
directly contested by documentary, testimonial, and expert evidence.
First, the evidence fails to show—let alone beyond dispute—that Harvard intentionally
discriminates against Asian-American applicants. SFFA cannot point to any testimonial or
documentary evidence to support its claim, and so it is forced to rely on statistical evidence. But
SFFA’s statistics cannot support a judgment in its favor. They are disputed in every relevant
respect by the far more comprehensive and statistically sound analysis submitted by Harvard’s
expert, David Card, and even taken at face value they are insufficient to support SFFA’s claim.
Unable to disagree in any serious way with Dr. Card’s analysis, SFFA distorts his work and
attacks him personally. SFFA also mischaracterizes other evidence on which it relies—
preliminary, incomplete work by employees of Harvard’s Office of Institutional Research,
reactions by Harvard administrators who realized how incomplete that work was, and a handful
of cherry-picked and misleadingly presented documents that do not even suggest, much less

-1-

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 6 of 51

establish, that the Harvard admissions office engaged in what SFFA’s complaint and papers
claim was a years-long, systematic effort to limit Asian-American enrollment. Indeed, SFFA
does not offer even a theory of how a committee comprising some 40 people at any given time
could have carried out the supposed scheme in a concerted fashion over many years—without
generating a shred of documentary or testimonial evidence of the alleged scheme.
Second, the evidence fails to show—let alone beyond dispute—that Harvard engages in
racial balancing. It is surprising that SFFA even presses this argument, for admissions data show
that the racial composition of the class admitted to Harvard fluctuates significantly from year to
year. SFFA’s own expert conceded that Harvard’s expert had shown “changes in the fraction of
admitted students by race/ethnicity over time” and never suggested that those changes were
statistically insignificant. Declaration of Felicia H. Ellsworth in Support of Defendant’s Motion
for Summary Judgment (Dkt. 419), Ex. 35 at 58. SFFA has identified no evidence that Harvard
seeks any particular racial balance for its classes, and there is none.
Third, the evidence fails to show—let alone beyond dispute—that Harvard considers race
in anything but the manner permitted by law. SFFA’s brief offers a new theory of liability, not
articulated in its complaint, that faults Harvard for failing to seek a “critical mass” of
underrepresented minority students. But the Supreme Court has never held that universities may
consider race only to attain a “critical mass” or that “critical mass” is a magic phrase that
universities must invoke. Indeed, Fisher II squarely rejects that proposition. Rather, the
Supreme Court has held that universities may consider race in the admissions process to
“obtain[] the educational benefits that flow from student body diversity.” Fisher v. Univ. of Tex.
at Austin (Fisher II), 136 S. Ct. 2198, 2210 (2016) (internal quotation marks omitted). The
evidence leaves no doubt that Harvard permissibly seeks those educational benefits in the

-2-

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 7 of 51

flexible, non-mechanical manner permitted by the Supreme Court. SFFA does not even mention
the unrebutted testimony of Harvard admissions officers explaining the limited way in which
race is considered, as one factor among many in the process. And the statistical evidence
confirms that testimony, by showing that numerous factors explain far more about admissions
decisions than does race.
Finally, the evidence fails to show—let alone beyond dispute—that Harvard could
achieve its educational objectives without considering race. SFFA inveighs against the process
by which Harvard considered race-neutral alternative practices but has notably little to say about
the results of Harvard’s analysis. SFFA’s objections to Harvard’s process are misplaced, and its
silence on the results speaks volumes. The process was an outgrowth of work that began before
this litigation commenced, and it was explicitly designed to allow consideration of all available
evidence, including the work done by both parties’ experts in this litigation. The result was a
report by the three most senior Harvard officials with direct responsibility for undergraduate
education and admissions. Harvard’s conclusions—that race-neutral alternatives would not
allow it to achieve the educational benefits of diversity that are required to meet its pedagogical
goals—are straightforward and unassailable.
For the reasons explained in Harvard’s summary judgment brief, the only conclusion
supported by the evidence is that SFFA’s claims fail and Harvard is entitled to judgment as a
matter of law. But if the Court disagrees, then Harvard welcomes the opportunity to expose at
trial the many fallacies of the narrative SFFA has offered. The truth is that Harvard devotes
extraordinary resources to extensive recruiting, a world-leading financial aid program, and a
painstaking review of individual applications, in order to admit and enroll a student body that is
diverse across all dimensions. It does so because it firmly believes that the perspective students

-3-

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 8 of 51

gain from living and learning alongside people who are not like themselves is essential to
fulfilling their potential as engaged citizens and leaders in our society. It does so in a manner
sensitive to and entirely consistent with Supreme Court precedent. And it does so without
discriminating against applicants of any race or ethnicity.
STANDARD OF REVIEW
In resolving a motion for summary judgment, the Court must view “the facts in the light
most favorable to … the party opposing summary judgment,” and must also draw “all reasonable
inferences” from those facts “in favor of the nonmovant.” Perry v. Roy, 782 F.3d 73, 75, 77 (1st
Cir. 2015). “Summary judgment is only appropriately granted where there is no genuine issue of
material fact, and the moving party is entitled to judgment as a matter of law.” Id. at 77-78
(internal quotation marks omitted). As long as “there exists evidence [] such that a reasonable”
factfinder could rule for the non-moving party, summary judgment should be denied. Id. at 78.
ARGUMENT
I.

HARVARD DOES NOT DISCRIMINATE AGAINST ASIAN-AMERICAN APPLICANTS
SFFA bases its claim of discrimination almost entirely on the statistical analysis offered

by its expert, Peter Arcidiacono. It undoubtedly does so because, notwithstanding the extensive
discovery SFFA took of Harvard in this litigation, SFFA has identified no documentary or
testimonial evidence that Harvard attempts to limit the number of Asian-American students who
enroll at the College. For the reasons explained in Harvard’s summary judgment brief (at 3844), the statistics in this case afford no legally sufficient evidence of intentional discrimination;
much less do they allow SFFA to carry its burden at summary judgment.
Aside from Dr. Arcidiacono’s deeply flawed statistical analysis, SFFA has nothing. Most
strikingly, SFFA fails to account for the absence from the record of any documentary or
testimonial evidence to support its theory that the roughly 40 current members of Harvard’s
-4-

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 9 of 51

Admissions Committee—and the many others who served on the committee in the preceding
years—engaged in an ongoing sub rosa scheme of intentional discrimination. It is inconceivable
that such a pervasive conspiracy would give rise to no evidence of its existence—not a single
admissions officer (current or former) who would testify to the nefarious scheme, nor a single
document alluding to it. SFFA’s resort to misleadingly presented language from a handful of
documents, generated during a period in which Harvard considered tens of thousands of
applicants, reveals the weakness of its case.
A.

The Statistical Evidence Does Not Entitle SFFA To Summary Judgment

As SFFA acknowledges, statistical evidence of disparate outcomes among applicants of
different races cannot prove intentional discrimination unless “the ‘statistical evidence of racially
disproportionate impact’” is “‘so strong that the results permit no other inference but that they
are the product of a racially discriminatory intent or purpose.’” SFFA Mem. 6 (quoting Spencer
v. Zant, 715 F.2d 1562, 1581 (11th Cir. 1983)); see also, e.g., 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.”); Mem. in Support of Def.’s Mot. for S.J.
(“Harvard Mem.”) 36. SFFA’s statistical evidence comes nowhere close to satisfying that high
standard. SFFA has shown no statistically significant disparity in the treatment of AsianAmerican and White applicants. And even if it had, it certainly has not proven—let alone with
unrebutted evidence—a “gross disparit[y] of the kind and degree sufficient to give rise to an
inference that” Harvard engaged in “a pattern or practice of discrimination,” Blunt v. Lower
Merion School District, 767 F.3d 247, 299-300 (3d Cir. 2014).

-5-

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 10 of 51

1.

The Deficiencies Of Dr. Arcidiacono’s Regression Analysis Preclude
Summary Judgment In SFFA’s Favor

Dr. Card and Dr. Arcidiacono agree that multivariate regression analysis is the proper
approach to estimating the effect of race in Harvard’s admissions process because it can control
for the effects of various other factors. Ellsworth Ex. 31 at 18; Ellsworth Ex. 33 ¶ 95. Yet
SFFA’s brief devotes exactly one paragraph (at 10) to Dr. Arcidiacono’s regression analysis.
That is for good reason: As Harvard’s summary judgment brief explains (at 38-44), Dr.
Arcidiacono’s analysis is fundamentally unreliable. Dr. Arcidiacono was able to arrive at his
conclusions only by excluding important information and treating the remaining data in a manner
seemingly calculated to reach the result SFFA desired.
Dr. Arcidiacono’s model fundamentally fails to account for key features of how
Harvard’s admissions process operates and is otherwise statistically unsound. To cite a basic
example, Dr. Arcidiacono chose to pool admissions data across all six admissions cycles for
which data were produced, rather than analyzing each cycle independently and computing
average effects across all cycles. Harvard Mem. 44. As Harvard’s summary judgment brief
explains (id.), that choice is methodologically indefensible because applicants in a given year
compete only against other applicants in the same year, not against applicants in different years.
Dr. Card’s analysis, employing the more appropriate year-by-year review, showed no statistically
significant evidence that Asian Americans are disadvantaged in Harvard’s admissions process,
either in any particular year or on average across the six years for which data were produced.
See Ellsworth Ex. 33 ¶ 92; Ellsworth Ex. 37 ¶ 103.
Dr. Arcidiacono’s model has other fundamental deficiencies as well. The two that most
severely skew his results are as follows:

-6-

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 11 of 51

a.

First, Dr. Arcidiacono performed his core analyses on a sample that excludes so-

called “ALDC” applicants—recruited athletes, children of Harvard College or Radcliffe alumni,
applicants on the Dean’s or Director’s interest lists, and children of Harvard faculty and staff
members. Harvard Mem. 41-42. It is clear why he did so: Excluding applicants in those
categories biases his model toward a result favorable to SFFA. Dr. Arcidiacono has disavowed
any claim that Harvard discriminates against applicants in the excluded categories, and his report
explicitly conceded that Harvard does not discriminate against such applicants. See Declaration
of Felicia H. Ellsworth in Opposition to SFFA’s Motion for Summary Judgment 1 Ex. 122 at
83:5-17, 88:22-90:20; Ellsworth Ex. 35 at 69 (“Harvard does not discriminate against AsianAmerican applicants who are in the special recruiting categories.”). In effect, therefore, Dr.
Arcidiacono chose to carve out the subset of the data that would have interfered with his ability
to achieve his desired result, rather than allowing the data to speak for itself in a statistically
sound manner. As discussed below, however, Dr. Arcidiacono’s choice makes it impossible for
SFFA even to articulate a coherent theory of how Harvard supposedly discriminated—let alone
to prove intentional discrimination on the basis of statistical evidence.
SFFA attacks (at 26-27) Dr. Card’s choice to include ALDC applicants in his analysis, on
the theory that “[t]he task here is to determine whether ‘similarly situated’ applicants have been
treated differently on the basis of race.” “‘[A]pples should be compared to apples,’” SFFA says
(at 27), and ALDC applicants “are not apples.” But that argument contradicts basic principles of
statistical analysis. The purpose of a regression analysis is to identify the roles of multiple

1

For ease of reference, the Declaration of Felicia H. Ellsworth in Support of Harvard’s
Motion for Summary Judgment ends with Exhibit 97 and the Declaration of Felicia H. Ellsworth
in Opposition to SFFA’s Motion for Summary Judgment begins with Exhibit 98, so exhibits
attached to both Declarations are referred to as “Ellsworth Exhibits.”
-7-

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 12 of 51

factors in complex processes. See, e.g., Daniel L. Rubinfeld, Reference Guide on Multiple
Regression, in Federal Judicial Center, Reference Manual on Scientific Evidence (“Reference
Guide”) 305, 305 (3d ed. 2011) (“Multiple regression analysis is a statistical tool used to
understand the relationship between or among two or more variables.”); Dkt. 414, Plaintiff’s
Statement of Undisputed Material Facts (“SFFA SJ SMF”) ¶¶ 650, 652 (logistic regression is a
means “to measure the effect of certain factors on a dependent variable,” and allows “the
marginal impact of each characteristic [to] be estimated in a way that controls for the impact of
the other characteristics”). In such an analysis, the way to deal with attributes (like ALDC
status) that distinguish some applicants from others is not to exclude any applicants who possess
the attribute; it is to control for the attribute, allowing the regression to distinguish the effect of
that attribute from the effects of other attributes. See, e.g., Rubinfeld, Reference Guide 336
(“The important variables that systematically might influence the dependent variable, and for
which data can be obtained, typically should be included explicitly in a statistical model.”).
SFFA’s argument makes as little sense as saying that, because employees with more advanced
degrees are paid more on average in a given field than those with less advanced degrees,
educational attainment should be excluded from a regression that attempts to discern how
numerous factors affect salaries. 2

2

SFFA’s argument in its brief for excluding ALDC applicants is not the one that Dr.
Arcidiacono invoked in his expert report, and for good reason: That rationale was fatally flawed,
as was revealed during Dr. Arcidiacono’s deposition. Dr. Arcidiacono’s theory was that ALDC
applicants are “subject to special admissions procedures,” Ellsworth Ex. 35 at 69, such that they
should not be compared in a regression model with applicants who are subject to the ordinary
admissions process. But at his deposition, Dr. Arcidiacono was unable to supply any basis for
his view that there is a separate admissions process for applicants with particular characteristics,
see Ellsworth Ex. 27 at 90:12-93:13—unsurprisingly, because that is not true, see Dkt. 420,
Local Rule 56.1 Statement of Undisputed Material Facts in Support of Defendant’s Motion for
Summary Judgment on All Remaining Counts (“Harvard SJ SMF”) ¶ 228; Ellsworth Ex. 149
(Declaration of Robin Worth (“Worth Decl.”)) ¶ 6. Rather, all applicants compete against each
-8-

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 13 of 51

Dr. Card does control for ALDC attributes, see Ellsworth Ex. 33 at 181 (Appendix E);
Ellsworth Ex. 37 at 116 (Appendix C), and finds no statistically significant effect of AsianAmerican ethnicity on applicants’ likelihood of admission, see Dkt. 420, Local Rule 56.1
Statement of Undisputed Material Facts in Support of Defendant’s Motion for Summary
Judgment on All Remaining Counts (“Harvard SJ SMF”) ¶¶ 219-220. Indeed, at Dr.
Arcidiacono’s suggestion, Dr. Card even includes in his model variables that “interact[]” ALDC
attributes with race, allowing the model to account for the possibility that the effect of race might
be different for ALDC applicants than for others. Ellsworth Ex. 37 ¶ 99. He “continue[s] to find
no evidence of bias” under that approach. Id.
The consequence of Dr. Arcidiacono’s decision to exclude ALDC applicants is that—
even if his model included all the relevant variables, which, as further discussed below, it does
not—the model would not address whether there is discrimination against Asian-American
applicants across the entire applicant pool. SFFA does not try to hide from this limitation of Dr.
Arcidiacono’s analysis; rather, SFFA rewrites its claim of discrimination to focus only on the
limited subset of the applicant pool included in Dr. Arcidiacono’s model. SFFA’s claim, it now
says (at 27), “‘is that the effects of Harvard’s use of race occur outside’” the so-called “‘special
categories’” of ALDC applicants—a proposition for which it cites Dr. Arcidiacono’s report
rather than the complaint. SFFA SJ SMF ¶ 751 (citing Ellsworth Ex. 35 at 19).
SFFA’s theory of discrimination is implausible and unsupported, however. SFFA does
not and cannot explain why, if Harvard truly intended to discriminate against Asian-American
applicants on the basis of their race, the supposed discrimination would affect only a subset of

other in the same process, and their characteristics may be considered as factors in that process.
That is presumably why SFFA abandons Dr. Arcidiacono’s “special admissions procedures”
explanation in its brief.

-9-

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 14 of 51

Asian-American applicants. 3 As courts have emphasized (see supra p. 5), it is exceptionally
difficult to prove intentional discrimination based only on statistics that reveal a disparity across
the entire allegedly targeted population; SFFA does not even attempt to do that, but rather tries to
make its case based on a conveniently selected subset of the population, without any explanation
for why the supposed discrimination would manifest itself in such a strange way.
SFFA tries to evade this implication of Dr. Arcidiacono’s flawed analysis by accusing
Dr. Card (at 27) of taking the view that “there is no penalty against Asian-American applicants
unless Harvard imposes a penalty on every Asian-American applicant.” But Dr. Card never said
that; SFFA misleadingly quotes Dr. Arcidiacono’s words (Ellsworth Ex. 35 at 3) as if they were
Dr. Card’s. Dr. Card’s actual point, and Harvard’s, is that where a statistical disparity in
outcomes is confined to limited subgroups of the allegedly targeted group, it is implausible to
infer that the disparity reflects intentional discrimination on the basis of race rather than the
effects of statistically unobserved factors—particularly where the plaintiff fails to explain why
the alleged discrimination would affect only a segment of the supposedly targeted group.
Ellsworth Ex. 33 ¶¶ 159, 173.
b.

A second flaw in Dr. Arcidiacono’s regression analysis is that it excludes an

integral component of the Admissions Office’s review of applicants: the personal rating. The
personal rating reflects a wide range of information in the application that bears on whether the
applicant will fulfill the ambitions Harvard has for its applicants: to be the kinds of roommates,
classmates, and teammates that will contribute to Harvard’s educational environment for all its

3

Nor does SFFA offer any response to Dr. Card’s findings that, for Asian-American
applicants who are female or those who apply from California—categories that together include
most of Harvard’s Asian-American applicants—there is a positive association between AsianAmerican ethnicity and the likelihood of admission in most years. See Harvard SJ SMF ¶¶ 221222. Those findings underscore the utter implausibility of SFFA’s theory of discrimination.
- 10 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 15 of 51

students and, once they graduate, to be citizen-leaders of our society. See, e.g., Harvard SJ SMF
¶¶ 59-60, 231; Ellsworth Ex. 149 (Worth Decl.) ¶ 9. Harvard recognizes that applicants’ ability
to succeed and contribute at Harvard and beyond is not just a function of their prior academic,
extracurricular, and athletic successes; it also reflects a broader range of personal characteristics.
Ellsworth Ex. 55 at 9-11; Ellsworth Ex. 149 (Worth Decl.) ¶ 9. The examination of applicants’
personal characteristics is therefore fundamental to Harvard’s admissions process, which
considers the whole person and does not seek simply to admit the applicants with the highest
grades or standardized test scores. Harvard SJ SMF ¶ 25; Ellsworth Ex. 55 at 10. That is
particularly true given the abundance of academically excellent applicants among whom the
Admissions Committee must choose. Harvard SJ SMF ¶¶ 1-10, 23. Including the personal
rating in a statistical model of the admissions process is the only way such a model could even
attempt to capture the intangible and unquantifiable factors that the rating reflects. See Harvard
Mem. 42. A model that disregards the personal rating and the information underlying that rating
thus does not reflect the actual admissions process and is fundamentally flawed and unreliable.
SFFA attacks (at 28-32) Dr. Card’s choice to include the personal rating in his analysis.
But as explained in Harvard’s summary judgment brief (at 42-44), that choice is not only
defensible; it is the only methodologically sound approach. There is no evidence, statistical or
otherwise, that the personal rating is discriminatory. Ellsworth Ex. 33 ¶¶ 145-156; Ellsworth Ex.
37 ¶¶ 35-57. At most, Dr. Arcidiacono’s analysis shows an association between Asian-American
ethnicity and the personal rating, not that any such association results from discrimination on the
basis of race. See Ellsworth Ex. 37 ¶ 41. In other words, it may show correlation rather than
causation. See id. Dr. Arcidiacono is well aware (as any economist would be) of the distinction
between correlation and causation. In his analyses of the academic and extracurricular ratings,

- 11 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 16 of 51

for example, he found a positive association between the ratings and Asian-American ethnicity—
yet he attributed that association to unobserved factors, not bias in favor of Asian-American
applicants. Id. ¶ 43. Yet when he found that the personal rating was negatively associated with
Asian-American ethnicity, he rushed to embrace bias instead of unobserved factors as the
explanation, with no explanation for the obvious inconsistency. Ellsworth Ex. 27 at 176:4177:18; see Harvard Mem. 43-44.
SFFA accuses Dr. Card (at 28) of having engaged in “offensive racial stereotyping” in
describing how the observed pattern in personal ratings could result from factors other than racial
discrimination. That accusation is beyond the pale even in hotly contested litigation. As SFFA
well knows, Dr. Card’s reports and testimony do not take the position that statistical analysis can
definitively explain why applicants receive the personal ratings they do. E.g., Ellsworth Ex. 33
¶ 147. To the contrary, Dr. Card recognized that the personal rating cannot reliably be modeled
because so many of the factors that inform the rating are difficult to quantify. E.g., id. Nor did
Dr. Card ever say—as SFFA wrongly insinuates (at 28)—that “Asian Americans have … a less
‘positive personality,’ ‘others like to be around’ them less, they have worse ‘character traits,’”
and so on. Those are not quotations from Dr. Card’s testimony or reports; SFFA uses quotations
drawn from other parts of the record, in what can only have been a deliberately misleading
manner, to suggest that they come from Dr. Card. See SFFA SJ SMF ¶ 90. Dr. Card did show,
including through the use of Dr. Arcidiacono’s own model, that the applications submitted by
White applicants to Harvard are on average slightly stronger in quantifiable non-academic
respects than those submitted by Asian-American applicants. See, e.g., Ellsworth Ex. 33 ¶¶ 7278, 136-137; Ellsworth Ex. 37 ¶¶ 28-30. That analysis does not reflect “offensive racial
stereotyping”; it reflects a careful and sound study of the available data. SFFA’s resort to

- 12 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 17 of 51

flagrant mischaracterizations of Dr. Card’s analysis reveals SFFA’s inability to offer any
substantive response to Dr. Card’s work and its apparent effort to create sensationalized
allegations for public consumption. 4
SFFA tries (at 30-31) to bolster its criticisms of Dr. Card by arguing that the personal
ratings assigned by alumni interviewers do not show the same disparity as those assigned by
admissions officers. But the alumni-interviewer and admissions-officer personal ratings reflect
fundamentally different information and serve fundamentally different purposes. The alumniinterviewer personal rating reflects only what the interviewer has heard from the applicant during
a single short interview. Ellsworth Ex. 54 at 28, 31. By contrast, the admissions officer’s
personal rating reflects a much broader range of information, including the applicant’s essays,
the recommendations of teachers and counselors who may have known the applicant for years,
and everything else in the application file related to the applicant’s background and life story.
See, e.g., Harvard SJ SMF ¶ 59. It is therefore unsurprising that admissions-officer and alumniinterviewer personal ratings may diverge; the ratings reflect different content. 5
SFFA further argues (at 30) that “[t]he scores submitted by teachers and counselors …
also show Asian-American applicants as having personal qualities that are comparable to white

4

It is equally baseless for SFFA to suggest (at 28-29) that Dr. Card’s analysis relies on
evidence to which SFFA lacked access. Dr. Card had access to the same application files and
summary sheets that SFFA and Dr. Arcidiacono were given. Dr. Card’s statements that (for
example) “the applicant’s essay is an important consideration in the personal rating,” Ellsworth
Ex. 33 ¶ 19, reflect his review of testimony and documents in the record, not any documents to
which Dr. Arcidiacono lacked access.
5

In addition, admissions officers have the perspective that comes from reviewing
thousands of application files over many years, whereas alumni interviewers may meet only a
handful of applicants. Perhaps for that reason, alumni-interviewer personal ratings of all
applicants are much higher, on average, than the admissions-officer personal ratings. Ellsworth
Ex. 33 ¶ 156 (alumni interviewers assign 62% of all applicants a personal rating of 1 or 2,
whereas admissions officers assign only 23% of all applicants a personal rating of 1 or 2).

- 13 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 18 of 51

applicants.” But because there is only one rating assigned to each letter from a teacher or
counselor, it is impossible to tell whether the ratings for a given applicant reflect the teachers’
and counselors’ perceptions of the applicant’s personal strengths, strengths in other dimensions
(including academic and extracurricular performance), or both. Harvard SJ SMF ¶ 231.
Finally, SFFA resorts to invoking data about applicants to Harvard from New York
City’s Stuyvesant High School, showing the admission rate of Asian-American and White
applicants from Stuyvesant over the six admissions cycles for which data were produced. See
SFFA Mem. 30; Declaration of Peter Arcidiacono tbl. 2. But as Dr. Card explains, Dr.
Arcidiacono fails to control for differences between the applications submitted by AsianAmerican and White applicants from Stuyvesant, other than the applicants’ race. Declaration of
David Card (“Card Decl.”) ¶ 3. When Dr. Card does control for those differences, he finds—
even using Dr. Arcidiacono’s own model, which excludes the personal rating—that Stuyvesant’s
Asian-American applicants were admitted at almost exactly the rate one would have expected
given the many factors considered in Harvard’s admissions process (indeed, a rate slightly higher
than one would have expected). Id. ¶ 4-8 & Ex. 1. He also finds that the differences between the
applications submitted by Asian-American and White applicants—rather than the race of the
Asian-American applicants—explained the lower rate at which the Asian-American applicants
were admitted. Id. ¶¶ 3-8 & Ex. 1. The cherry-picked Stuyvesant numbers prove nothing at all.
Beyond the misleading nature of its statistics, SFFA also does not try to explain—nor
could it—how data from one high school, among the thousands from which Harvard receives
applications, could shed light on whether it is proper to consider the personal rating in a
regression model of Harvard’s admissions process. Nor does SFFA even try to argue that the
Stuyvesant example is anything more than anecdotal; SFFA is surely aware that, at several of the

- 14 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 19 of 51

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 20 of 51

Those comparisons shed no light on the issues in this case. In addition to the many
individual problems with the analyses on which SFFA relies (discussed below), they share an
overarching flaw: a myopic focus on academic qualifications, to the exclusion of all else that
matters in Harvard’s admissions process. As Dr. Card explains, Harvard’s applicant pool is full
of academically excellent candidates; indeed, academic excellence is among the qualities most
abundant in the pool. See Ellsworth Ex. 33 ¶¶ 29-32. Harvard’s admissions process therefore
turns, to a considerable degree, on the innumerable non-academic factors that might set one
applicant apart from another. Id. ¶¶ 33-44. The purpose of a regression analysis is to attempt
(however imperfectly) to account for the many factors that distinguish one applicant from
another. By contrast, SFFA’s descriptive statistics ignore the many non-academic considerations
that may explain why applicants within the same academic decile have different outcomes.
SFFA tries to sidestep the limitations of the crude comparisons on which it relies by
invoking (at 30) the supposed presumption that unobserved characteristics will typically “move
in the same direction as observed characteristics.” If that presumption were correct, and if it held
true in this case, it would mean that one might expect a group of applicants to be strong in nonacademic dimensions (many of which are statistically unobservable) if they are strong in
academic dimensions (which generally are observable). In essence, SFFA suggests that instead
of actually trying to analyze the many non-academic considerations in the applicant process, the
Court should simply assume they redound to the benefit or detriment of particular applicants.
But even if it were often true that unobserved characteristics “move in the same direction as
observed characteristics”—a proposition by no means obvious, for which SFFA cites only Dr.
Arcidiacono’s bare assertions, see SFFA SJ SMF ¶ 765; Ellsworth Ex. 31 at 68, 77—that is not
true in the context of this case. In reality, Dr. Card found—including by using Dr. Arcidiacono’s

- 16 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 21 of 51

own model—that the applications submitted by Asian-American applicants were slightly
stronger across academic measures, but slightly less strong than those submitted by White
applicants across a range of observable non-academic measures. See Ellsworth Ex. 33 ¶¶ 72-78,
136-137; Ellsworth Ex. 37 ¶¶ 28-30 & Card Exs. 2, 3. 7 There is no reason to doubt that the same
pattern would hold for unobservable non-academic attributes. See, e.g., Ellsworth Ex. 37 ¶ 45.
SFFA first argues (at 7-9) that the personal rating must be biased against Asian-American
applicants because “alumni interviewers … rate Asian Americans, on average, at the top with
respect to personal ratings,” whereas admissions officers assign Asian-American applicants
slightly lower personal ratings on average. But the alumni-interviewer and admissions-officer
personal ratings reflect fundamentally different information and serve fundamentally different
purposes. See supra p. 13. Moreover, according to Dr. Arcidiacono’s own data, alumni
interviewers in fact assign Asian-American applicants slightly lower personal ratings than White
applicants both on average across the entire applicant pool and in the top nine deciles of
academic performance. See Ellsworth Ex. 31 at Table B.5.5.
SFFA next invokes (at 9) Dr. Arcidiacono’s assertion that Asian-American applicants
received slightly lower overall ratings, on average, than White applicants within each academic
decile. That shows nothing, however. The overall rating, like Harvard’s admissions process
more generally, considers much beyond applicants’ prior academic success. See, e.g., Harvard
SJ SMF ¶ 64. The fact that applicants within an academic decile receive higher or lower overall
ratings than others within the same decile does not mean the overall rating is somehow deceptive
or discriminatory. Rather, it means nothing more remarkable than that the applicants with higher

7

Dr. Arcidiacono himself has made a similar observation. See Ellsworth Ex. 122 at 27:1830:23 (addressing a paper concluding that students attending certain MBA programs were
stronger on observable measures but potentially less strong in unobservable respects).

- 17 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 22 of 51

overall ratings were stronger in non-academic respects than the applicants with lower overall
ratings. Regression analysis can attempt to control for those non-academic differences among
applicants, whereas the crude comparison on which SFFA relies cannot.
As with the personal rating, SFFA also compares the overall ratings assigned to
applicants by admissions officers and those assigned by alumni interviewers, claiming (at 9) that
alumni interviewers, unlike admissions officers, awarded Asian-American applicants “overall
scores … virtually identical to those of white applicants in the top” academic decile. But again,
that claim ignores the fundamental differences between the ratings assigned by admissions
officers and alumni interviewers—particularly the fact that admissions officers have access to a
broad range of application materials, including comments from teachers, guidance counselors,
and others, whereas interviewers meet with the applicant for an hour or less.
Finally, SFFA invokes (at 9-10) Dr. Arcidiacono’s assertion that Asian-American
applicants were admitted at a lower rate than applicants of other races within each academic
decile except one. Again, however, that proves nothing other than SFFA’s mistakenly inflated
view of the role academic factors play in the admissions process. Because excellent academic
credentials are abundant in Harvard’s applicant pool, Ellsworth Ex. 33 ¶¶ 26-32 & Card Exs. 2-3,
and Harvard relies on many other factors to distinguish among academically excellent
candidates, it is unsurprising that applicants within a particular academic decile are admitted at
different rates. That simply reflects differences among the applicants in non-academic respects.
In sum, the crude descriptive comparisons on which SFFA relies come no closer to
proving intentional discrimination than Dr. Arcidiacono’s pervasively flawed—and now
apparently abandoned—regression analysis. SFFA’s statistical evidence is not even strong
enough to allow its claim to survive Harvard’s summary judgment motion.

- 18 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 23 of 51

B.

The OIR Documents Are Not Evidence—Let Alone Undisputed Evidence—
Of Discrimination

SFFA also relies (at 11-20) on documents generated by individuals in Harvard’s Office of
Institutional Research (OIR). As discussed in Harvard’s summary judgment brief (at 38),
however, those documents no more support summary judgment in SFFA’s favor than does the
statistical record. To the extent SFFA is relying on those documents for the substance of the
preliminary and incomplete analysis they contain, that analysis is rendered irrelevant by Dr.
Card’s far more comprehensive, informed, and reliable work. More generally, SFFA’s
interpretation of those documents is both unwarranted and plainly disputed.
a.

SFFA begins by mischaracterizing the documents themselves. It breathlessly

claims (at 11) that they are the products of an “internal investigation … into Harvard’s treatment
of Asian Americans,” that the supposed “investigation” was “conducted in consultation with the
Admissions Office,” and that “OIR concluded that Harvard’s admission system is, in fact, biased
against Asian Americans and that there is no neutral explanation for it.” None of that is remotely
true.
OIR was not conducting an “internal investigation” of anything. The documents
originated within OIR and not in response to a request from the Admissions Office or the Office
of the General Counsel. Ellsworth Ex. 150 (Declaration of Erin Driver-Linn (“Driver-Linn
Decl.”)) ¶¶ 9, 17, 30. OIR Director Erin Driver-Linn testified, for example, that the analysis in
question was not directed to “whether there is bias against Asians in college admissions at
Harvard,” Ellsworth Ex. 118 at 165:9-16; that it focused instead on “evaluating factors that play
a role in Harvard College admission,” id. at 166:4-13; and that “no person outside of OIR asked”
OIR to conduct the analysis, id. at 145:7-15. Consistent with its origins, the work done by OIR
employees was not intended to address whether Asian-American applicants were experiencing

- 19 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 24 of 51

discrimination and did not answer that question. Ellsworth Ex. 150 (Driver-Linn Decl.) ¶¶ 9, 17,
30. For example, when asked whether “any of the work that [he] did while at OIR showed that
Harvard College discriminates against Asian[] Americans,” former OIR employee Mark Hansen
answered “[n]o.” Ellsworth Ex. 20 at 193:24-194:4. He explained that, because the admissions
process considers factors not observable in the data to which he had access, he lacked “enough
information” to competently assess whether there was evidence of discrimination against AsianAmerican applicants. Ellsworth Ex. 116 at 137:20-138:21.
The documents were not prepared “in consultation with the Admissions Office,” as SFFA
contends (at 11) without a single citation to the record. The OIR employees performed their
analysis on their own, not together with employees of the Admissions Office, and acknowledged
that they did not know enough about the admissions process to render reliable conclusions. See,
e.g., Ellsworth Ex. 115 at 134:19-138:17 (OIR “oversimplified the [admissions] process”),
156:12-24 (“[T]his does not reflect the process by which we do admissions … [b]ecause we
review many factors, some of which can be data and some of which are not.”); Ellsworth Ex. 116
at 116:20-117:4, 137:20-138:21, 195:23-196:18 (admissions process considers factors and data
not reflected in the OIR models).
Indeed, the OIR documents on their face acknowledge that any analysis would need
much more information to approach a reliable understanding of Harvard’s admissions process.
One document states, for example, that “[t]here are a variety of factors that quantitative data is
likely to miss or ratings do not capture,” and lists factors analysts would “like to better
understand.” Ellsworth Ex. 65 at 36. Another lists factors for which OIR’s model did not
account: “Children faculty/staff,” “Search for socioeconomic diversity,” “High school
quality/opportunities open to student,” and “Dockets.” Ellsworth Ex. 66 at 17. Consistent with

- 20 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 25 of 51

all that evidence, Ms. Driver-Linn testified that the analysis was “exploratory, preliminary, and
limited” in nature. Ellsworth Ex. 23 at 196:8-10. Dr. Card’s analysis—which found no
statistically significant evidence of discrimination against Asian-American applicants—
incorporated much of the information that OIR acknowledged was missing from its model, since
Dr. Card (and Dr. Arcidiacono) had access to substantially broader data than did the OIR
employees. See Harvard SJ SMF ¶ 218.
SFFA tries (at 14) to spin the gaps in OIR employees’ understanding of the admissions
process—such as the fact that one draft analysis contained blank slides labeled “Conclusions”
and “Possible Explanations”—as incriminating. But those blank slides are hardly surprising
given how incomplete the documents were. For example, the very document that contained the
blank slides also had the subtitle left blank on the cover page. Declaration of Michael Connolly
in Support of Plaintiff’s Motion for Summary Judgment (“Connolly”) (Dkt. 421), Ex. 145 at 1
(“Admissions Part II Subtitle”); Ellsworth Ex. 150 (Driver-Linn Decl.) ¶ 11. And to the extent
the blank slides have any meaning at all, they prove Harvard’s point: They show that the OIR
employees were well aware of the significant limitations of their work. SFFA is wrong to
suggest (at 17), moreover, that no one “question[ed] the quality or thoroughness of OIR’s
findings when they were made”; the OIR employees themselves explicitly noted the
shortcomings of their analysis on the face of the documents.
b.

SFFA also mischaracterizes the reactions to the OIR documents of others at

Harvard, including William Fitzsimmons, Dean of Admissions and Financial Aid; and Rakesh
Khurana, Danoff Dean of Harvard College. SFFA makes the headline-grabbing claim (at 15)
that, “[f]aced with an internal investigation showing systemic discrimination against AsianAmerican applicants, Harvard killed the study and quietly buried the reports.” But there was no

- 21 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 26 of 51

investigation to kill and no adverse findings to bury; the analysis shown in the documents neither
sought to address whether there was discrimination against Asian-American applicants nor
suggested that such discrimination was taking place. See supra pp. 19-20. Moreover, the
evident limitations of the analysis are all the explanation that is needed for the way in which the
documents were perceived by those who viewed them. For example, Dean Fitzsimmons testified
that “with a very, very incomplete model with a great deal of information missing … it’s unclear
what to do with it.” Ellsworth Ex. 26 at 407:5-11. 8 Indeed, even the OIR employees who
showed the documents to Dean Fitzsimmons expressed the concern that they might have tried “to
take complex things and reduce them to overly simplistic” analyses, Ellsworth Ex. 118 at
192:24-193:8, for they understood that their analysis did not take into account many factors that
might have affected admissions decisions, Ellsworth Ex. 20 at 196:7-18. And Dean Khurana—
who was shown the OIR analysis simply as an example of the types of analyses OIR periodically
conducts with respect to Harvard College, see Connolly Ex. 162, and whose academic expertise
is in social science—likewise testified he “didn’t think that th[e] … analysis was done
appropriately” because there are “a lot of limitations to doing … models like this.” Ellsworth
Ex. 8 at 253:13-20. 9

8

Dean Fitzsimmons was also intimately aware of how the admissions process operated,
the training given to admissions staff, and the role of the 40-person committee, all of which
allowed him to be confident that the limited OIR work did not require further action on his part.
9

SFFA attacks (at 16-17) the supposed “amnesia” of Harvard witnesses—Dean
Fitzsimmons, Dean Khurana, Ms. Driver-Linn, and former OIR employee Erica Bever—who
were asked about the OIR documents. But it is hardly surprising that those busy people
remembered only some of this partial and preliminary analysis that they had seen or worked on
years earlier—especially when they explained why they found the analysis unilluminating. See,
e.g., Morse v. Hanks, 172 F.3d 983, 984 & n.1 (7th Cir. 1999) (“understandable” that a
prosecutor could not recall events from six years earlier); Johnson v. Dismore, 2012 WL
1855031, at *6 (S.D. Ill. Apr. 24, 2012) (“perceived discrepancy” in a witness’s testimony was
not “damaging to [its] overall truthfulness,” since “[t]he events in question occurred almost two
years ago” and the witness could not “be expected to remember every detail”), adopted, 2012
- 22 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 27 of 51

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 28 of 51

“standard strong” is not an epithet; it is simply a phrase “used to describe an applicant who is
very well qualified academically and likely has a good deal of extracurricular involvement as
well but isn’t distinguished in Harvard’s incredibly, incredibly competitive applicant pool.”
Connolly Ex. 25 at 229:10-15; see also Ellsworth Ex. 102 at 73:20-74:6 (“standard strong” refers
to “the average student who applies to Harvard with qualities that make the student an
appropriate applicant to the school”). In other words, it refers to an applicant who is very good
but who does not rise to the top of an exceptionally competitive pool. Nor is “standard strong” a
racially coded phrase. The record contains numerous examples of excellent applicants of other
races who were described as “standard strong.” See, e.g., Ellsworth Ex. 139; Ellsworth Ex. 136;
Ellsworth Ex. 140. In its 900 supposedly undisputed material “facts,” SFFA neglected to include
those examples.
Nor is SFFA correct that Asian-American applicants who were described as “standard
strong” were stronger than applicants of other racial backgrounds described in the same way
(even if the sample of summary sheets were sufficiently broad to allow reliable conclusions to be
drawn, which SFFA has previously suggested it is not, see Letter from William S. Consovoy to
Court at 3 (Feb. 3, 2017)). SFFA asserts (at 20) that Asian-American applicants described as
“‘standard strong’” were “‘substantially more qualified academically’” than applicants of other
races who were so described. But as noted above, “standard strong” refers not to academic
strength but to strength across the full range of factors the Admissions Committee considers.
The expert analysis conducted by Dr. Card found that Asian-American and White applicants who
were described as “standard strong” in the produced sample of summary sheets were in fact
equally strong across all dimensions, as measured by the sum of their academic, extracurricular,

- 24 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 29 of 51

athletic, and personal ratings. See Ellsworth Ex. 33 ¶ 75 n.65. Asian-American applicants were
not held to a higher standard.
Second, SFFA points (at 20) to a handful of comments on summary sheets in which
Asian-American applicants were described in terms that SFFA claims to be discriminatory. But
there is nothing discriminatory or racially coded about those comments. For example, SFFA
points (at 20) to a single Asian-American applicant described as “busy and bright,” as if that
were somehow a racial epithet. But elsewhere SFFA highlights an African-American applicant
described as “bright and perfectly busy.” SFFA SMF ¶ 695. SFFA also points to a single AsianAmerican applicant described as needing to “fight it out with” similar candidates. SFFA Mem.
20 (citing SFFA SMF ¶ 690). But similar comments were applied to candidates of other racial
backgrounds as well. See, e.g., Ellsworth Ex. 132 (questioning how a White applicant would
“compare in the pool”); Ellsworth Ex. 134 (noting that a Hispanic applicant might not “surviv[e]
this competition”); Ellsworth Ex. 135 (observing that a White applicant would not “jump[] out in
our pool”); Ellsworth Ex. 137 (referring to a White applicant as “very SS [standard strong] given
the competition”).
SFFA also claims (at 20) that the ethnicity of Asian-American applicants “is rarely seen
as a positive factor in the chances of admissions,” but the record shows the opposite. See, e.g.,
Ellsworth Ex. 141 (comment that “[applicant] is a very deserving student from a first generation
Vietnamese background who is valedictorian for this city-wide magnet school”); Ellsworth Ex.
131 (applicant of Nepali descent “[c]ertainly would bring a fascinating perspective to campus”);
Ellsworth Ex. 142 (“Tug for BG [background] here, she writes well about the plight of exiled
Tibetans and T2 [second teacher recommendation] lets us know that both of her parents were
born in Tibetan refugee camps in India.”); Ellsworth Ex. 130 (“B/G [background] of interest as

- 25 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 30 of 51

he [applicant of Indian origin] would be someone who would add to the mix at H.”); Ellsworth
Ex. 143 (comment that applicant was “involved in the Asian community as EIC [editor-in-chief]
of local journal”). Moreover, the record shows that Harvard does not (as SFFA’s arguments
suggest) treat Asian-American candidates simply as a bloc; rather, Harvard’s admissions officers
have a nuanced understanding of the many differences within Asian-American communities and
of the ways in which those differences may enrich the diversity of the Harvard student body.
See, e.g., Ellsworth Ex. 127 (training presentation urging admissions officers to “be wary of
aggregated data for Asian American student populations” and to “[h]onor the nuance of both
identity and context”).
Third, SFFA points (at 21) to a handful of instances in which individuals—a single OIR
employee, two alumni interviewers, and a single high-school student—raised questions about
how Harvard treats Asian-American applicants, and criticizes Harvard for not launching an
investigation in response to those inquiries. But those examples do not even permit an inference
of intentional discrimination by Harvard; much less do they support summary judgment in
SFFA’s favor on the claim. Harvard’s admissions process is so competitive, and the hopes
associated with the possibility of admission so intense, that individuals regularly complain about
admissions outcomes; such episodic complaints are hardly sufficient to warrant the kind of
investigation SFFA seems to think was justified.
SFFA also relies (at 22-23) on an email exchange between the Director of Admissions
and

and another exchange between the

Director and a more than ninety-year-old alumnus who urged Harvard to adopt an “informal
quota[]” to limit the number of Asian-American students. That correspondence comes no closer
to suggesting discriminatory intent, and certainly not undisputedly so. Moreover, just as in the

- 26 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 31 of 51

analogous context of employment discrimination, “‘stray workplace remarks’ … normally are
insufficient, standing alone, to establish … discriminatory animus,” Gonzalez v. El Dia, Inc., 304
F.3d 63, 69 (1st Cir. 2002), and any “probativeness” of “such ‘stray remarks’ … is circumscribed
if they … were not related to the … decision in question,” Straughn v. Delta Air Lines, Inc., 250
F.3d 23, 36 (1st Cir. 2001) (emphasis omitted). Even under SFFA’s unreasonable construction
of the documents, the “stray remarks” they contain cannot show that the members of Harvard’s
Admissions Committee conspired to discriminate against Asian-American applicants. Much less
can those episodes be considered undisputed evidence of discrimination, given the overwhelming
record evidence that Harvard did not discriminate against Asian-American applicants.
Indeed, SFFA’s reliance on a handful of isolated documents (even if SFFA’s construction
of those documents were reasonable, which it is not) underscores its inability to point to any
genuine evidence of the years-long scheme of intentional discrimination it alleges. SFFA lacks
even a theory of how such a conspiracy could have been accomplished among the many
members of the Admissions Committee. The Admissions Office procedures and training
materials are all in the record (e.g., Ellsworth Exs. 51-57), and they contain no hint of a
concerted plot against Asian Americans. The many members and former members of the
Admissions Office who have been deposed have uniformly denied the allegations. See, e.g.,
Ellsworth Ex. 102 at 33:12-16, 186:17-187:12; Ellsworth Ex. 105 at 207:5-23; Connolly Ex. 11
at 157:9-17; Ellsworth Ex. 114 at 128:1-7; Ellsworth Ex. 14 at 54:2-23; Ellsworth Ex. 113 at
63:9-13, 105:16-25; Ellsworth Ex. 120 at 336:12-15, 443:5-13; Ellsworth Ex. 1 at 231:9-232:3;
Ellsworth Ex. 12 at 83:12-84:2; Ellsworth Ex. 3 at 239:13-15. It is inconceivable that a yearslong conspiracy among more than 40 people would produce no evidence of its existence.

- 27 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 32 of 51

D.

As The Court Has Already Recognized, Harvard’s Treatment Of Jewish
Applicants In The 1920s Is Irrelevant To This Case

Finally, SFFA invokes (at 23-26) the century-old history of Harvard’s treatment of
Jewish applicants in the 1920s, arguing that it somehow renders unlawful the entire enterprise of
whole-person admissions employed not only at Harvard but at many other universities across the
country. It does so even though this Court has rightly and repeatedly held that history to be
irrelevant to SFFA’s claims, see Dkt. 181 at 2-3; Dkt. 193 at 15:17-17:12. Indeed, it does so
without even acknowledging the Court’s ruling, let alone the requirement that summary
judgment be based only on relevant, admissible evidence, see Fed. R Civ. P. 56(c); Fed. R. Evid.
402. And it does so even though Justice Powell’s opinion in Regents of the University of
California v. Bakke, 438 U.S. 265 (1978), pointed to Harvard’s process as the paradigm of the
lawful consideration of race, id. at 316-317.
SFFA apparently thinks that events of a century ago help its case based on the Supreme
Court’s statement in Village of Arlington Heights v. Metropolitan Housing Development Corp.,
429 U.S. 252 (1977), that the background of a policy can shed light on whether it is intentionally
discriminatory. But even assuming Arlington Heights is relevant here—and even if the 1920sera history had any probative value today, which this Court has already ruled it does not—
SFFA’s invocation (at 23) of what it calls Harvard’s “original sin” ignores the principle that
“‘[p]ast discrimination cannot, in the manner of original sin, condemn … action that is not itself
unlawful.’ The ‘ultimate question remains whether a discriminatory intent has been proved in a
given case.’” Abbott v. Perez, 138 S. Ct. 2305, 2324-2325 (2018) (quoting Mobile v. Bolden,
446 U.S. 55, 74 (1980); citation omitted). As this Court has recognized, the events of nearly a
century ago shed no light on that question. SFFA’s invocation of ancient history should be seen

- 28 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 33 of 51

for what it is: a publicity-seeking attempt to distract from its lack of any evidence that Harvard
discriminates against Asian-American applicants.
In sum, SFFA has no evidence—and certainly no undisputed evidence—that Harvard has
violated Title VI’s prohibition against racial discrimination. The Court should deny its motion
for summary judgment on Count I.
II.

HARVARD DOES NOT ENGAGE IN RACIAL BALANCING
SFFA next contends (at 34) that it is entitled to summary judgment on Count II because

“Harvard has a desired racial balance and aims for that target.” As support for that assertion,
SFFA claims that “Harvard’s admissions statistics remain stubbornly consistent across every
racial group.” Id. A reasonable factfinder could only conclude that both assertions are untrue.
At the very least, the record demonstrates that SFFA is not entitled to summary judgment on this
count.
It is surprising that SFFA makes this claim, for not even its own expert supports the
argument. The data clearly show that the racial composition of Harvard’s admitted class
fluctuated meaningfully from year to year over a 19-year period. Ellsworth Ex. 33 ¶¶ 193-194 &
Card Ex. 31-34. SFFA’s expert, Dr. Arcidiacono, does not dispute that fact; to the contrary, Dr.
Arcidiacono’s rebuttal report admitted that the data showed “changes in the fraction of admitted
students by race/ethnicity over time” and never suggested that those changes were statistically
insignificant. Ellsworth Ex. 35 at 58. As explained in Harvard’s summary judgment brief (at
18-20), the undisputed variation in the racial composition of the class warrants summary
judgment for Harvard on Count II.
SFFA’s only supposed statistical evidence of Harvard’s alleged balancing is a chart (at
34) that shows the racial composition of Harvard’s admitted classes over four years. But the
chart proves Harvard’s point—it shows significant fluctuations from year to year in the racial
- 29 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 34 of 51

composition of the class. For example, SFFA’s chart shows an 11% increase in the AsianAmerican share of the admitted class between the Class of 2015 and the Class of 2016, a 17%
decrease in the African-American share of the admitted class between the Class of 2015 and the
Class of 2016, and a 20% increase in the Hispanic share of the admitted class between the Class
of 2014 and the Class of 2015. Far from showing “stubbornly consistent” enrollment patterns,
the data show meaningful fluctuations over time. Cf. Grutter v. Bollinger, 539 U.S. 306, 336
(2003) (variations in the total number of minority students in each class were “inconsistent with
a quota”). And, again, SFFA mounts no argument whatsoever based on the 19 years of data
available to it.
SFFA also fails to identify any documentary or testimonial evidence that Harvard seeks a
particular racial balance for its incoming class. SFFA points to Director McGrath’s statement
that Harvard seeks to “mak[e] certain that there’s good representation of the various [racial]
groups whose numbers we track.” SFFA Mem. 34 (quoting McGrath 2015 Dep. 248:12-14).
But that unremarkable statement suggests merely that Harvard seeks broad diversity in its
classes—not any specific targets. Moreover, SFFA conveniently ignores Director McGrath’s
clear and unrebutted testimony that the Admissions Committee is “not ever looking for a
particular number or percentage.” Ellsworth Ex. 98 at 240:18-19; see also id. 252:4-10 (“Q.
Your testimony is that [Harvard] doesn’t have any specific target or minimum level for racial
groups on campus? A. We do not. Q. And it does not have a minimum level for individual
racial groups on campus? A. No. Not a specific number.”).
SFFA’s allegation (at 34-35) that Harvard’s recruiting efforts demonstrate racial
balancing is no better supported. The Supreme Court has endorsed recruiting as a means of
achieving the educational benefits of diversity. See Fisher II, 136 S. Ct. at 2213 (noting the

- 30 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 35 of 51

University of Texas at Austin’s “extensive … outreach efforts” to minority students); id. at 2236
(Alito, J., dissenting) (suggesting that the university should have “intensif[ied] its outreach
efforts” instead of considering race). Indeed, SFFA and its expert, Mr. Kahlenberg, suggest that
Harvard could and should engage in even more recruiting to achieve the educational benefits of
diversity. See, e.g., Dkt. 1 ¶¶ 332-335; Ellsworth Ex. 32 at 39-40. Harvard sends informational
material to more than

applicants of all racial and ethnic backgrounds each year.

Connolly Ex. 113 at HARV00023564. In recognition of the fact that

But nothing about that fact suggests that
applicants of the same race compete only against each other for admission, let alone that Harvard
has some preconceived or desired racial balance. Rather, the mailings reflect Harvard’s efforts
to reach the broadest spectrum of applicants who might be competitive in Harvard’s admissions
pool, in which all applicants compete against each other. See Ellsworth Ex. 26 at 335:5-7 (“The
point of recruitment is to get the strongest possible class across many dimensions, including
ethnicity.”). 10

10

SFFA also contends (at 35-36) that Harvard’s participation in meetings of the Association
of Black Admissions and Financial Aid Officers of the Ivy League and Sister Schools
(“ABAFAOILSS”) is evidence of racial balancing because admissions officers from
participating schools share admissions statistics at an annual meeting. SFFA implies that after a
Harvard admissions officer attends the meeting, Harvard somehow manipulates admissions rates
based on information gleaned from its peers. There is no evidence that the information shared at
ABAFAOILSS has any effect on admissions decisions. To the contrary, the admissions officer
who attended those ABAFAOILSS meetings testified that the numbers were “never” shared in
Harvard subcommittee or full committee meetings, nor were they ever mentioned in other
meetings in the Harvard admissions office. Ellsworth Ex. 105 at 158:24-159:10. Dean
Fitzsimmons testified that the information, to the extent he received it, was simply “useful in
getting to know what happens beyond Harvard” and never suggested that this information had
any effect on Harvard’s admissions decisions. Connolly Ex. 9 at 452:18-19.

- 31 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 36 of 51

SFFA also contends (at 35-38) that Harvard’s use of a target number of total number of
students of all races to admit each year somehow amounts to racial balancing. Using such a
target based on predicted yield rates is an essential and unremarkable feature of any admissions
process. Harvard is a residential college with only so many beds available in each class—usually
1,662. Ellsworth Ex. 120 at 283:11-13. Harvard must therefore be certain that it does not offer
admission to too many applicants and that those offered admission do not yield too many
enrolled students. Ellsworth Ex. 120 at 285:8-10; Connolly Ex. 16 at 212:14-213:2.
Estimating yield is therefore critical to the Harvard admissions process. Harvard uses a
variety of information to estimate the yield from those it offers admission. Harvard starts with
the baseline of the prior year’s application cycle. Connolly Ex. 16 at 211:21-23; Ellsworth Ex.
120 at 281:4-8, 18-21. Moreover, Harvard’s senior admissions officers know that yield rates
vary substantially for certain types of applicants. See, e.g., Ellsworth Ex. 120 at 338:2-6
(“people from the South tend to yield at a lower rate”); Connolly Ex. 16 at 211:22-212:4
(“people who are farther away [are] less likely to take up your offer. … [A]thletes yield at a very
high rate. Humanities students have yielded at a very high rate.”), 213:12-20 (yield may vary
based on race and whether an applicant’s parent attended Harvard). Accordingly, when the
composition of the likely admitted class varies from prior years in a way that could affect the
estimated yield, senior admissions officers may need to adjust the overall target for the number
of admitted students to reflect changes in the estimated yield. Connolly Ex. 9 at 339:3-17;
Connolly Ex. 16 at 212:5-13.
SFFA also argues (at 37) that the Admissions Office engages in racial balancing when it
engages in the “lop” process, which occurs in the final days of evaluation of applicants. There is
nothing discriminatory about that process. Not surprisingly, given the abundance of excellent

- 32 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 37 of 51

candidates who apply to Harvard, the Admissions Committee may find, near the end of its
consideration, that it has tentatively decided to offer admission to too many applicants, given the
fixed number of beds available and the likely yield rate. When that happens, the Admissions
Office must determine which of those applicants cannot be admitted. Ellsworth Ex. 149 (Worth
Decl.) ¶ 20. That is an unremarkable feature of Harvard’s admissions process—and most likely
of any admissions process in which thousands of highly qualified individuals apply for a fixed
and limited number of slots.
SFFA also points (at 36-37) to the “one-pagers” received by Dean Fitzsimmons and
Director McGrath at various points in the process as evidence of racial balancing, implying that
Harvard seeks to match the racial composition of the class admitted in one year to that of the
prior year’s class. The record does not support that assertion—and certainly not undisputedly so.
The “one-pagers” contain a snapshot of the tentatively admitted class and the prior class,
breaking down the number of applicants by gender, geography, intended concentration, whether
the applicant is a recruited athlete, whether the applicant’s parent attended Harvard, whether
Harvard waived the applicant’s application fee, whether the applicant was flagged as
socioeconomically “disadvantaged” by Harvard’s admissions staff, whether the applicant applied
for financial aid, citizenship, permanent residency, and race. See, e.g., SFFA SJ Ex. 43, 70. That
is hardly evidence of a myopic or undue focus on race.
Moreover, Admissions Office leaders review the one-pagers not to ensure that they are
precisely matching numbers from prior years but to ascertain whether there are any significant
trends worth noting and to make sure that the Admissions Committee is not overlooking strong
candidates. Dean Fitzsimmons testified that variations from prior years might lead the
Admissions Office to look carefully at applicants from groups that appear to be surprisingly

- 33 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 38 of 51

underrepresented. Ellsworth Ex. 120 at 313:5-6. Director McGrath likewise testified that if (for
example) at the end of Early Action “there’s no one from the Northwest … we might say, ‘Let’s
look at the stronger ones who seem not to be in’” or if there were “no engineers in the class [and]
we are used to having early action more engineers than that, we would say … ‘If there are some
strong ones, please let’s look at those again and see if they look any better or not.’” Ellsworth
Ex. 98 at 201:19-202:6. Although race is one of numerous categories that may be reviewed in
the one-pagers—the full list is above (at 33)—that hardly amounts to an obsession with race or
racial balancing. Indeed, the Supreme Court recognized in Grutter that it is permissible for an
admissions committee to consult “daily reports” showing the racial composition of the admitted
class, 539 U.S. at 336, and stressed that “[s]ome attention to numbers, without more, does not
transform a flexible admissions system into a rigid quota,” id. (internal quotation marks omitted).
What SFFA has failed to note is the unrebutted testimony that Harvard has no targets by
race (see supra p. 30) and the unrebutted statistical evidence showing significant fluctuations in
the racial composition of the class from year to year (see supra pp. 29-30). 11 Harvard, not
SFFA, is entitled to summary judgment on Count II.

11

In a single paragraph (at 38-39), SFFA briefly restates Dr. Arcidiacono’s claim that the
admissions rate for African-American applicants, as measured using one methodology for
counting applicants by race, was very similar to the admission rate of all other domestic
applicants for the Classes of 2017 through 2019. As explained in Harvard’s summary judgment
brief (at 20-21), that bizarre claim serves only to show the lengths to which SFFA will go to
conjure a claim of racial balancing. Dr. Card explained that the phenomenon that SFFA has
identified is statistically unremarkable. Ellsworth Ex. 37 ¶ 161. Moreover, SFFA offers no
reason Harvard would seek to create such a pattern or any evidence that it deliberately did so.
See Ellsworth Ex. 5 at 137:15-138:6; Ellsworth Ex. 26 at 100:23-101:9; Ellsworth Ex. 33 ¶ 190;
Ellsworth Ex. 37 ¶ 153. SFFA has nearly abandoned Dr. Arcidiacono’s bizarre argument in its
brief, and for good reason given the unassailable rebuttal offered by Dr. Card.

- 34 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 39 of 51

III.

HARVARD CONSIDERS RACE FLEXIBLY, ALONG WITH MANY OTHER FACTORS, TO
ACHIEVE THE EDUCATIONAL BENEFITS OF DIVERSITY
SFFA next contends (at 39-40) that Harvard’s race-conscious admissions program

violates Title VI because Harvard is not seeking a “critical mass of underrepresented minority
students.” That flawed theory is not alleged in SFFA’s complaint and should be rejected for that
reason alone. See Dkt. 1 ¶¶ 456-465 (Count III); Estrada v. Progressive Direct Ins. Co., 53 F.
Supp. 3d 484, 497 (D. Mass. 2014) (“Plaintiffs cannot now introduce an entirely new theory of
liability in their summary judgment papers. … In doing so, they impermissibly seek[] to amend
[their] complaint without ever filing a motion for leave to amend pursuant to Fed. R. Civ. P. 15.”
(internal quotation marks omitted)).
SFFA’s argument also rests on a fundamental misreading of the law. The Supreme Court
has held that universities have a compelling interest “in obtaining the educational benefits that
flow from a diverse student body,” Grutter, 539 U.S. at 343, and that they may consider race to
achieve that interest. But the Supreme Court has never held that universities may use race to
achieve that compelling interest only by seeking to attain a “critical mass” of underrepresented
students. Seeking a “critical mass” is one way that universities may permissibly consider race,
but the Court has never suggested it is the only one, and the quotation SFFA uses to suggest that
pursuit of a “critical mass” is a legal requirement comes from the recitation of facts at the outset
of Grutter, not the Court’s holding. SFFA Mem. 39 (quoting Grutter, 539 U.S. at 318). 12
The Supreme Court’s decision in Fisher II confirms that pursuit of a “critical mass” is not
the relevant legal requirement. There, the Court rejected the argument that the University of

12

The sentence that SFFA quotes reads in full: “This was done, [the Michigan Director of
Admissions] testified, to ensure that a critical mass of underrepresented minority students would
be reached so as to realize the educational benefits of a diverse student body.” Grutter, 539 U.S.
at 318.

- 35 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 40 of 51

Texas at Austin had not sufficiently defined “the level of minority enrollment that would
constitute a ‘critical mass.’” 136 S. Ct. at 2210. The Court explained that “the compelling
interest that justifies consideration of race in college admissions is not an interest in enrolling a
certain number of minority students. Rather, a university may institute a race-conscious
admissions program as a means of obtaining the educational benefits that flow from student body
diversity.” Id. (internal quotation marks omitted). The Court thus rejected any notion that
universities must define their objectives in terms of critical mass—instead, their goal must be to
pursue the educational benefits of diversity.
Harvard’s goal is just that. In 2015, for example, the Committee to Study the Importance
of Student Body Diversity, chaired by Dean Khurana, “emphatically embrace[d] and
reaffirm[ed] the University’s long-held view that student body diversity—including racial
diversity—is essential to our pedagogical objectives and institutional mission.” Ellsworth Ex. 45
at 22. Harvard’s interest in the educational benefits of diversity is reflected in the mission of
Harvard College, the 1996 report authored by Harvard University President Neil Rudenstine, and
the “Harvard Admissions Program” submitted to the Supreme Court and endorsed by Justice
Powell in Bakke. See Ellsworth Ex. 68; Ellsworth Ex. 41 at 53; Bakke, 438 U.S. at 316-317
(opinion of Powell, J.).
SFFA also briefly contends that race is “a ‘predominant factor’” in Harvard’s admissions
process. It is not. The Court’s teaching in Grutter was that a university may consider race if its
consideration is “‘flexible enough to consider all pertinent elements of diversity in light of the
particular qualifications of each applicant, and to place them on the same footing for
consideration, although not necessarily according them the same weight.’” 539 U.S. at 334
(quoting Bakke, 438 U.S. at 317); see also id. (“[T]ruly individualized consideration demands

- 36 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 41 of 51

that race be used in a flexible, nonmechanical way.”). The record conclusively establishes that
Harvard’s consideration of race satisfies those requirements because the unrebutted testimony of
Harvard’s admissions officers—which SFFA again does not even mention—is that race is but
one factor among many that are considered flexibly in the admissions process. 13
The record also makes clear that Harvard’s admissions officers carefully consider each
applicant in his or her entirety, seeking a full picture of the whole person in context. See
Ellsworth Ex. 26 at 50:16-20, 51:24-52:20, 55:14-21, 233:12-234:12; Ellsworth Ex. 52;
Ellsworth Ex. 53; Ellsworth Ex. 55 at HARV00001400-1402; Ellsworth Ex. 12 at 46:6-48:5,
60:23-24; Ellsworth Ex. 98 at 231:9-233:15. Race is but a small part of that consideration. 14

13

Ellsworth Ex. 26 at 158:24-159:7 (“[I]f you’re looking for a particular formula, there are
no formulas of any sort … because people are multidimensional.”); Ellsworth Ex. 1 at 231:15232:1 (“[E]ach applicant is really considered as an individual, including … many factors, family
background, which will include whatever we know of race, whatever else we know about family
circumstance and education, whatever we can know about the nature of the school and the kind
of community the student grew up in. Those context features, those features of the student’s
setting are always important to us in imagining how well he’s achieved in the circumstances that
he started with to us as a candidate.”); Ellsworth Ex. 6 at 91:5-8 (“There are many, many factors
that are used in combination to advocate for a student [to be admitted], not any one alone
guarantees anything.”); Ellsworth Ex. 12 at 60:23-24 (“[W]e view each file individually based on
achievement and areas of excellence.”); Ellsworth Ex. 13 at 42:24-43:2 (“[W]e consider race as a
factor among many that we are looking at in the whole-person review process.”); Ellsworth Ex.
14 at 54:11-14 (“Every student would be reviewed on a variety of factors. The impact that any
individual characteristic would have would vary by student and depend on the student.”);
Ellsworth Ex. 3 at 239:13-14 (“Race was just one factor of many factors that were considered in
an applicant’s folder.”); see also Ellsworth Ex. 4 at 12:3-7 (“Harvard has a holistic admissions
policy where we look at the variety of attributes of every applicant, life experience, education,
activities, the whole range of who an individual is.”).
14

SFFA accuses Harvard (at 41) of “label[ing] every applicant by race.” That assertion is
misleading at best. Harvard applicants have the option, on standard application forms in wide
use in college admissions and not generated by Harvard, to disclose information about their race
or ethnicity. Ellsworth Ex. 98 at 145:11-147:12, 149:18-150:23, 152:19-153:2; Ellsworth Ex. 26
at 145:19-20, 240:3-11; Ellsworth Ex. 49 at HARV00003562; Ellsworth Ex. 50 at
HARV00003557; Ellsworth Ex. 95. But thousands of applicants choose not to disclose their race
each year (Ellsworth Ex. 33 ¶ 111), and when they do not, Harvard does not take any steps to
determine the applicants’ race or ethnicity (Ellsworth Ex. 26 at 240:3-11; Ellsworth Ex. 9 at
190:10-14). When applicants do disclose information about their race or ethnicity, Harvard does
- 37 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 42 of 51

Moreover, all applicants are compared against all others. Ellsworth Ex. 120 at 297:13-22;
Ellsworth Ex. 102 at 187:10-12; 236:19-23; Ellsworth Ex. 98 at 231:14-232:3; Ellsworth Ex. 16
at 173:2-9. There are no racial quotas, and race does not have any automatic or mechanical
effect—unlike in cases where the Supreme Court has invalidated a university’s consideration of
race. Cf. Gratz v. Bollinger, 539 U.S. 244, 271 (2003) (University of Michigan’s
unconstitutional system “automatically distributes 20 points to every single applicant from an
‘underrepresented minority’ group”); Bakke, at 319-320 (the University of California at Davis
Medical School’s unconstitutional system set aside “a specific percentage of the seats in an
entering class” for applicants of certain racial groups). 15
The statistics also rebut SFFA’s argument. As explained in Harvard’s summary
judgment brief, Dr. Card found that applicants must have multiple areas of strength to be
admitted to Harvard. Dr. Card also showed that numerous applicant characteristics—including
the applicant’s academic rating, extracurricular rating, personal rating, teacher ratings, alumni
interviewer ratings, intended concentration, and intended career—explain more about admissions
decisions than race. Ellsworth Ex. 33 ¶ 178 & Card Ex. 27; Ellsworth Ex. 37 ¶¶ 65, 140 & Card
Ex. 22. SFFA fails to address that evidence in its brief, preferring instead to present an
inaccurate and caricatured version of Harvard’s admissions process.
Dr. Card also demonstrated that, even for the most competitive applicants, race no more
influences outcomes than do many other factors, such as a top academic or extracurricular rating.

not attempt to check or verify it (Ellsworth Ex. 98 at 148:20-150:22; Ellsworth Ex. 6 at 114:9115:18; Ellsworth Ex. 9 at 190:10-14), simply considering an applicant’s self-identified race as
one factor among many.
15

SFFA contends (at 41) that Harvard intends to consider race indefinitely, but that is
patently incorrect. The report of the Committee to Study Race-Neutral Alternatives in Harvard
College Admissions specifically recommended reevaluating the need to consider race in five
years. Ellsworth Ex. 47 at HARV00097328.

- 38 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 43 of 51

Ellsworth Ex. 37 ¶¶ 141-146 & Card Ex. 23-24. Indeed, SFFA’s own expert conceded that many
factors besides race may be “determinative” for competitive applicants. Ellsworth Ex. 27 at
279:21-280:4. 16
Accordingly, SFFA is not entitled to summary judgment on Count III. If anything, the
documentary, testimonial, and statistical evidence demonstrates that a reasonable factfinder
could find only for Harvard on that count.
IV.

HARVARD COULD NOT ACHIEVE ITS EDUCATIONAL OBJECTIVES WITHOUT
CONSIDERING RACE IN THE ADMISSIONS PROCESS
SFFA finally accuses Harvard of not sufficiently considering whether race-neutral

alternatives would permit Harvard to achieve its educational objectives. SFFA does not
seriously contend that Harvard could meet its educational goals without considering race; indeed,
it barely discusses any actual race-neutral alternatives. Instead, SFFA makes a variety of
formalistic or inaccurate complaints about the process that Harvard has undertaken to evaluate
and implement race-neutral means to achieve diversity. None of those arguments suggests that
Harvard could achieve its educational objectives without considering race.
SFFA first complains that Harvard did not consider race-neutral measures “before” it
considered race. SFFA Mem. 42. In fact, Harvard has for decades engaged in a variety of race-

16

SFFA stresses (at 41) that if race were not considered in the admissions process, the
proportion of African-American and Hispanic students in the admitted class would drop
significantly. But in a highly competitive process like Harvard’s, where applicants must have
multiple areas of strength to be admitted, any factor—even when it is just one among many—
may be enough to tip the applicant from being denied admission to being admitted. The fact that
race has a marginal effect of that nature, for many otherwise very strong African-American and
Hispanic applicants, does not mean that race plays a role beyond that approved by the Supreme
Court. See Grutter, 539 U.S. at 339 (noting that race may be “outcome determinative” in “the
Harvard plan discussed approvingly by Justice Powell in Bakke, and indeed [in] any plan that
uses race as one of many factors”).

- 39 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 44 of 51

neutral efforts—including several that SFFA’s own expert suggests—to achieve a class that is
diverse along many dimensions, including race.
As explained in Harvard’s summary judgment brief (at 26-27) and the report of the
Committee to Study Race-Neutral Alternatives in Harvard College Admissions (Ellsworth Ex. 47
at HARV00097313), Harvard devotes significant resources to recruitment and outreach, seeking
to encourage candidates from a variety of backgrounds to apply to Harvard and matriculate if
admitted. Harvard has long been a national leader in financial aid, offering one of the most
generous need-based financial aid policies of any university. Id. at HARV00097314. Once
applications are received, admissions officers pay close attention to applicants’ socioeconomic
backgrounds; they ensure that no applicant is “disadvantaged in the application process because
of [his or her] lack of resources and opportunities,” and they seek “to recognize the particular
achievement of students who have excelled when coming from a modest background.” Id.; see
also Ellsworth Ex. 57 at HARV00015424. Harvard also eliminated its Early Action program for
the Classes of 2012 through 2015 with the expectation that the change would cause an even more
diverse group of candidates to apply and matriculate; it reinstated the program when it
determined that the policy change had actually hindered Harvard’s efforts to achieve racial
diversity and other goals. Ellsworth Ex. 47 at HARV00097314-97315, HARV00097324;
Ellsworth Ex. 64 at HARV00030303-HARV00030332.
Fisher I did not adopt a process-based theory of liability. The Court stated that a
university has the “ultimate burden of demonstrating, before turning to racial classifications, that
available, workable race-neutral alternatives do not suffice.” Fisher v. Univ. of Tex. at Austin
(Fisher I), 570 U.S. 297, 312 (2013). The Court nowhere suggested the remarkable proposition
SFFA is advancing: that every university that was then considering race in its holistic admissions

- 40 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 45 of 51

process was suddenly violating the law if it had not previously conducted a formal analysis of
race-neutral alternatives. Instead, the Court held that if the legality of a university’s raceconscious admissions practices were challenged, the university would have to demonstrate to the
court that its consideration of race was necessary to achieve its educational objectives. Id.
(“Narrow tailoring also requires that the reviewing court verify that it is ‘necessary’ for a
university to use race to achieve the educational benefits of diversity. This involves a careful
judicial inquiry into whether a university could achieve sufficient diversity without using racial
classifications.” (citation omitted)).
SFFA’s argument that Harvard must have considered race-neutral alternatives “before”
considering race also ignores that SFFA seeks only forward-looking injunctive relief in this case.
Dkt. 1 at 119. As discussed in more detail below and in Harvard’s summary judgment brief (at
25-34), there is no doubt that Harvard has undertaken a thorough review of potential race-neutral
alternatives and concluded that it could not achieve its educational objectives without
considering race. Even under SFFA’s procedural theory, therefore, Harvard has done exactly
what is required, and SFFA’s argument therefore provides no basis for any injunctive relief
against Harvard. See United States v. Oregon State Med. Soc., 343 U.S. 326, 333 (1952) (“The
sole function of an action for injunction is to forestall future violations.”); Steir v. Girl Scouts of
the USA, 383 F.3d 7, 16 (1st Cir. 2004) (“prospect of future harm” is “the essential prerequisite
for equitable relief”).
SFFA also criticizes (at 42-44) the process by which Harvard considered race-neutral
alternatives in the wake of the Supreme Court’s decisions in Fisher I and Fisher II. There is no
dispute that in 2014—shortly after the Supreme Court decided Fisher I and before SFFA filed
this lawsuit—Harvard formed a university-wide committee, chaired by then-Dean of the

- 41 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 46 of 51

Graduate School of Education James Ryan, which was charged with examining the educational
benefits of diversity and whether Harvard’s various schools could obtain those benefits without
considering race. Ellsworth Ex. 47 at HARV00097311; Ellsworth Ex. 42 at HARV0007236572368. It is hardly surprising that the Committee paused its work after SFFA filed this lawsuit,
which makes allegations with respect to Harvard College regarding the very issues that the
Committee was charged with considering. Harvard recognized that—in this unique situation—
this litigation would produce information and analysis that would inform the evaluation of many
of the questions that the Committee was considering. Ellsworth Ex. 47 at HARV00097311.
The Committee’s work with respect to Harvard College was divided between two new
committees. First, in the spring of 2015, the Committee to Study the Importance of Student
Body Diversity, chaired by Dean Khurana, considered the importance of a diverse student body
to Harvard’s educational goals. Ellsworth Ex. 45 at HARV00008048, HARV00008069;
Ellsworth Ex. 104 at 163:19-164:10. Second, in 2017—so as to have the benefit of the expert
reports in this case—Harvard convened a Committee to Study Race-Neutral Alternatives in
Harvard College Admissions, chaired by Dean of the Faculty of Arts and Sciences Michael
Smith. Ellsworth Ex. 47 at HARV00097310-97311.
SFFA apparently has no quarrel (nor could it) with the process and findings of the
Committee to Study the Importance of Student Body Diversity. The Committee’s findings were
clear and were unanimously endorsed by Harvard’s Faculty of Arts and Sciences (Ellsworth Ex.
45 at HARV00008069; Ellsworth Ex. 71), and SFFA has not challenged them.
Instead, SFFA focuses on the Committee to Study Race-Neutral Alternatives and accuses
Harvard of trying to “create a record … after the fact.” SFFA Mem. 43. But the course that
Harvard chose enabled it to consider and address the specific race-neutral alternatives advanced

- 42 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 47 of 51

by SFFA and the evaluations of those alternatives in the parties’ expert reports. Nothing in the
Supreme Court’s jurisprudence suggests that a university is required to proceed hastily and
without valuable information to meet some nonexistent deadline.
SFFA complains (at 43-44) that the Committee to Study Race-Neutral Alternatives had
too few members, did not “t[ake] testimony” or “conduct[] interviews,” and did not determine
whether “preferences were helping Harvard students succeed academically” (even though, as the
committee members well understood, Harvard College has an extremely high graduation rate,
with 97 percent of students graduating (Ellsworth Ex. 126 at HARV00003906; Ellsworth Ex.
100 at 293:24)). SFFA cites no authority to suggest that Harvard was required to do any of those
things; no decision suggests that a university’s consideration of race-neutral alternatives must
resemble a legislative hearing. Harvard sensibly entrusted the decision to an expert committee
composed of three of its most senior deans, each responsible for relevant aspects of admissions
and undergraduate education. Those committee members spent many hours reviewing academic
literature, studying the complaint and SFFA’s and Harvard’s expert reports, and discussing the
impact of existing and potential race-neutral measures at seven meetings over ten months.
Ellsworth Ex. 47 at HARV00097312. Since SFFA had already initiated this lawsuit and the
meetings concerned compliance with a legal obligation imposed by the Supreme Court’s Fisher
decisions, it is unsurprising and unremarkable that counsel were involved in the process.
SFFA seeks (at 44) to minimize the effect of ending race-conscious admissions by
ignoring the Committee’s full analysis. For example, SFFA singles out the Committee’s
conclusion that ending the consideration of race would cause a decline of more than 50% in the
proportion of African-American students—suggesting that would be the only effect of such a
change. That effect by itself would be quite harmful to Harvard’s educational objectives. Yet

- 43 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 48 of 51

SFFA ignores the Committee’s further conclusion that ending the consideration of race in
admissions would cause yet broader declines in diversity—reducing “the population of students
who self-identify as African-American, Hispanic, or ‘Other’ … by nearly 50%.” Ellsworth Ex.
47 at HARV00097317. SFFA also fails to mention the Committee’s conclusion that the
“corresponding increase in students of other races” would “primarily” benefit “White students”
and that “the non-White percentage of the student body would decline substantially.” Id. The
Committee explained that such a decline in diversity “would prevent Harvard from achieving its
diversity-related educational objectives,” id.—a conclusion SFFA simply ignores.
The Committee also concluded that the proportion of admitted students with the highest
academic ratings would decline by 19% if Harvard gave even greater weight to socioeconomic
circumstances and abandoned consideration of race. Ellsworth Ex. 47 at HARV00097323.
SFFA argues that dramatic drop in academic excellence cannot justify the Committee’s
determination that Harvard’s educational objectives would suffer if it could not consider race.
But the Supreme Court has explained that the law “does not force universities to choose between
a diverse student body and a reputation for academic excellence,” and SFFA offers no actual
reason to doubt—let alone reject—Harvard’s conclusion that this decline is inimical to the
College’s pursuit of academic excellence. Fisher II, 136 S. Ct. at 2213; see Grutter, 539 U.S. at
339. SFFA also ignores the Committee’s determination that “the proportion of students given
the highest extracurricular, personal, and athletic ratings … would also decline substantially” if
Harvard gave even greater weight to socioeconomic circumstances. Ellsworth Ex. 47 at
HARV00097323. Based on that determination, the Committee concluded that the change SFFA
advocates would have a substantial adverse impact on Harvard’s fundamental educational
objectives. SFFA points to nothing to suggest that the Committee’s conclusion was erroneous.

- 44 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 49 of 51

SFFA’s blithe suggestion (at 45) that Harvard should further increase financial aid is
meritless. Harvard already devotes $200 million annually to one of the nation’s leading financial
aid programs, which it publicizes widely. Ellsworth Ex. 47 at HARV00097319. The Committee
also recognized that a strong majority of African-American and Hispanic households in America
already qualifies for financial aid under that program. Id. SFFA ignores the Committee’s wellreasoned conclusion that further increases in financial aid would not materially increase
diversity. Instead, SFFA points to Harvard’s endowment and demands (at 45) that Harvard
spend more—ignoring both the sacrifices that such spending would require of other Harvard
programs and the question whether it would in fact increase diversity. SFFA tellingly does not
identify any data or analysis—let alone data or analysis specific to Harvard—to suggest that the
increase in financial aid it demands would have any material effect on diversity. Like much else
in SFFA’s brief, this point consists of speculation and conclusory assertions of what SFFA
would like the record to look like and the choices SFFA wishes Harvard would make. But SFFA
must support its case with facts and evidence, not grandstanding and bluster.
SFFA does not even bother to discuss the many other race-neutral alternatives considered
by the Committee. See generally Ellsworth Ex. 47 at HARV00097315-97327. The lack of
substance to SFFA’s arguments about race-neutral alternatives demonstrates why Harvard is
entitled to summary judgment on this count or, at the very least, why it is an issue for trial.
CONCLUSION
The Court should deny SFFA’s motion for summary judgment and, for the reasons stated
in Harvard’s memorandum in support of its summary judgment motion, the Court should grant
summary judgment to Harvard on all remaining counts of the Complaint.

- 45 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 50 of 51

Respectfully submitted,
/s/ Seth P. Waxman
Seth P. Waxman (pro hac vice)
Paul R.Q. Wolfson (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]
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]
William F. Lee (BBO #291960)
Felicia H. Ellsworth (BBO #665232)
Andrew S. Dulberg (BBO #675405)
Elizabeth Mooney (BBO #679522)
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]
Dated: July 27, 2018

Counsel for Defendant President and
Fellows of Harvard College

- 46 -

Case 1:14-cv-14176-ADB Document 435 Filed 07/27/18 Page 51 of 51

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, and
that the sealed version of this document will be served on counsel for SFFA by email.
/s/ Seth P. Waxman
Seth P. Waxman

- 47 -