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

Filing 627

Proposed Findings of Fact by Students for Fair Admissions, Inc.. (Consovoy, William)

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 1 of 85

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

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

PRESIDENT AND FELLOWS OF HARVARD
COLLEGE,
Defendant.
PLAINTIFF’S RESPONSE TO DEFENDANT’S PROPOSED
FINDINGS OF FACT AND CONCLUSIONS OF LAW
Adam K. Mortara
Scott J. McBride
Krista J. Perry
BARTLIT BECK LLP
54 West Hubbard Street, Suite 300
Chicago, IL 60654
312.494.4400
[e-mail]
[e-mail]
[e-mail]
John M. Hughes
Katherine L.I. Hacker
Meg E. Fasulo
BARTLIT BECK LLP
1801 Wewatta Street, Suite 1200
Denver, CO 80202
303.592.3100
[e-mail]
[e-mail]
[e-mail]
Paul M. Sanford BBO #566318
BURNS & LEVINSON LLP
One Citizens Plaza, Suite 1100
Providence, RI 02903
617.345.3000
[e-mail]
January 23, 2019

William S. Consovoy
Thomas R. McCarthy
J. Michael Connolly
Cameron T. Norris
CONSOVOY MCCARTHY PARK PLLC
3033 Wilson Boulevard, Suite 700
Arlington, Virginia 22201
703.243.9423
[e-mail]
[e-mail]
[e-mail]
[e-mail]
Patrick Strawbridge BBO #678274
CONSOVOY MCCARTHY PARK PLLC
Ten Post Office Square
8th Floor South PMB #706
Boston, MA 02109
617.227.0548
[e-mail]
Michael H. Park
CONSOVOY MCCARTHY PARK PLLC
745 Fifth Avenue, Suite 500
New York, NY 10151
212.247.8006
[e-mail]

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 2 of 85

TABLE OF CONTENTS
INTRODUCTION ........................................................................................................................................... 1
RESPONSE TO HARVARD’S PROPOSED FINDINGS OF FACTS ................................................ 4
RESPONSE TO HARVARD’S PROPOSED CONCLUSIONS OF LAW ........................................ 47
A.

Harvard Intentionally Discriminates Against Asian-American Applicants. ................ 47

B.

Harvard’s Use Of Race Does Not Comply With Precedent. ........................................ 69
i.
ii.

Harvard failed to consider or implement workable race-neutral
alternatives................................................................................................................ 70

iii.

Harvard’s use of race violates Grutter. .................................................................. 74

iv.
C.

Harvard is not entitled to any deference. ............................................................ 69

Harvard engages in racial balancing. .................................................................... 75

SFFA Has Standing. ............................................................................................................. 75

CERTIFICATE OF SERVICE .................................................................................................................... 77

ii

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 3 of 85

TABLE OF AUTHORITIES
Cases
Adams v. Ameritech Servs., Inc.,
231 F.3d 414 (7th Cir. 2000) ................................................................................................................. 53
Adarand Constructors, Inc. v. Pena,
515 U.S. 200 (1995) ................................................................................................................................ 49
Aguilar v. ICE,
510 F.3d 1 (1st Cir. 2007) ...................................................................................................................... 52
Ahern v. Shinseki,
629 F.3d 49 (1st Cir. 2010) .................................................................................................................... 49
Aiken v. City of Memphis,
37 F.3d 1155 (6th Cir. 1994) ................................................................................................................. 72
Ala. Legislative Black Caucus v. Alabama,
231 F. Supp. 3d 1026 (M.D. Ala. 2017) ............................................................................................... 48
Aman v. Cort Furniture Rental Corp.,
85 F.3d 1074 (3d Cir. 1996) ................................................................................................................... 62
Anderson ex rel. Dowd v. City of Boston,
375 F.3d 71 (1st Cir. 2004) .............................................................................................................. 48, 69
Atonio v. Wards Cove Packing Co.,
810 F.2d 1477 (9th Cir. 1987) ............................................................................................................... 56
Bazemore v. Friday,
478 U.S. 385 (1986) ................................................................................................................... 50, 54, 55
Berger v. Iron Workers Reinforced Rodmen Local 201,
843 F.2d 1395 (D.C. Cir. 1988)............................................................................................................. 50
Blunt v. Lower Merion Sch. Dist.,
767 F.3d 247 (3d Cir. 2014) ................................................................................................................... 52
Boykin v. Georgia-Pac. Corp.,
706 F.2d 1384 (5th Cir. 1983) ............................................................................................................... 60
Brown v. Gaston Cty. Dyeing Mach. Co.,
457 F.2d 1377 (4th Cir. 1972) ............................................................................................................... 56
Bryant v. Indep. Sch. Dist. No. I-38 of Garvin Cty.,
334 F.3d 928 (10th Cir. 2003) ............................................................................................................... 52
Burgett v. Kan. City Bd. of Police Comm’rs,
613 F. App’x 553 (8th Cir. 2015) .......................................................................................................... 52
Bush v. Vera,
517 U.S. 952 (1996) ................................................................................................................................ 66
Castaneda v. Partida,
430 U.S. 482 (1977) ................................................................................................................... 50, 51, 60

iii

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 4 of 85

Chadwick v. WellPoint, Inc.,
561 F.3d 38 (1st Cir. 2009) ....................................................................................................... 58, 59, 68
Chaffin v. Kan. State Fair Bd.,
348 F.3d 850 (10th Cir. 2003) ............................................................................................................... 72
Chin v. Runnels,
343 F. Supp. 2d 891 (N.D. Cal. 2004) ................................................................................................. 68
City of Cleburne v. Cleburne Living Ctr.,
473 U.S. 432 (1985) ................................................................................................................................ 48
City of Richmond v. J.A. Croson Co.,
488 U.S. 469 (1989) ................................................................................................................................ 71
Columbus Bd. of Educ. v. Penick,
443 U.S. 449 (1979) ................................................................................................................................ 64
Cooper v. Fed. Reserve Bank of Richmond,
467 U.S. 867 (1984) ................................................................................................................................ 61
Cooper v. Harris,
137 S. Ct. 1455 (2017) ............................................................................................................................ 49
Desert Palace, Inc. v. Costa,
539 U.S. 90 (2003)................................................................................................................................... 54
EEOC v. Gen. Tel. Co. of Nw., Inc.,
885 F.2d 575 (9th Cir. 1989) ................................................................................................................. 55
EEOC v. Tex. Roadhouse, Inc.,
215 F. Supp. 3d 140 (D. Mass. 2016) ................................................................................................... 56
Fed. Hous. Fin. Agency v. Nomura Holding Am., Inc.,
104 F. Supp. 3d 441 (S.D.N.Y. 2015) .................................................................................................. 55
Fisher v. Univ. of Tex. at Austin (Fisher I),
570 U.S. 297 (2013) ........................................................................................................................ passim
Fisher v. Univ. of Tex. at Austin (Fisher II),
136 S. Ct. 2198 (2016) ............................................................................................................... 70, 71, 73
Franchina v. City of Providence,
881 F.3d 32 (1st Cir. 2018) .................................................................................................................... 59
Fudge v. City of Providence Fire Dep’t,
766 F.2d 650 (1st Cir. 1985) .................................................................................................................. 54
Goodman v. Bowdoin Coll.,
380 F.3d 33 (1st Cir. 2004) .............................................................................................................. 48, 65
Goodman v. Lukens Steel Co.,
482 U.S. 656 (1987) ................................................................................................................................ 66
Gratz v. Bollinger,
539 U.S. 244 (2003) ................................................................................................................... 47, 49, 66

iv

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 5 of 85

Griffin v. Carlin,
755 F.2d 1516 (11th Cir. 1985) ............................................................................................................. 56
Grutter v. Bollinger,
539 U.S. 306 (2003) .......................................................................................................................... 70, 75
Hazelwood Sch. Dist. v. United States,
433 U.S. 299 (1977) ................................................................................................................................ 50
Hemmings v. Tidyman’s Inc.,
285 F.3d 1174 (9th Cir. 2002) ............................................................................................................... 55
Hunt v. Wash. State Apple Advert. Comm’n,
432 U.S. 333 (1977) ................................................................................................................................ 61
Hunter v. Underwood,
471 U.S. 222 (1985) ................................................................................................................................ 67
In re Neurontin Mktg. & Sales Practices Litig.,
712 F.3d 21 (1st Cir. 2013) .................................................................................................................... 54
Int’l Bhd. of Teamsters v. United States,
431 U.S. 324 (1977) ........................................................................................................................ passim
J.E.B. v. Ala. ex rel. T.B.,
511 U.S. 127 (1994) ................................................................................................................................ 48
Jana-Rock Const., Inc. v. N.Y. State Dep’t of Econ. Dev.,
438 F.3d 195 (2d Cir. 2006) ...................................................................................................... 48, 49, 50
Johnson v. Bd. of Regents of Univ. of Ga.,
263 F.3d 1234 (11th Cir. 2001) ............................................................................................................. 71
Johnson v. Transp. Agency, Santa Clara Cty.,
480 U.S. 616 (1987) ................................................................................................................................ 66
Jones v. City of Boston,
752 F.3d 38 (1st Cir. 2014) .................................................................................................................... 54
Keller v. Prince George’s Cty.,
827 F.2d 952 (4th Cir. 1987) ................................................................................................................. 66
Kimble v. Wis. Dep’t of Workforce Dev.,
690 F. Supp. 2d 765 (E.D. Wis. 2010) ................................................................................................. 69
Lam v. Univ. of Haw.,
40 F.3d 1551 (9th Cir. 1994) ................................................................................................................. 67
Latinos Unidos De Chelsea En Accion (LUCHA) v. Sec’y of HUD,
799 F.2d 774 (1st Cir. 1986) ............................................................................................................ 50, 61
Lipsett v. Univ. of P.R.,
864 F.2d 881 (1st Cir. 1988) .................................................................................................................. 53
Loving v. Virginia,
388 U.S. 1 (1967) ..................................................................................................................................... 48

v

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 6 of 85

Lowery v. Circuit City Stores, Inc.,
206 F.3d 431 (4th Cir. 2000) ................................................................................................................. 64
Lujan v. Franklin Cty. Bd. of Educ.,
766 F.2d 917 (6th Cir. 1985) ........................................................................................................... 50, 51
Lynn v. Regents of Univ. of Cal.,
656 F.2d 1337 (9th Cir. 1981) ............................................................................................................... 52
Mangold v. Cal. Pub. Utilities Comm’n,
67 F.3d 1470 (9th Cir. 1995) ................................................................................................................. 62
McCleskey v. Kemp,
481 U.S. 279 (1987) .......................................................................................................................... 50, 51
McGuire v. Reilly,
386 F.3d 45 (1st Cir. 2004) .................................................................................................................... 51
McReynolds v. Sodexho Marriott Servs., Inc.,
349 F. Supp. 2d 1 (D.D.C. 2004) .......................................................................................................... 64
Miller v. Johnson,
515 U.S. 900 (1995) .......................................................................................................................... 65, 66
Mister v. Ill. Cent. Gulf R. Co.,
832 F.2d 1427 (7th Cir. 1987) ......................................................................................................... 51, 63
Nieves-Márquez v. Puerto Rico,
353 F.3d 108 (1st Cir. 2003) .................................................................................................................. 52
Palmer v. Shultz,
815 F.2d 84 (D.C. Cir. 1987) ........................................................................................................... 51, 55
Parents Involved in Community Sch. v. Seattle School Dist. No. 1,
551 U.S. 701 (2007) ........................................................................................................................ passim
Parham v. Sw. Bell Tel. Co.,
433 F.2d 421 (8th Cir. 1970) ................................................................................................................. 54
Pegues v. Miss. State Emp’t Serv.,
699 F.2d 760 (5th Cir. 1983) ................................................................................................................. 56
Pers. Adm’r of Mass. v. Feeney,
442 U.S. 256 (1979) ................................................................................................................................ 64
Phillips v. Martin Marietta Corp.,
400 U.S. 542 (1971) ................................................................................................................................ 59
Police Officers for Equal Rights v. City of Columbus,
644 F. Supp. 393 (S.D. Ohio 1985) ...................................................................................................... 64
Resare v. Raytheon Co.,
981 F.2d 32 (1st Cir. 1992) .................................................................................................................... 62
Robinson v. Metro-N. Commuter R.R. Co.,
267 F.3d 147 (2d Cir. 2001) ...................................................................................................... 52, 60, 61

vi

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 7 of 85

Roebuck v. Drexel Univ.,
852 F.2d 715 (3d Cir. 1988) ................................................................................................................... 67
Rosen v. Thornburgh,
928 F.2d 528 (2d Cir. 1991) ................................................................................................................... 62
Santiago-Ramos v. Centennial P.R. Wireless Corp.,
217 F.3d 46 (1st Cir. 2000) .................................................................................................................... 72
Segar v. Smith,
738 F.2d 1249 (D.C. Cir. 1984)..................................................................................................... passim
Shaw v. Reno,
509 U.S. 630 (1993) .......................................................................................................................... 65, 66
Smith v. Barton,
914 F.2d 1330 (9th Cir. 1990) ............................................................................................................... 52
Spath v. NCAA,
728 F.2d 25 (1st Cir. 1984) .................................................................................................................... 51
Stender v. Lucky Stores, Inc.,
803 F. Supp. 259 (N.D. Cal. 1992) ....................................................................................................... 64
Sweeney v. Bd. of Trustees of Keene State Coll.,
604 F.2d 106 (1st Cir. 1979) .................................................................................................................. 57
Thomas v. Eastman Kodak Co.,
183 F.3d 38 (1st Cir. 1999) ....................................................................................................... 62, 65, 66
U.S. Postal Serv. Bd. of Governors v. Aikens,
460 U.S. 711 (1983) ................................................................................................................................ 54
Udo v. Tomes,
54 F.3d 9 (1st Cir. 1995) ........................................................................................................................ 49
United States ex rel. Williams v. City of Brockton,
No. 12-CV-12193-IT, 2016 WL 742917 (D. Mass. Dec. 23, 2016) ................................................ 52
United States v. Cty. of Maricopa,
889 F.3d 648 (9th Cir. 2018) ................................................................................................................. 64
United States v. Fisher,
624 F.3d 713 (5th Cir. 2010) ................................................................................................................. 49
United States v. Mass. Mar. Acad.,
762 F.2d 142 (1st Cir. 1985) .................................................................................................................. 72
United States v. W. T. Grant Co.,
345 U.S. 629 (1953) ................................................................................................................................ 72
United States v. Windsor,
570 U.S. 744 (2013) ................................................................................................................................ 67
Vill. of Arlington Heights v. Metro. Hous. Dev. Corp.,
429 U.S. 252 (1977) ............................................................................................................. 48, 50, 51, 69

vii

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 8 of 85

Washington v. Davis,
426 U.S. 229 (1976) ................................................................................................................................ 51
Watson v. Fort Worth Bank & Tr.,
798 F.2d 791 (5th Cir. 1986) ................................................................................................................. 60
Wayte v. United States,
470 U.S. 598 (1985) ................................................................................................................................ 60
Wessmann v. Gittens,
160 F.3d 790 (1st Cir. 1998) ............................................................................................................ 71, 73
White v. Vathally,
732 F.2d 1037 (1st Cir. 1984) ................................................................................................................ 49
Woods v. City of Greensboro,
855 F.3d 639 (4th Cir. 2017) ........................................................................................................... 58, 65
Other Authorities
13A Fed. Prac. & Proc. Juris. § 3531.9.5 (3d ed.)........................................................................................ 61
Deborah L. Rhode, The Subtle Side of Sexism, 16 Colum. J. Gender & L. 613 (2007) ............................. 58
Fed. Judicial Ctr., Reference Manual on Scientific Evidence (3d ed. 2011) ....................................................... 53
Jerome Karabel, The Chosen: The Hidden History of Admission and Exclusion at Harvard, Yale,
and Princeton (2005) .................................................................................................................................. 58
Joan C. Williams, The Social Psychology of Stereotyping, 7 Emp. Rts. & Emp. Pol’y J. 401 (2003) ............. 58
Lex K. Larson, Employment Discrimination § 40.04 (2d ed. 1996)................................................................ 58

viii

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 9 of 85

INTRODUCTION
At long last, Harvard concedes that being Asian American has a “negative effect” on an
applicant’s “personal rating” and that these “negative associations between Asian-American ethnicity
and the personal rating” significantly undermine their chances of admission. Harvard Findings of Fact
(“FF”) 34, 135-137. The Court therefore must answer a question of great importance to the parties
and Asian Americans across the country: Why? Why does being Asian American systematically lead
to a lower rating of an applicant’s personal qualities? SFFA’s answer is that Harvard engages in racial
stereotyping, a form of intentional discrimination, and that the record contains extensive statistical
and other circumstantial evidence to substantiate that charge. SFFA Findings of Fact (“SFFA-FF”)
21-133. Harvard’s answer (to the extent it even offers one) is that Asian-American applicants simply
deserve it—they deserve lower personal ratings. According to Harvard, Asian-Americans applicants—
as a group, year after year—compare unfavorably to their African-American, Hispanic, and white
peers in “integrity, helpfulness, courage, kindness, reaction to setbacks, concern for others, selfconfidence, leadership abilities, and maturity.” FF 46.
The Court thus faces an unenviable task. It must find that there is something defective about
the character of Asian-American applicants or something defective with Harvard’s admissions
process. Harvard asks the Court to enshrine in law old and offensive stereotypes that Asian-American
students are timid, quiet, shy, passive, withdrawn, one dimensional, hard workers, perpetual foreigners,
and so-called “model minorities.” SFFA-FF 94. SFFA, on the other hand, asks the Court to tell
Harvard the truth based on the evidence. Either Harvard discriminates against Asian Americans or
the stereotypes are true. The Court must choose.
This is not a hard choice. To state the disreputable proposition upon which Harvard’s legal
defense depends is to refute it. Moreover, the experts agree that if the personal rating (like the overall
rating) is influenced by race—let alone reflects racial stereotyping of Asian-American applicants—

1

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 10 of 85

Harvard has imposed a statistically significant penalty on Asian Americans. Unless the Court is willing
to find that the clear racial hierarchy of the personal rating is just a coincidence, SFFA must prevail.
It is unimaginable that the Court would hold, under any legal framework, that there is a statistically
significant penalty on Asian Americans yet no violation of Title VI. No more than the Court would
look at an analogous record in a Title VII sex-discrimination case and find that an employer had not
engaged in unlawful discrimination against women.
Indeed, to find for Harvard, the Court would have to disregard the history of discrimination
that Asian Americans have suffered, the stereotypes that continue to plague them, and the obvious
toxicity of a system that combines racial preferences with a subjective assessment of an applicant’s
personality. It would have to reject the Department of Education’s finding that Harvard uses race in
the personal rating and racially stereotypes Asian-American applicants. It would have to ignore the
Office of Institutional Research’s finding that Harvard penalizes Asian Americans, that the penalty
cannot be attributed to legacy and athlete preferences, and that Dean Fitzsimmons did not find this
conclusion troubling. It would have to believe that the glimpses of racial stereotyping that Harvard’s
recruiting policies, written comments, and testimony reveal are not symptomatic of a more profound
malady. It would have to ignore the changes that Harvard made to its Reading Procedures on the eve
of trial because of racial-stereotyping concerns. It would have to credit—in derogation of half a
century of civil-rights law—Harvard’s prayer against statistical analysis and circumstantial evidence,
accepting that SFFA cannot win without a “smoking gun.” It would have to give decisive weight to
Harvard’s cherry-picked anecdotal evidence. And the Court would have to assume Harvard’s good
faith notwithstanding all evidence to the contrary. None of this is remotely tenable. SFFA is entitled
to judgment on Count I.
SFFA is entitled to judgment on its other counts too. Harvard claims that it is not engaged in
racial balancing. Yet the admissions numbers barely change, Fitzsimmons obsesses over his “ethnic

2

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 11 of 85

stats” throughout the process, Harvard closely monitors the racial balance year over year, and Harvard
admits that it would exceed its bed count if the racial balance ever deviated. Harvard claims that it
cares about diversity in all of its dimensions. Yet it myopically obsesses over racial diversity while
paying virtually no attention to religious, geographic, and socioeconomic diversity. Harvard claims
that, for underrepresented minorities, race is merely a “plus” factor. But Harvard concedes that the
use of race increases the chance of admission for Hispanics and African Americans by 141% and
324%, respectively. Harvard claims that its use of race has a logical stopping point. Yet Harvard—
contrary to the many assurances it gave the Supreme Court over the last forty-plus years—never
examined race-neutral alternatives until SFFA filed suit, has not complied with the requirement to set
a “sunset” date, and whispers what its amici have the courage to say aloud: that race should always be
considered because it is an essential part of an applicant’s identity. Last, Harvard claims that it lacks
workable race-neutral alternatives. Yet Harvard can achieve student body diversity by ending legacy,
Dean’s and Director’s List, and children of staff and faculty preferences, and by increasing the
preference for socioeconomically disadvantaged applicants to only half the size of the preference given
to recruited athletes.
Harvard’s position, ultimately, is a mass of contradictions. SFFA overemphasizes the role of
academics—but race-neutral alternatives are unworkable because they purportedly would decrease
“the proportion of admitted students with the highest academic ratings.” FF 250. Harvard does not
engage in racial balancing and considers each applicant as a “whole person”—but it knows the exact
percentage that African-American and Hispanic enrollment will decrease absent racial preferences.
Harvard Conclusions of Law (“CL”) 316. Harvard does not “set a target for the racial composition of
the class”—but any diminution (however slight) in the number of African Americans admitted is
unacceptable even if Harvard’s overall racial diversity would increase. FF 211. Race is only a “plus”
factor—but the African-American share of the class would plummet from 14% to 6% without using

3

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 12 of 85

race. CL 323. And race-neutral alternatives (that would keep the African-American share of the class
at 10%) must be rejected in the name of “diversity”—even though implementing them would make
the college less white, less wealthy, and more geographically diverse. There is no way that these
pretextual arguments can withstand review under strict scrutiny.
That is why Harvard resorts—as it has from the beginning—to invective against SFFA and
anyone else who has the temerity to believe that the “way to stop discrimination on the basis of race
is to stop discriminating on the basis of race.” Parents Involved in Community Sch. v. Seattle School Dist. No.
1, 551 U.S. 701, 748 (2007) (plurality). Given how one-sided the evidence is, Harvard’s name-calling
is understandable. But the not-so-subtle suggestion from Harvard and its amici that this Court should
sweep Harvard’s illegal behavior under the rug to save Bakke and Grutter is both illogical and immoral.
Virtue signaling is not a defense, and Harvard’s with-us-or-against-us reductionism is unpersuasive.
Decent people can in good conscience disagree with SFFA’s broadest aims—and even enthusiastically
support existing Supreme Court precedent—but understand that this honest difference of opinion
does not excuse intentional discrimination against Asian Americans or Harvard’s failure to otherwise
use race in a narrowly tailored way. It would be perverse indeed if swearing allegiance to precedent
could save Harvard from the consequences of its flagrant noncompliance with those same decisions.
SFFA, its 20,000-plus members, and the 269 Asian-American organizations that support it are entitled
to a fair resolution of these claims whether or not their ultimate legal and policy views are popular in
certain circles. The Court should grant judgment on liability to SFFA.
RESPONSE TO HARVARD’S PROPOSED FINDINGS OF FACTS
1-2. FF 1-2 are incomplete and misleading. Harvard myopically focuses on race and does not
pursue diversity in all of its dimensions. SFFA-FF 142-144.
3-5. FF 3-5 are inaccurate, incomplete and misleading. Harvard does not value religious,
socioeconomic, and geographic diversity the way that it values racial diversity. SFFA-FF 142-144.

4

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 13 of 85

Harvard could be more diverse if it focused on socioeconomic diversity and abandoned preferences
for legacies, the children of donors, and the children of Harvard faculty and staff (“LDC”)—who are
overwhelmingly white and wealthy. SFFA-FF 158. By “racial diversity,” Harvard means the mere fact
that a particular racial box is checked, since it gives African Americans and Hispanics a “tip” regardless
of whether their race is important to them or has affected them in some specific way. SFFA-FF 12.
6-9. FF 6-9 are incomplete and misleading. Harvard convened a committee “to study the
importance of student body diversity” in 2015—after the launch of “Harvard Not Fair” in April 2014
and after this suit was filed in November 2014. T7.34:20-22; P340; Doc. 1 (complaint).
10-13. FF 10-13 are inaccurate, incomplete, and misleading. SFFA introduced substantial
evidence that Harvard does not have a “long-held and fundamental institutional interest” in obtaining
all forms of diversity, including religious, socioeconomic, and geographic diversity. Supra ¶¶ 1-5.
14-16. FF 14-16 are incomplete and misleading. Harvard convened a committee to study
“inclusion and belonging” in May 2014—after the launch of “Harvard Not Fair.” And the committee
released its report in November 2015—after this suit was filed. D13.1, 9; P340; Doc. 1 (complaint).
17-18. FF 17-18 are incomplete and misleading. Harvard convened a “University-wide Task
Force on Inclusion and Belonging” in May 2016 and issued its findings in March 2018—after the
launch of “Harvard Not Fair” and this suit was filed. D740.1-2; T14.211:6-9; P340; Doc. 1 (complaint).
19. FF 19 is inaccurate. Supra ¶¶ 1-5.
20-22. FF 20-22 are inaccurate, incomplete, and misleading. Harvard’s recruiting is not equally
“directed at students of all demographic and socioeconomic backgrounds.” Asian Americans must
score higher on standardized tests than African Americans, Hispanics, and whites (in Sparse Country).
SFFA-FF 15, 105-106. These efforts do have an “effect on admissions decisions.” They allow Harvard
to “consciously shape the makeup of [its] student body.” T1.129:5-22, 142:24-143:13. Recruited
students are “about twice as likely to be admitted as other applicants.” T1.132:7-17.

5

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 14 of 85

23-24. FF 23-24 are incomplete and misleading. Harvard does a poor job of recruiting,
admitting, and enrolling students from modest socioeconomic backgrounds and first-generation
college students. SFFA-FF 143, 157-158; T6.47:19-48.6.
25-27. FF 25-27 are inaccurate, incomplete, and misleading. The cited evidence does not
support the assertion that Harvard’s financial aid is “among the most generous in the country.” There
is no financial barrier to admitting more non-wealthy students. Doc. 597-1, 46:5-47:4; T6.56-7-58:25.
Harvard is “literally the richest university in the entire country,” with a $37 billion endowment that is
“bigger than the gross domestic product of half the world’s countries.” T6.56:7-58:25.
28-29. FF 28-29 are inaccurate. Harvard ignores attributes that would serve its stated
“educational objectives.” Supra ¶¶ 1-5. And Harvard admits certain applicants for reasons unrelated
to “academic excellence, extracurricular distinction, and personal qualities,” D5.10, or the potential to
“significantly contribute to and benefit from the educational experience on campus.” FF 29. Harvard
gives tips to LDC applicants because of their family connections—not because of what they have
accomplished or may contribute to the Harvard community. SFFA-FF 14; infra ¶ 223.
30-31. FF 30-31 are inaccurate, incomplete, and misleading. Harvard does not employ “wholeperson” review that “considers the full range of available information.” Harvard blinds itself to an
applicant’s religion, often crucial information about a how a person will contribute to diversity. SFFAFF 142. Harvard also gives a “tip” to the children of donors. SFFA-FF 14, 20. Harvard could increase
it socioeconomic diversity by ending LDC preferences. SFFA-FF 157-158.
32. FF 32 is accurate.
33-34. FF 33-34 are inaccurate, incomplete, and misleading. Academic qualifications are a key
to admission; 82% of admits get an academic rating of 1 or 2, SFFA-FF 25, and applicants with an
academic 1—but a 3 or worse on all other ratings—still have a 40% chance of admission. D672. By

6

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 15 of 85

contrast, applicants with an extracurricular 1 or a personal 1 (but a 3 or worse on all other ratings)
have only a 7% and 0% chance of admission, respectively. DX672; P316.14; P316.12; T7.97:1-24.
35-36. FF 35-36 are accurate.
37. FF 37 is inaccurate. Harvard penalizes Asian Americans, SFFA-FF 21-133, engages in
racial balancing, SFFA-FF 134-140, and awards large racial preferences to African Americans and
Hispanics, SFFA-FF 141-152.
38. FF 38-39 are incomplete and misleading. The Common Application and Universal College
Application also may include information about the applicant’s religion, which Harvard does not
include in the applicant’s file. SFFA-FF 142.
39-40. FF 39-40 are inaccurate, incomplete, and misleading. Harvard’s ratings are not “often”
assigned before the application file is complete. T6.185:18-186:1. And the ratings are key to admission.
SFFA-FF 25-26, 29; PD1, PD2, PD4, PD8, PD11, PD38.10, P620, P621, P622, P623.
41. FF 41 is inaccurate, incomplete, and misleading. The cited testimony is that applicants are
not “admitted or rejected based solely on the profile ratings,” which Harvard distorts. FF41. Harvard’s
ratings strongly correlate with admission. Supra ¶¶ 39-40; DD10.6, DD10.8, T13.33:16-24; 102:11-20.
That is why applicants with overall ratings of 3 and below typically are not deemed “competitive,”
P1.14, and are not discussed in committee meetings, P1.14; T4.9:20-10:7. Harvard concedes that an
applicant’s ratings are a “handle” for the committee discussion and the decision made about each
applicant. T8.156:1-6; P555.18; SFFA-FF 17-19.
42. FF 42 is accurate.
43-45. FF 43-45 are incomplete and misleading. The academic and extracurricular ratings are
not, as Harvard implies, mostly subjective. They are almost entirely based on objective criteria. SFFAFF 5-6; T2.12:21-13:2. Outside of recruited athletes (who receive a rating of 1), the athletic rating is
not strongly correlated with admission. SFFA-FF 7, 28; T2.12:21-13:2. Indeed, applicants whose two

7

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 16 of 85

“strengths” are in the academic, extracurricular, or personal rating have a 62% higher chance of
admission than applicants whose two strengths include the athletic rating; similarly, applicants whose
three strengths are in the academic, extracurricular, and personal rating have a 55% higher chance of
admission than applicants whose three strengths include the athletic rating. DD10.11.
46. FF 46 is inaccurate, incomplete, and misleading. Harvard quotes the Class of 2023 Reading
Procedures, which were adopted in the fall of 2018 and changed the way admissions officers assign
the personal rating because of concerns that the lack of substantive guidance was leading to the racial
stereotyping of Asian-American applicants. SFFA-FF 113-114. The Reading Procedures for all other
years—dating back to at least the 1990 OCR investigation—provide no substantive guidance on how
to assign the rating, aside from general descriptions like “very strong,” “generally positive,” and
“bland.” SFFA-FF 8, 97-99, 113-114, e.g., D744.4; P1.6; P555.15. Admissions officers assign the rating
using highly subjective measures of their own choosing, some of which FF 46 identifies. Id. The
personal rating also measures whether someone is “likeable” and “a good person.” SFFA-FF 99.
Admissions officers then “sort of add it all up and get a feeling.” Id.
47. FF 47 is misleading and incomplete. The school-support ratings are not assigned by
teachers and guidance counselors; they are Harvard’s interpretations of the level of school support.
SFFA-FF 9; T2.12:21-13:2; T4.204:4-23. Professor Arcidiacono found that race influences Harvard’s
assignment of school-support ratings and penalizes Asian-American applicants. Professor Card said
nothing in response to this finding. SFFA-FF 62.
48. FF 48 is incomplete and misleading. The overall rating “reflects the reader’s judgment as
to the applicant’s likelihood of admission, based on a combination of the other ratings along with the
reader’s sense of how strong or weak the applicant is.” P555.13; T4.205:21-206:2; SFFA-FF 10;
T2.12:21-13:2. The overall rating is strongly correlated with admission. SFFA-FF 29.

8

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 17 of 85

49-50. FF 49-50 are incomplete and misleading. Readers may include written comments in
the application file, but they are careful about what they say. Harvard warns them: “REMEMBER:
your comments may be open to public view at a later time.” P1.13; T4.165:20-166:17.
51. FF 51 is inaccurate. Subcommittees do not “examine the entire docket” several times.
They examine those applicants deemed to be “legitimate contenders.” T4.10:2-7.
52-54. FF 52-54 are incomplete and misleading. The Admissions Office leadership, especially
Dean Fitzsimmons and Director McGrath, wield enormous influence over who is admitted. If the
committee process is nearing the end and the share of a certain racial group is “surprisingly or notably
underrepresented,” they will direct the committee to “go back and look at those cases,” T5.198:24199:13, to “make sure that we’re not having a dramatic drop-off in some group who we did at a certain
level with last year,” T5.200:22-202:3. Applicants that Fitzsimmons and McGrath prefer are far more
likely to be admitted than others. SFFA-FF 69-71.
55. FF 55 is incomplete and misleading. The profile ratings and overall rating are crucial to
admission to Harvard. Supra ¶ 41.
56-57. FF 56-57 are accurate.
58. FF 58 is incomplete and misleading. See SFFA-FF 12-13.
59-60. FF 59-60 are inaccurate. Harvard always knows and “considers” an applicant’s race to
determine whether a “tip” is warranted. SFFA-FF 12; T5.192:13-193:14; T10.106:25-107:9; P316.1;
e.g., D24.13. Only underrepresented minorities receive a “tip” based on the race they check on their
application. Supra ¶ 58. For African-American and Hispanic applicants, race is not just one “factor
that contributes to the admission of an applicant”; it dramatically increases their chances of admission.
Harvard also uses race to penalize Asian Americans. Supra ¶ 37.
61. FF 61 is inaccurate. The cited evidence does not support the assertion that Harvard gives
a racial “tip” only if the applicant “would be highly competitive no matter what their race.” For many

9

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 18 of 85

African Americans and Hispanics, that “tip” is what makes them competitive. SFFA-FF 145-150.
Professor Arcidiacono’s analysis proves the point. In the fourth through seventh academic deciles—
the middle of the pack in terms of academic qualifications—African Americans are at least ten times
more likely to be admitted than Asian Americans. PD38.21. Indeed, African Americans in the bottom
40% of the pool academically have the same chance of admission as Asian Americans in the top decile,
and are more than 14 times more likely to be admitted than Asian Americans in the same decile.
PD38.21. Similarly, Hispanic applicants in the fifth decile have a 9.1% chance of admission—on par
with Asian-American applicants in the second decile, i.e., near the top academically; African-American
applicants in the fifth decile have a stunningly high 22.4% admission rate, which is twice as high as
Asian-Americans in the top decile. PD38.21.
62. FF 62 is inaccurate. Race influences the overall rating, the personal rating, and the schoolsupport ratings. SFFA-FF 44-62.
63-65. FF 63-65 are inaccurate. Harvard’s use of race is not “flexible and non-mechanistic.”
It awards large racial preferences to African Americans and Hispanics, penalizes Asian Americans,
and engages in racial balancing. Supra ¶ 37. Harvard does consider “the fact of an applicant’s race” in
assigning the personal rating. SFFA-FF 29-34, 49-58, 95-96, 98, 101-102.
66-68. FF 66-68 are misleading and incomplete. Harvard’s training on how to use race is
deficient. SFFA-FF 97-98. The casebook is just “lightly edited application files,” FF 66, and provides
no instruction on how to use race, D2. The guide to the casebook contains almost no instruction on
the use of race. T5.258:20-259:3, e.g., D24.7, 13. To the extent that Harvard trains admissions officers
on the use of race via “conversation,” it was unmemorable to most admissions officers. SFFA-FF 100.
69. FF 69 is inaccurate. Harvard’s “diversity training” examined “testing data and percentile
breakdowns by race” so officers had “context” about how applicants’ test scores compared intraracially. P81. Officers also were informed that Asian Americans are “overrepresented” in terms of

10

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 19 of 85

how they perform on the SAT relative to their population. P99.2; D19.10. Although the training
discussed the discrimination that African Americans face, the discrimination that Asian Americans
face was not discussed. Compare D19.50, with D19.28; SFFA-FF 94-95.
70. FF 70 is accurate.
71. FF 71 is incomplete and misleading. Although new Reading Procedures are issued each
year, they rarely changed in form or substance until this case was filed. Compare, e.g., P1 with P71. Until
then, the instructions for assigning the personal rating had remained unchanged for decades—at least
as far back as 1990. Compare, e.g., P1.6 with P555.15. It was not until the fall of 2018—on the eve of
trial—that significant changes were made. SFFA-FF 113-114.
72. FF 72 is incomplete and misleading. Until September 2018, the Reading Procedures
contained no discussion—“explicit” or “implicit”—about “how to use race in evaluating applicants
or assigning ratings.” SFFA-FF 97-98, 114, 133. These changes were made, in part, to warn admissions
officers not to penalize applicants for personality traits that are often associated with Asian-American
stereotypes. SFFA-FF 114.
73. FF 73 is inaccurate. The Class of 2023 Reading Procedures do not “codify longstanding
practices” and do “reflect [a] change in policy.” SFFA FF 113-114; supra ¶ 37. But Harvard’s contrary
argument, and its decision to rely on those new procedures to defend itself, make clear that Harvard
has waived any Rule 407 objection. SFFA Conclusions of Law (“SFFA-CL”) 193 n.5.
74. FF 74 is inaccurate. Harvard penalizes Asian Americans, engages in racial balancing, and
awards excessively large preferences to African-Americans and Hispanics. Supra ¶ 37. Fitzsimmons
and McGrath exert enormous influence over admissions decisions. Supra ¶¶ 52-54.
75-78. FF 75-78 are incomplete and misleading. Harvard is bad at recruiting, admitting, and
enrolling applicants from modest socioeconomic backgrounds and first-generation college students.
Supra ¶ 23-24. Harvard also does a poor job of obtaining geographic diversity. Supra ¶ 1-5.

11

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 20 of 85

79. FF 79 is inaccurate, incomplete, and misleading. Harvard maintains a “consistent racial
composition of the student body.” Over the ten-year period before SFFA filed suit (Class of 20092018), the share of African-American admits was always between 10.0% and 11.7% (a 1.7% range);
the share of Hispanic admits was always between 8.2% and 11.5% (a 3.3% range); and the share of
Asian-American admits was always between 17.5% and 20.3% (a 2.8% range). D713.2; P319.2-4.
There was similar stability by race for matriculated students. P319.3-4.
80. FF 80 is inaccurate, incomplete, and misleading. As to racial balancing, Harvard’s “expert
testimony” was just simple graphs and charts showing the variations in admission and enrollment
numbers. D711, D713; DD10.100-104; T12.118:15-124:5. Professor Card did not perform regression
analysis to assess whether Harvard was engaging in racial balancing. T13.118:15-124:5. More broadly,
Professor Card’s testimony was not credible. SFFA-FF 63-93. Among other things, his incomplete
analysis never attempted to measure the impact of race on the personal rating, T13.188.23-189:11, or
the overall rating, T14.76:3-11. He also never disclosed his coefficients in his reports, contrary to
standard practice in the field (including his own). T14.14:5-24:6. Perhaps worst of all, he did not
independently investigate whether Harvard considers race in the personal rating. He just accepted the
word of Dean Fitzsimmons. SFFA-FF 64; T14.80:19-81.1; James v. Stockham Valves & Fittings Co., 559
F.2d 310, 332 (5th Cir. 1977) (rejecting expert’s statistical model because he “made no attempt to
control or check for racial bias” in some variables he included).
81. FF 81 is inaccurate. See Supra ¶ 79. Harvard uses tables with scales that are misleadingly
enlarged to give the appearance of large year-to-year fluctuations. D711. The fluctuations in the
admitted class actually demonstrate that Harvard has engaged in racial balancing. For the past two
decades, virtually every time there was a slight drop in the proportion of African-American and
Hispanic admits, there was an increase the following year of a corresponding (and sometimes identical)
magnitude. D711.3-4.

12

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 21 of 85

82. FF 82 is inaccurate, incomplete, and misleading. The share of African-American applicants
was between 7.5% and 10.4% (a 2.9% range), which is nearly twice as large as the range of AfricanAmerican admits during that time period (1.7%); and while the range for Hispanic applicants (2.2%)
was lower than Hispanic admits (3.3%), the range for Asian-American applicants (2.9%) was higher
than Asian-American admits (2.8%). D713; supra ¶¶ 79-81.
83. FF 83 is inaccurate. Harvard’s statistics are cherry picked to rely on the Class of 2019—
the class admitted the year after this suit was filed. SFFA-FF 37, 128. In the prior ten years (Classes
of 2009-2018), there was no change in the share of Asian-American admits: 19.1% of the class in 2008
and 19.1% of the class in 2018. D713.2.
84. FF 84 is incomplete and misleading. The racial-balancing claim can be resolved without
any expert analysis. Supra ¶¶ 79-80; SFFA-CL 207-214.
85-86. FF 85-86 are inaccurate, incomplete, and misleading. Harvard regularly reviews onepagers throughout the year and, in particular, at key points in the admissions process. SFFA-FF 135.
One-pagers do not just summarize the racial characteristics of the admitted class for the current year;
they summarize the racial characteristics for the prior year too. Id. Harvard’s one-pagers display more
information about race than any other characteristic; indeed, nearly half the information displayed on
one-pagers is about race. E.g., P156.
87. FF 87 is incomplete and misleading. Although one-pagers are not physically distributed
to admissions officers, Fitzsimmons reads the racial statistics contained on them aloud to admissions
officers throughout the full-committee process. SFFA-FF 136-137, 139.
88. FF 88 is inaccurate. The Admission Office uses one-pagers to “manage the number of
offers that can be extended” by ensuring that the racial characteristics of the admitted class are stable
year to year. Supra ¶ 79. This consistency ensures that Harvard’s target number of admits (which it
sets in January) accurately predicts the number of applicants who will accept their offer. SFFA-FF

13

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 22 of 85

134-135; P177; T10.117:21-118:3; Doc. 597-1, 291:19-293:25. There is no evidence that Harvard ever
admits fewer applicants than anticipated because it discovered at the end of the process that it was
admitting a “larger than expected” number of applicants from a racial group that yield at a high rate
(like Asian Americans). FF 88.
89. FF 89 is inaccurate. It is true that the Admissions Office uses one-pagers to see how the
class is “shaping up.” If the full-committee process is nearing the end and the share of a certain racial
group is “surprisingly or notably underrepresented,” the full committee will “go back and look at those
cases” with the goal of preventing a “dramatic drop-off in some group who we did at a certain level
with last year.” SFFA-FF 137. It is not credible that Admissions Leaders closely monitor the changes
in the class to “inform the next year’s recruitment efforts.” See PD20.22; e.g., P152 (Dec. 2, 2013),
P153 (Dec. 5, 2013), P154 (Dec. 6, 2013), P155 (Dec. 10, 2013), P156 (Dec. 13, 2013). That close
monitoring is unnecessary to get a “rough idea” about recruiting. T1.160:3-16; T4.81:23-82:13. There
is no evidence that Harvard has ever altered its recruitment of a particular racial group because of the
one-pagers Dean Fitzsimmons received during the admissions cycle. FF 19-27; T8.212:19-218:9.
Fitzsimmons certainly does not ask for his “ethnic stats” at the end of the lopping process because he
wants information about “next year’s recruitment efforts.” SFFA-FF 139; T7.181.
90. FF 90 is inaccurate. Dean Fitzsimmons regularly (and not just “in some years”) reads
aloud to the full committee the admitted percentage of the class by race as of that date and the admitted
percentage of the class by race from the prior year. It is not credible that Fitzsimmons did not want
these announcements to affect the full committee’s decisionmaking. Nor is it credible that these
announcements—from the Dean of Admissions—would not affect the committee’s decisionmaking.
SFFA-FF 136, 137, 139; T5.197:14-17; T8.83:7-16.
91. FF 91 is inaccurate. Harvard does seek a class of a “particular racial composition.” Supra
¶¶ 81-83. Admissions officers did not credibly testify otherwise. SFFA-FF 134-140.

14

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 23 of 85

92-93. FF 92-93 are inaccurate. Harvard does not employ a “whole-person” admissions
process that considers the full range of available information. Supra ¶¶ 30-31. Race is not “considered
as just one factor among many.” SFFA-FF 145-150. Race does not “make a difference only for
applicants who would be highly competitive no matter what their race.” Supra ¶ 61. Race is not
considered as “an effort to understand each applicant as a whole person,” and admissions officers did
not credibly testify otherwise. SFFA-FF 141-144.
94. FF 94 is inaccurate, incomplete, and misleading. The use of race has a “meaningful effect”
on applicants who are not “highly qualified” and who would not be “highly competitive for admission
regardless of their race.” Supra ¶ 61. Parental occupation and intended career do not play a larger role
in admissions than race. As Professor Card testified, the average marginal effect of being Hispanic on
the probability of admission is 3.73%, and the average marginal effect of being African American on
the probability of admission is 6%. Professor Card’s own analysis demonstrates, in other words, that
Harvard’s use of race—by itself—doubles a Hispanic applicant’s chance of admission, and triples an
African-American applicant’s chance of admission. SFFA-FF 146.
95. FF 95 is inaccurate, incomplete, and misleading. Race plays an important role for
applicants who are not otherwise “highly qualified” and who would not be “highly competitive for
admission regardless of their race.” Supra ¶ 61. Race does not play a minimal role in admissions
decisions. SFFA-FF 145-150; PD38.39. It is true that “the effect of race is comparable to” certain
“other factors.” SFFA-FF 149.
96. FF 96 is accurate, incomplete and misleading. Harvard’s attempts to smear Professor
Arcidiacono are disappointing. Professor Arcidiacono is respected, fair, unbiased, and highly qualified.
SFFA-FF 23; T14.13:5-8. Indeed, Professor Michael Keane—a recognized leader in discrete choice
modeling—supports Professor Arcidiacono and his work in this case. T14.90:14-98:11; Doc 624,
Amicus Brief of Michael Keene et al. (Jan. 9, 2019) & Ex. A. SFFA likewise could have cherry-picked

15

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 24 of 85

from Professor Card’s academic work to unfairly attack him. See, e.g., T14.12:22-13:24. Harvard’s
efforts to smear Professor Arcidiacono should be considered in assessing its credibility.
97. FF 97 is inaccurate. Professor Arcidiacono did not testify that “race has no effect on the
vast majority of applicants.” T9.200:1-3; T25.200:1-17, 201:12-25. Nor did he testify that “race makes
a difference only for applicants who would otherwise be competitive.” T9.200:1-17. Race can make a
difference for applicants who would not “otherwise be competitive.” Supra ¶ 61.
98. FF 98 is inaccurate. There is powerful evidence that Harvard intentionally discriminates
against Asian-American applicants. FF 21-133.
99. FF 99 is incomplete and misleading. Both experts built logit regression models to evaluate
the role of race Harvard admissions. SFFA-FF 21-23; T9.13:16-21. Both excluded applicants from
international dockets and foreign students on domestic dockets. T14.47:11-48:12. And both agreed
that the overall rating was influenced by race and must be excluded. T9.91:24-92.2; T14.77:22-78.4.
Professor Arcidiacono also explained that, although his model does not capture every factor that can
contribute to admission, including certain qualitative information, that does not make it unreliable. It
“would be the downfall of empirical economics if that were the case because all models have
unobservables.” T9.81:20-21. Professor Card testified that these logistic regression models are “the
standard . . . to use in analyzing something like an admissions decision which is -- in the context of a
case like this with many different attributes of candidates that need to be taken into consideration and
making a determination about one particular feature like ethnicity.” T12.101:5-17.
100. FF 100 is accurate
101. FF 101 is inaccurate, incomplete, and misleading. Professor Card did find that AsianAmerican ethnicity has a negative effect on admission when he removed the personal rating from his
model. SFFA-FF 68.

16

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 25 of 85

102-103. FF 102-103 are inaccurate. Professor Arcidiacono more accurately modeled the role
of race in Harvard’s admission system—i.e., the task assigned to the experts. Supra ¶ 99. If the experts
were merely tasked with replicating “Harvard’s admissions process as closely as possible” they would
not have excluded the overall rating; nor would they have excluded thousands of applicants (such as
foreign applicants). SFFA-FF 21, 74; T14.47:11-48:12. And, if the experts were tasked with trying to
include all variables that “actually play a role in the Harvard admissions process,” they would not have
excluded variables (such as the overall rating) that are influenced by race. SFFA-FF 43. Professor
Arcidiacono’s conclusions are reliable and support a finding that Harvard intentionally discriminates
against Asian-American applicants. SFFA-FF 21-93.
104. FF 104 is incomplete and misleading. Once the personal rating is removed, Professor
Card’s own model shows that Harvard imposes a statistically significant penalty on Asian Americans.
SFFA-FF 68.
105. FF 105 is inaccurate, incomplete, and misleading. Professor Card’s finding that Harvard
does not penalize female Asian-American applicants and Asian-American applicants from California
has all the same flaws as his primary analysis and suffers from an overfitting problem. SFFA-FF 93.
106. FF 106 is inaccurate. Professor Arcidiacono’s finding of a statistically significant penalty
on Asian Americans was not the result of unsound methodological choices. SFFA-FF 63-92. But even
if the Court accepts all of Professor Card’s flawed modeling choices—except inclusion of the personal
rating—there is still a statistically significant penalty on Asian Americans. Supra ¶ 104.
107. FF 107 is inaccurate, incomplete, and misleading. Professor Arcidiacono has a “baseline”
model that excludes ALDCs and an “expanded” model that includes LDCs. Even including LDC
applicants, Professor Arcidiacono’s model shows a statistically significant penalty on Asian Americans
SFFA-FF 77. The same is true when Professor Arcidiacono includes ALDCs in his regression model.

17

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 26 of 85

T10.80:4-21. Professor Card’s model includes all ALDC applicants; and, once the personal rating is
excluded, it shows a statistically significant Asian-American penalty. SFFA-FF 77.
108. FF 108 is inaccurate, incomplete, and misleading. Professor Arcidiacono did not concede
“that Asian-American ethnicity has a positive effect on the likelihood of admission for ALDC
applicants.” He acknowledged that, for LDC applicants, the coefficient for Asian-American ethnicity
in admissions is positive, T9.121:1-3, but was not statistically significant, T9.126:4-5, which means that
its “effect is not distinguishable from zero,” T9.190:4-5. To claim that a positive coefficient by itself
means a “positive effect” suggests that Harvard either misunderstands statistical analysis or is being
purposely disingenuous, given that Professor Card repeatedly explained that a positive effect that is
not statistically significant is “not statistically different than zero, so we could think of it as roughly
zero.” T12.109:12-110:7, 128:5-9. What the data actually mean is that Harvard penalizes 98% of AsianAmerican applicants (nearly 40,000 applicants) in admissions decisions; and for the less than 2% of
Asian-American applicants who are LDCs, their ethnicity has no significant effect on their admissions
outcome because the LDC preference is so large that it overwhelms any Asian-American penalty
imposed. SFFA-FF 71-72.
109. FF 109 is incomplete and misleading. Removing ALDC applicants from Professor
Arcidiacono’s model does result in a “more negative effect on Asian-American ethnicity,” but this is
because of the distorting effect that including ALDC applicants has on the ratings (in particular, the
academic and extracurricular ratings), infra ¶ 110, not because of the 2% of Asian-American applicants
who are ALDC, T10.70:13-71:2.
110. FF 110 is inaccurate. Professor Arcidiacono did not exclude ALDCs merely because they
receive a “tip.” He did so to ensure “an apples-to-apples comparison” of applicants. T9.27:8-10.
ALDCs are afforded procedural and substantive advantages in the admissions process that make them
quite unlike non-ALDC applicants. SFFA-FF 69-73. Moreover, ALDCs have a distorting effect on

18

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 27 of 85

how the ratings (especially the academic and extracurricular ratings) affect the admission decisions of
non-ALDC applicants, thus preventing an “accurate estimate[] for the 95 percent of applicants who
are not in one of those groups.” Because academic and extracurriculars are less important for ALDCs,
it gives the erroneous impression that they are less important for non-ALDCs too. Because Asian
Americans excel on these ratings, undervaluing them underestimates—and thus conceals—the true
magnitude of the Asian-American penalty. SFFA-FF 72-73. Professor Arcidiacono also excluded
ALDCs because Harvard historically has defended itself against claims of discrimination against Asian
Americans by relying on the preferences it gives to athletes and legacies. T9.27:17-22. If that defense
were valid, a regression analysis using the 95% of applicants who are non-ALDCs would show “no
penalty against Asian-Americans.” T9.28:1-6. This is how both OCR and OIR investigated claims of
Asian discrimination at Harvard. SFFA-FF 76. Professor Card never did so even though Harvard
previously supported analyzing the data this way. FF 107-116.
111. FF 111 is inaccurate. In the admissions process, ALDCs have procedural and substantive
advantages compared to their non-ALDC peers and Professor Arcidiacono did not exclude ALDCs
merely because they receive a “tip.” Supra ¶ 110. Attempting to “control” for ALDC status instead of
excluding these applicants would have been improper. First, the experts excluded—and did not
“control” for—foreign applicants because they are not similarly situated to the rest of the applicant
pool and thus inhibit an apples-to-apples comparison. SFFA-FF 21; T14.47:11-48:12. Second,
controlling for ALDC status would not eliminate the distorting effect that including ALDCs has on
the profile ratings. Because the profile ratings operate differently for ALDC applicants, the only way
to properly attempt to control for ALDC status would be to include interactions between ALDC
status and all the profile ratings. T9.85:4-14. But Professor Card’s yearly model lacks sufficient power
to accommodate all of these additional interaction variables, T9.143:5-10, and would not have altered
Professor Arcidiacono’s findings in any event, SFFA-FF 80.

19

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 28 of 85

112. FF 112 is inaccurate, incomplete, and misleading. Professor Arcidiacono did not testify
that ALDCs “participate in a different process.” ALDCs have procedural and substantive advantages
in the admissions process compared to their non-ALDC peers. Supra ¶ 110. Moreover, describing
ALDCs as merely having “higher-than-average rates of admission” understates the massive disparities
in ALDC and non-ALDC admit rates. Legacies are admitted at more than five times the rate of all
other applicants (33.6% vs. 5.9%); children of Harvard faculty/staff are admitted at more than six
times the rate of all other applicants (46.7% vs. 6.6%); applicants on the Dean’s or Director’s lists are
admitted at nearly seven times the rate of all other applicants (42.2% vs. 6.1%); and recruited athletes
are admitted at more than fourteen times the rate than all other applicants (86.0% vs. 6.0%). SFFAFF 71. It is also misleading to claim that ALDCs are stronger than the “overall pool.” Admitted nonALDCs, as a group, are stronger than their ALDC counterparts; non-ALDC admits have a higher
share of scores of 2 or better on every rating except for the athletic rating. Compare P621 with P623.
113. FF 113 is inaccurate, incomplete, and misleading. ALDC admits, as a group, are weaker
than non-ALDCs, and their exclusion does not make Professor Arcidiacono’s model unreliable
because they are a “significant part of the admissions process.” Supra ¶¶ 110-112. Professor Card
excluded from his model all applicants from international dockets, as well as foreign applicants from
domestic dockets. This group is almost the same size as the entire set of ALDC applicants. P319. And
as Professor Card conceded, the number of international/foreign admits is approximately the same as
the number of Hispanic admits. T14.51:7-11; P319.4.
114. FF 114 is inaccurate. Professor Arcidiacono did not exclude ALDCs from the analysis
because they receive a “tip,” nor did he exclude them to increase the size of the Asian-American
penalty. Supra ¶¶ 110-113. Professor Arcidiacono also conducted regression analyses that included
LDCs (his expanded dataset), and those results confirmed his findings with respect to non-ALDC

20

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 29 of 85

applicants. SFFA-FF 77. Among OIR, OCR, and Professor Arcidiacono and Professor Card, only
Professor Card refused to look separately at the non-ALDC pool. Supra ¶ 110.
115. FF 115 is inaccurate, incomplete, and misleading. Professor Arcidiacono did not exclude
ALDCs because they receive a “tip” or because they are “admitted at higher-than-average rates.” Supra
¶¶ 110-113. Professor Arcidiacono controls for other tips because they do not have a distorting effect
or because removing them makes no difference. D693; T9.234:14-15 (excluding applicants with “an
academic rating of 1” “would make no difference”). For example, he included early-action applicants
(who receive a preference for applying early) because, unlike ALDCs, “the way the early-action process
rewards all characteristics, including race, was similar between the early-action applicants and the
[regular-admission] applicants.” T9.245:3-6.
116. FF 116 is inaccurate. Excluding ALDCs is the proper way to determine whether Harvard
is penalizing Asian Americans. Supra ¶¶ 107-115. But even including LDCs, Professor Arcidiacono’s
model shows a statistically significant penalty on Asian-American applicants. As does Professor Card’s
analysis including athletes, once the personal rating is removed. SFFA-FF 77.
117-118. FF 117-118 are incomplete and misleading. Professor Arcidiacono’s pooled model
controls for year-to-year variations; in contrast, Professor Card’s yearly model does not capture any
of the many benefits of the pooled approach. SFFA-FF 78-79. For example, pooling applicants from
six years gives the model “more statistical power” which allows for the inclusion of interactions
between race and disadvantaged status in order to capture “the subtleties for how race operates.”
T9.143:5-10. Professor Card’s yearly model lacks the power to be able to include these interactions.
Id. OIR also conducted a pooled analysis and included interactions between race and low-income
status. T5.118:21-119:11; 110:21-111:8.
119-120. FF 119-120 are incomplete and misleading. Professor Card’s year-by-year model
does have a “slightly lower standard error.” But his approach leaves his model without enough

21

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 30 of 85

statistical power to account for the ways in which race operates differently across other variables;
Professor Card thus cannot include interaction variables in his model to capture the way that race
interacts differently with disadvantaged status. T9.143:4-10. So it is misleading to suggest that
Professor Card’s yearly model is more precise—it has a slightly lower standard error, but it also fails
to capture the true effect that race has on admissions at Harvard. In any event, yearly versus pooled
makes no difference, as Professor Card testified. T13.90:12-14 (“If one was to take that model and
estimate the model in a pooled framework, which Professor Arcidiacono does, essentially it doesn’t
make any difference.”). Moreover, Harvard imposes a statistically significant penalty on Asian
Americans even using the yearly approach that Professor Card prefers. SFFA-FF 80.
121-123. FF 121-123 are inaccurate, incomplete, and misleading. The experts agree that it was
proper to omit certain variables and both, in fact, omitted variables that play a role in the admissions
process (like the overall rating). Supra ¶¶ 102-103. The dispute is whether it was proper to omit these
variables. Professor Arcidiacono correctly omitted parental occupation, intended career, the staffinterview indicator, and the personal rating from his baseline model. Omitting them did not make the
Asian-American penalty bigger. It revealed the true magnitude of the penalty. SFFA-FF 63-68, 81-89.
124. FF 124 is inaccurate, incomplete, and misleading. The cited evidence does not support
the assertion that intended career and parental occupation are “correlated with race.” There are racial
variations for certain careers and occupations, but these variations are highly unusual in the way they
change year over year. SFFA-FF 81, 85. These variations make these data unsuitable for inclusion in
the model if the object is to meet the standard in the field for a peer-reviewed publication. SFFAFF 81-83, 85-86; T9.146:4-148:12, T9.147:10-15. Professor Card did not even use parental occupation
as it exists in Harvard’s database. He recoded these variables in a way that yields nonsensical results.
SFFA-FF 81-82. For example, he would “code the owner of a small convenience store the same way
[he] would code one of the Koch brothers,” and he would give “the Secretary of Energy . . . the same

22

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 31 of 85

code as . . . a regional sales manager for Dunder Mifflin.” T14.30:15-18, 31:23-32:3. Professor Card
conceded that Harvard would not treat these occupations similarly. T14.30:19-22, 32:4-9. Regardless,
Harvard imposes a statistically significant penalty on Asian Americans even including intended career
and parental occupation in the analysis. SFFA-FF 84, 87.
125. FF 125 is inaccurate, incomplete, and misleading. Professor Arcidiacono did not exclude
the intended career and parental occupational variables from his baseline model because of “year-toyear fluctuations” in the data. He excluded them because, among other reasons, the data varied in
dramatic and unexplainable ways that made them unreliable. Professor Card’s attempt to recode these
data to make them usable does not solve the problem; if anything, it exacerbates the unreliability of
these data because it yields results that even he acknowledges are nonsensical. Supra ¶ 124.
126-127. FF 126-127 are inaccurate, incomplete, and misleading. Receiving a staff interview
is affected by other preferences such as ALDC status. For example, ALDC applicants are far more
likely to receive a staff interview and are afforded them outside the published timeframe. T5.181:18184:16; PD38.3. The Admissions Office also has a target number for athlete interviews. T5.184:3-11.
Unsurprisingly then, more than 20% of ALDCs receive staff interviews, while only 1.26% of nonALDCs do. P619. Receiving a staff interview—irrespective of how it is scored—has an outsized effect
on admission. More than 50% of applicants who receive staff interviews are admitted; and ALDCs
who receive staff interviews are more than twice as likely to be admitted as non-ALDCs who receive
them (78.55% versus 29.48%). T9.37:6-38:20; P619; PD38.2. The evidence thus shows “that staff
interviews mean something different for ALDC applicants versus non-ALDC applicants, which is not
in the model.” T10.23:22-25. Given the high admissions rate for applicants who receive staff
interviews, it is suspicious that Asian Americans are afforded staff interviews at the lowest rate of any
race, T9.57:22-24, especially when they seek and obtain alumni interviews at the highest rate, P623.
Staff interviews thus are not properly included as a variable. SFFA-FF 88-89, T9.38:11-20; T9.148:13-

23

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 32 of 85

18. Regardless, Harvard imposes a statistically significant penalty on Asian Americans even including
staff interviews. SFFA-FF 89.
128-129. FF 128-129 are inaccurate, incomplete, and misleading. Professor Arcidiacono
acknowledged that staff interviews can provide valuable information about an applicant, but that is
beside the point. The dispute regarding the inclusion of the staff-interview variable is about who
receives them and why—not how the interview is scored. Professor Card did not even use the actual
staff interview rating—only the fact that one took place. Supra ¶¶ 126-127; T10.13:18-20. Professor
Arcidiacono did not exclude staff interviews because Asian-American applicants are less likely to
receive them. Nor did he exclude staff interviews because ALDC applicants are merely “more likely
to participate in staff interviews.” ALDCs are 5% of the applicant pool; yet they are 42% of those
who receive staff interviews (again, sometimes outside of the process available to non-ALDCs).
Indeed, ALDCs are nearly 16 times as likely as non-ALDCs to receive a staff interview. Supra ¶¶ 126128; P619.
130. FF 130 is inaccurate, incomplete, and misleading. Professor Card’s inclusion of staff
interviews is flawed for many reasons. Supra ¶¶ 126-129. Professor Card’s attempt to control for this
variable does not solve the problem. Most notably, the variable does not reflect any evaluation of the
applicant at all; it reflects only whether the applicant received a staff interview. SFFA-FF 88-89. In
other words, Professor Card does not control for how the applicant performed in the interview; he
controls for a binary result: “participating in a staff interview.” FF 130. Professor Card testified that a
staff interview might yield “personal information and insight” to “someone who is going to be present
at the subcommittee and [full] committee.” T13.26:3-9. But this goes to the ratings assigned at staff
interview—not to whether or not an applicant received a staff interview.
131-132. FF 131-132 are inaccurate. Professor Arcidiacono did not exclude the personal rating
because “applicants of different races received ‘slightly’ different personal scores.” The racial

24

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 33 of 85

disparities are stark; and they exhibit a pattern that closely tracks the racial hierarchy in the overall
rating—which Professor Card agreed is influenced by race and should be excluded. SFFA-FF 48.
Professor Arcidiacono thus excluded the personal rating because race significantly influences it. SFFAFF 49-58, 63-68. Professor Card studiously avoided testifying in his highly scripted direct testimony
that the personal rating is not influenced by race; indeed, he testified that he could not “rule out racial
bias” as the explanation for “the unexplained gap between whites and Asians in the personal rating.”
T13.189:12-190:3. Harvard uses the personal rating to grant a massive preference to African
Americans and Hispanics and to impose a statistically significant penalty on Asian Americans. SFFAFF 49-58; PD38.30; PD38.31. Harvard’s consideration of race increases an African-American
applicant’s likelihood of receiving a 2 or higher on the personal rating by 27%; at the same time,
Harvard imposes a penalty on Asian-American applicants that makes them 17.6% less likely to receive
a 2 or higher on the personal rating. PD38.31.
133.

FF 133 is inaccurate, incomplete, and misleading. Harvard’s witnesses, especially

Christopher Looby, did not credibly testify that they do not consider race in assigning the personal
rating. SFFA-FF 132. Nor did their testimony comport with what OCR and OIR found, the confusion
among Harvard admissions officers about when and how race should be used, and the reasons given
for making changes to the Reading Procedures. SFFA-FF 101-103, 113-114. Several witnesses—
including Dean Fitzsimmons and Director McGrath—testified falsely as to whether Harvard has
written guidelines on the use of race and thus initially concealed that Harvard had made substantial
revisions regarding the personal rating beginning in the summer of 2018 and continuing up to the eve
of trial. SFFA-FF 194-95, 199. Harvard and its witnesses lack credibility on this issue.
134. FF 134 is inaccurate. There is powerful statistical evidence that “the personal score is
informed by an applicant’s race.” Supra ¶¶ 131-132. Given that Professor Card conceded that he could
not rule out the possibility of racial bias in the personal rating, T13.125:22-126:13, his decision to rely

25

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 34 of 85

on Fitzsimmons’ denials for why he failed to statistically examine the issue is especially indefensible,
supra ¶ 80.
135-136. FF 135-136 are inaccurate, incomplete, and misleading. Professor Arcidiacono found
that the stronger performance of Asian Americans on the academic and extracurricular rating is not
attributable to racial bias, while the “negative association between Asian-American ethnicity and the
personal score” is due to racial bias. Those are not inconsistent findings. It is what the data shows to
an econometrician. Asian Americans perform the best of any racial group on the factors that drive the
academic and extracurricular ratings. SFFA-FF 24-26. Professor Card also does not believe that Asian
Americans receive a racial preference in the academic or extracurricular ratings. T14.85:9-13. In
contrast, the data cannot explain why African Americans perform the best on the personal rating,
followed by Hispanics, then by whites, with Asian Americans faring the worst, when the observable
data would have the hierarchy run in exactly the opposite fashion. Only the use of race—to the benefit
of African Americans and Hispanics and to the detriment of Asian Americans—can explain the racial
divergence of observable measures from personal ratings outcomes. SFFA-FF 52-57.
137. FF 137 is accurate.
138. FF 138 is inaccurate, incomplete, and misleading. Asian Americans are stronger than
white applicants on the observable factors relating to the personal rating. SFFA-FF 55. Professor
Card, moreover, does not directly rebut Professor Arcidiacono’s finding; he never models the personal
rating. Supra ¶ 80. He created an index focused on admission, and this “non-academic admissions
index” inappropriately includes the personal rating, T13.73:10-13, the very rating at issue.
139. FF 139 is inaccurate, incomplete, and misleading. “Asian-American applicants were” not
“weaker than White applicants on ratings assigned by alumni interviewers for . . . overall strength.”
The share of Asian-American applicants receiving a 2 or better on the alumni overall rating (51.14%)
is significantly larger than the share of white applicants (46.65%). P621; PD38.9. And though white

26

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 35 of 85

applicants have slightly higher shares of 1s and 2s on the alumni personal rating (63.13% v. 62.25%),
the difference is tiny by comparison to Asian Americans’ superior scores on the alumni overall rating.
P621; PD38.9. It also is inaccurate to claim that Asian-American applicants are weaker on the “ratings
assigned by admissions officers to recommendation letters written by applicants’ high school teachers
and guidance counselors.” On these school-support ratings, white and Asian-American applicants
scored essentially the same; on all three ratings, the shares of white and Asian-American applicants
receiving a 2 or better were within 1 percentage point of the other: white applicants scored slightly
higher on the counselor rating and the teacher 2 rating (0.45% and 0.82% differences, respectively)
and Asian Americans scored slightly higher on the teacher 1 rating (0.18% difference). P621; PD38.9.
Professor Card tries to obscure this similarity by summing the ratings. But as he conceded, his
summing technique has no basis in how Harvard actually evaluates applicants and incorrectly assumes
that an applicant having ratings of “1 and a 2 and a 3 . . . is the same as 2, 2, 2.” T13.125:14-126:20.
Even if that were a valid way to look at the issue, that “whites have slightly higher scores than Asian
Americans on the school-support ratings does not justify an inference that Asian Americans score
substantially worse than whites on other unobserved variables that influence the personal rating. Yet
this is precisely the inference that would be necessary to conclude that unobserved variables could
explain the substantially lower personal-rating scores of Asian-American applicants.” Doc 624, Keane
Amicus Brief at 7-8. Professor Card also conceded, by refusing to rebut it, Professor Arcidiacono’s
finding that the assignment of the school support rating determinations is influenced by race and
impose a penalty on Asian-American applicants. Supra ¶ 47. Only by making all of these errors could
Harvard makes the circular claim that Asian-American applicants are “weaker, on average, across the
full range of non-academic factors in the model.”
140. FF 140 is inaccurate. Harvard’s use of race—not factors outside the data—best explains
the association between Asian-American ethnicity and weaker personal ratings. Supra ¶¶ 135-139.

27

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 36 of 85

Harvard’s theory that the racial disparity in the personal rating could be explained by the differences
in how the racial groups score on the alumni and schools-support ratings falls apart once the scores
of the African American and Hispanics on these ratings are incorporated into the analysis. If Harvard’s
theory were correct—i.e., that the alumni and school-support ratings explain the racial hierarchy of
the personal rating—then African Americans and Hispanics should be significantly outscoring their
white and Asian-American peers on these school-support ratings because they are outscoring whites
and Asian Americans on the personal rating. But they are not. Asian Americans get substantially higher
scores than African Americans and Hispanics on all five of these ratings. P621. The only “factual,
logical, and statistical” explanation for the racial disparity in the personal rating is the influence of race.
Professor Arcidiacono’s model of the personal rating includes the school-support and alumni ratings,
moreover, so whatever effect they have cannot explain the Asian-American penalty Harvard imposes
in assigning the personal rating. SFFA-FF 55.
141. FF 141 is incomplete and misleading. The economist amici supporting Harvard recited
Professor Card’s conclusions without addressing his numerous errors. That is unsurprising given that
these amici are not experienced in discrete choice modeling—the type of economic modeling the
experts performed. T10.50:1-9. In contrast, the amici who agree with Professor Arcidiacono (including
Professor Keane) are leading econometricians who are experienced in discrete choice modeling.
Professor Card relies on Professor Michael Keane’s work in connection with his teaching duties: he
has assigned Professor Keane’s work when he teaches discrete choice modeling to his labor economics
students; and in the handbook of labor economics that Professor Card edits, “the discrete choice
modeling chapter is by [Professor] Keane.” T14.90:23-92.19.
142. FF 142 is incomplete and misleading. That Asian-American ALDCs suffer on the
personal rating compared to white ALDCs supports Professor Arcidiacono’s findings. It proves that
Harvard’s racial stereotyping harms all Asian-American applicants. SFFA-FF 94-95, 105-112. Asian-

28

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 37 of 85

American ALDC applicants are lucky enough to be preferred among a penalized group and therefore
suffer no statistically significant penalty in their admissions outcomes—an unsurprising result given
that this is Harvard’s donor and employee base. SFFA-FF 75; supra ¶ 108
143-144. FF 143-144 are inaccurate, incomplete, and misleading. Professor Card’s position on
the personal rating contradicts his opinion on the overall rating, which he agreed should be removed
from the model. SFFA-FF 43, 48. Moreover, there is no race effect to be removed from the academic
and extracurricular ratings; both Professor Card and Professor Arcidiacono agree that race does not
affect those ratings. Supra ¶¶ 135-136. Professor Card also did not include a race-adjusted overall
rating, or race-adjusted school-support ratings; Professor Card’s virtual rating model is worthless.
SFFA-FF 66. But if Professor Card truly believed that all of these ratings are influenced by race, then
he should have run a model excluding all the ratings. Professor Arcidiacono ran that model and it
shows a statistically significant penalty against Asian Americans. SFFA-FF 67.
145. FF 145 is inaccurate, incomplete, and misleading. Omitting the parental occupation,
intended career, staff interview, and personal ratings did not make the Asian-American penalty bigger.
It revealed the true magnitude of the penalty. Supra ¶¶ 121-123. Further, Harvard imposes a statistically
significant penalty on Asian Americans even including the parental occupation, intended career, and
staff interview variables in the analysis. SFFA-FF 84, 87, 89.
146. FF 146 is inaccurate. Professor Arcidiacono’s analysis supports an inference that Harvard
intentionally discriminates against Asian-American applicants. He determined that Harvard imposes a
statistically significant penalty on Asian Americans that cannot be explained on non-racial grounds.
SFFA-FF 49-58.
147. FF 147 is inaccurate and misleading. To take the position that a regression model based
on real-world data cannot be used to prove the existence of discrimination is to take the position that
there is no place whatsoever for regression analysis in discrimination cases (or for that matter, in any

29

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 38 of 85

evaluation of racial discrimination). Infra CL ¶ 271-273. This is an untenable position that undermines
Professor Card’s credibility—especially because it contradicts all of Professor Card’s work, academic
writings, and testimony in this case. Supra ¶ 99.
148. FF 148 is inaccurate, incomplete, and misleading. There are always unobservable factors
that cannot be accounted for in a regression model; but that does not mean that sound regression
modeling cannot produce reliable conclusions. Supra ¶ 99. In fact, the data set here is much richer
than in most econometric models. T9.20:2-19. Further, the application files that Harvard produced
corroborate Professor Arcidiacono’s findings. T9.133:7-142:12. Professor Card, in contrast, never
reviewed a single application file. T12.81:16-82:13.
149. FF 149 is inaccurate. Professor Arcidiacono did not concede that Asian-American ALDC
applicants are admitted at a higher rate than white ALDC applicants. He acknowledged that AsianAmerican ALDCs are admitted at a slightly higher rate without regard to the relative strength of those
applicants. T9.121:7-122:4. And the claim that Asian-American ethnicity has a positive effect on
admission for legacy applicants is based on a model that improperly includes recruited athletes.
T12.141:4-14. In any event, it is unsurprising that Harvard’s Asian-American penalty is concentrated
among non-ALDCs. SFFA-FF 75-76. And even including LDCs, Professor Arcidiacono’s model
shows a statistically significant penalty on Asian-American applicants. SFFA-FF 77.
150. FF 150 is inaccurate, incomplete, and misleading. That Professor Card’s model includes
certain variables that Professor Arcidiacono excluded does not make it more reliable—it makes it less
reliable. Supra ¶¶ 102-103, 121-123.
151. FF 151 is inaccurate. The statistical evidence supports a finding that Harvard intentionally
discriminates against Asian-American applicants. SFFA-FF 21-93.
152. FF 152 is inaccurate. There was a “growing concern within the Asian American
community over the possibility of discrimination in the selection of applicants for admission by some

30

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 39 of 85

of the most prestigious colleges and universities in the country.” P555.1. But OCR also investigated
Harvard because “specific concerns about the undergraduate admissions program at Harvard were
raised directly to the Department of Education and to OCR.” P555.2.
153-154. FF 153-154 are inaccurate, incomplete, and misleading. OCR found that the mean
score over a ten-year period on the personal rating was 2.84 for Asian-American applicants and 2.79
for white applicants (a 0.05 difference). P555.33. Thus, it found that “the magnitude of the difference
between the two groups [on the personal rating] was small” and that “the Asian American and white
applicant pools are similar in overall quality.” P555.34. OCR “could not conclude that the disparity in
Asian American and white admit rates was attributable to differences in the quality of the respective
applicant pools.” P555.34. OCR did not review application files and summary sheets to “investigate
the cause of the difference” in personal ratings. P555.19-28. Nor did OCR conclude that Harvard’s
personal ratings “fairly assessed each individual application file.” It found that “[g]iven the subjective
nature of the process, . . . it was almost impossible to determine that a rating was inaccurate. OCR
could not determine, for example, whether an applicant, based on all the information in a file, should
have received a ‘1’ rather than a ‘2’ on ‘personal qualities’, when a 1 was defined in the Reading
Procedures as ‘Outstanding’ and a 2 as ‘Very Strong.’” P555.45.
155. FF 155 is inaccurate. OCR found “a number of cases” in which “Asian American
applicants were described as ‘quiet, shy, reserved, self-contained, soft spoken’”; that these descriptions
“were underlined for added emphasis by the reader”; and that these descriptions were “ascribed to
Asian American applicants more frequently” than white applicants. P555.25. OCR also found
“insensitivity” to the challenges that Asian Americans face. Id.; supra ¶ 154.
156. FF 156 is incomplete and misleading. OCR found that Asian Americans faced a “small
but statistically significant disadvantage in the admissions process,” but that “this disadvantage is
virtually eliminated if legacies and recruited athletes (groups with few Asian Americans) are removed

31

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 40 of 85

from the Asian American and white samples.” P555.45-46. That is no longer Harvard’s view. In fact,
Harvard now claims that it is improper to remove ALDCs from the analysis. SFFA-FF 69. OCR also
found that while Asian Americans performed better on the academic rating, whites performed better
on the extracurricular rating. P555.34. That is no longer true; today, Asian Americans perform better
in both categories. SFFA-FF 24-26.
157. FF 157 is inaccurate. The cited evidence does not support the assertion that Dean
Fitzsimmons reviews NLNA statistics because he wants “to ensure that Asian-American applicants
are treated fairly in the process.” The evidence is to the contrary. SFFA-FF 107, 121, 125-127. The
cited evidence does not support Harvard’s assertion that the admission rate of Asian-American NLNA
applicants has been higher than that of white NLNA applicants in “many admissions cycles between
1990 and the present.” That Asian-American NLNA applicants in some years were admitted at higher
rates than NLNA white applicants does not prove the absence of discrimination given how much
stronger Asian-American applicants are than white applicants. SFFA-FF 36.
158. FF 158 is incomplete and misleading. OCR also found that Harvard uses race in the
personal score and stereotypes Asian-American applicants. SFFA-FF 101, 192.
159. FF 159 is incomplete and misleading. OCR found that Harvard did not discriminate
against one applicant because, among other reasons, the applicant’s SAT score of 1350 “fell below the
median and mean scores of the applicants to whom Harvard offered admission.” D44.3.
160. FF 160 is incomplete and misleading. OIR does “very, very good” work. T3.102:24103:10, 147:14-17. Harvard leaders, including Dean Fitzsimmons, regularly rely on OIR’s statistical
analysis when making important decisions relating to admissions policies, including, for example, the
decision to reinstate early action. P288.1, 42, 44-45, 51-53, 56, 61-71, 72, 79, 117-136. On these
questions, OIR “collaborated closely with Dean Bill Fitzsimmons and Director of Financial Aid Sally
Donahue to understand the climate and nuances of the data.” P288.1.

32

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 41 of 85

161. FF 161 is inaccurate, incomplete, and misleading. The cited evidence does not support
the assertion that Hansen alone created the models shown in the February 25, 2013 presentation to
Dean Fitzsimmons (“February 2013 Report”). OIR employees Driver-Linn, Bever, and Hansen
worked on the February 2013 Report and presented their findings to Fitzsimmons. SFFA-FF 119120; T8.21:4-7; P15; T3.102:7-11; T5.96:1-17. The purpose of the February 2013 Report was not to
“simulate the racial composition of the admitted class if the admissions process considered only a
limited set of factors.” The purpose was to analyze whether “the admissions process disadvantage[s]
Asians.” P12.3; T8.13:2-18:21. The February 2013 Report analyzed the “important variables” in the
admissions process. T3.90:1-10. OIR and Fitzsimmons believed the report was accurate, complete,
and reliable. T4.211:7-24, 238:4-7, 238:22-240:21, 241:7-243:2; P35.1-7; T5.12:15-14:10, 39:6-8, 96:117, 111:9-25; T3.102:24-103:10, 147:14-17.
162. FF 162 is inaccurate, incomplete, and misleading. The academic index is “considered in
the admissions process.” P112; T9.43:11-19. The academic index is printed in bold on the summary
sheet and the components of this index clearly affect the admissions process. P112.2, T9.43:11-19,
57:22-58:11. The academic index allows those components to be summarized in one variable, just as
Professor Card’s “admissions index” encompasses multiple variables that affect admissions decisions.
Regardless, the academic index—a weighted average of an applicant’s GPA and standardized test
scores—is an appropriate way to sort applicants on the basis of objective academic qualifications.
SFFA-FF 24; T9.57:11-58:11.
163. FF 163 is inaccurate, incomplete, and misleading. The February 2013 Report was not
“consistent with OCR’s 1990 Report.” It showed that the personal rating penalized Asian Americans.
SFFA-FF 120. OCR made no finding that the personal rating penalized Asian Americans. Supra ¶ 154.
OCR found that preferences for legacies and recruited athletes explained their “disadvantage in the
admissions process.” P555.31-46. If Dean Fitzsimmons believed that the February 2013 Report was

33

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 42 of 85

“consistent” with his understanding, then he understood the personal rating to decrease the share of
Asian Americans by more than 5 percent—and he intended this result.
164. FF 164 is inaccurate, incomplete, and misleading. The cited evidence does not support
the assertion that Hansen alone created the models shown in the February 2013 Report. Driver-Linn,
Bever, and Hansen worked on it and presented their findings to Dean Fitzsimmons. Supra ¶ 161. As
Harvard concedes, Models 1, 2, and 3 show the effect of various factors, such as legacy status and the
personal rating, on the admissions chances of each racial group. FF 163. Model 4 demonstrated that
the one factor not included in those models—racial “tips”—accounted for the difference between
what Models 1-3 collectively showed and the racial composition “of the actual admitted class.” Model
4 thus is not “circular.” P12.34; T8.77:7-78:7. As OIR explained: “Once we account for ratings and
demographic factors, we can closely predict what the admitted class will look like.” P12.36.
165. FF 165 is inaccurate. The February 2013 Report did show “the effect of any particular
factor on an applicant’s chance of admission.” Because Dean Fitzsimmons knew that Asian Americans
did better on the extracurricular rating, he knew that the personal rating was penalizing Asian
Americans. SFFA-FF 120. OIR also was aware of each factor’s effect because it had created
coefficients for each variable. T8.73:20-77:6.
166-168. FF 166-168 are inaccurate, incomplete, and misleading. The February 2013 Report
was not solely Hansen’s work. Supra ¶ 161. The Admissions Office worked with OIR to investigate
whether Harvard’s “admissions process disadvantage[s] Asians.” T3.45:19-49:13, 76:21-77:2; T5.87:620; T8.12:19-14:4; T7.208:7-15; P230; T7.208; P236.1; P238.1. For all its projects with the Admissions
Office, OIR “collaborated closely with Dean Bill Fitzsimmons and Director of Financial Aid Sally
Donahue to understand the climate and nuances of the data.” P288.1. The February 2013 Report
analyzed the “important variables” in the admissions process. T3.90:1-10. It investigated whether there
was “bias against Asians in college admissions” and whether “the admissions process disadvantage[d]

34

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 43 of 85

Asians.” P12.3, 38; T3.119:5-16. OIR created the report in response to an accusation of discrimination
against Asian Americans. SFFA-FF 115-117. OIR’s related work, “Admissions Part II,” P9, performed
various analyses to determine whether being Asian American was negatively associated with the
likelihood of admission, SFFA-FF 118; T8.46:17-47:11. OIR and Fitzsimmons believed the February
2013 Report was accurate, complete, and reliable. Supra ¶ 161.
169-170.

FF 169-170 are incomplete and misleading. OIR did not meet with Dean

Fitzsimmons to “discuss . . . the four models.” OIR met with Fitzsimmons to tell him that Asian
Americans were being disadvantaged in the admissions process and, in particular, that the personal
rating was penalizing Asian Americans. SFFA-FF 119-120; P12.3, 34-38; T3.79:7-86:23, 91:5-92:22.
171. FF 171 is inaccurate, incomplete, and misleading. OIR regularly labeled its work product
“preliminary” because “almost everything [OIR did] was up for discussion and review.” T4.213:5214:13. The February 2013 Report did “reflect input from the Admissions Office.” Supra ¶¶ 166-167.
OIR and Fitzsimmons believed the report was accurate, complete, and reliable. Supra ¶ 161. OIR was
confident enough in its report to suggest that it be shared with President Faust and Dean Smith,
P12.37; T3.94:16-23, and to share it with senior Harvard officials in Fall 2013 and Spring 2014,
T4.238:4-7, 238:22-240:21, 241:7-243:2; P35.1-7; T5.7:9-14:10; P41.47-66; P35, P36.
172-173. FF 172-173 are inaccurate, incomplete, and misleading. The February 2013 Report
was not “consistent” with OCR’s 1990 Report insofar as OCR cleared Harvard and OIR indicted it.
Supra ¶ 163. Nor is it credible that the purpose of OIR’s presentation was to show Dean Fitzsimons
that as “additional factors [are] considered in the process, the racial composition of the simulated class
more closely reflect[s] that of the actual class.” OIR met with Fitzsimmons to inform him that AsianAmerican applicants were being disadvantaged and, specifically, that the personal rating was penalizing
them. Supra ¶¶ 169-170.

35

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 44 of 85

174. FF 174 is incomplete and misleading. Hansen was not, as Harvard implies, incapable of
doing the work contained in the Admissions Part II Presentation. Hansen was fully capable of
performing logistic regressions, did high quality work, and received uniformly positive performance
evaluations. T8.10:4-12:3; T5.88:11-14; T4.215:10-216:1. No one has ever found any errors in
Admissions Part II. T5.95:3-12; T8.48:2-12. Hansen did not investigate these issues on his own
initiative. T8.13:19-23.
175. FF 175 is incomplete and misleading. Admissions Part II provided Fitzsimmons with
extensive statistical evidence that Harvard penalizes Asian-American applicants. SFFA-FF 118.
176-177. FF 176-77 are inaccurate, incomplete, and misleading. Fitzsimmons was shown
Admissions Part II before this litigation, as were Driver-Linn and Bever. SFFA-FF 118; T3.60:1861:10, P16. It is true that, after laying out the numerous ways Asian Americans were penalized vis-àvis whites, P9.5-12, OIR left blank the page entitled “Possible Explanations,” P9.14.
178. FF 178 is inaccurate, incomplete, and misleading. Fitzsimmons asked OIR to provide
him with “empirical proof” that Harvard gives a “tip” to low-income applicants. SFFA-FF 122-123;
T3.106:16-107:1, 125:19-127:2; T5.35:10-36:13. OIR told Fitzsimmons that “low income students
clearly receive a ‘tip’ in the admissions process.” P26.4; T3.111:12-16. Fitzsimmons understood that
“being low-income was positively associated with being admitted to Harvard.” T3.130:17-25, 131:24132:9, 106:8-107:1.
179-180. FF 179-180 are incomplete and misleading. OIR’s regression analysis showed “a
negative and statistically significant coefficient for Asian-American ethnicity.” FF 179; P26.4. That
meant that being Asian American was “negatively correlated” with being admitted. T4.226:1-21;
SFFA-FF 124-125; T3.134:5-135:8; T5.104:2-24. OIR specifically warned against “sharing these results
publicly” because their “descriptive analysis and regression models” show that “there are demographic
groups [i.e., Asian Americans] that have negative effects.” P26.4; T3.134:5-135:8. Although OIR

36

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 45 of 85

acknowledged that a “wider set of variables might result in a better fitting model,” it nevertheless
found that “[i]n spite of these limitations, the logistic regression model results are consistent with [its]
descriptive analysis.” P26.3. Driver-Linn, Bever, and Hansen did not have a “limited understanding
of the admission process,” as they “collaborated closely with Dean Bill Fitzsimmons and Director of
Financial Aid Sally Donahue to understand the climate and nuances of the data.” P288.1; supra ¶¶ 166167; T8.47:12-17. OIR and Fitzsimmons believed the May 1, 2013 Memorandum was accurate,
complete, and reliable. T5.12:15-14:10; T4.211:7-24, 238:4-7, T5.39:6-8, 96:1-17, 111:9-25; T3.102:24103:10, 147:14-17. OIR was confident enough in the memorandum to share it with the Dean of
Harvard College, Rakesh Khurana, in the spring of 2014. P41.59-66; T5.7:9-14:10.
181. FF 181 is accurate.
182. FF 182 is inaccurate, incomplete, and misleading. The cited evidence does not support
Harvard’s assertion that Fitzsimmons did not “understand the [May 1, 2013 Memorandum] to show
bias or discrimination against Asian-American applicants.” He understood from the memorandum
that being Asian American was “negatively associated” with being admitted to Harvard just like he
understood that being low income was positively associated with being admitted. There is no other
way to interpret OIR’s chart and OIR’s description of its findings. Supra ¶¶ 178-180; SFFA-FF 124;
T3.134:5-135:8, 138:25-139:9; P26.4.
183-184. FF 183-184 are inaccurate, incomplete, and misleading. It is not credible that the
May 30, 2013 Report “reassured” Fitzsimmons that Harvard was “treating Asian-Americans in an
evenhanded manner.” FF 184. The report prominently showed a negative coefficient for being Asian
American, P28.7-8, meaning there was a “negative chance of getting into Harvard by virtue of being
Asian,” SFFA-FF 126; T3.143:7-11.
185-187. FF 185-187 are inaccurate, incomplete, and misleading. OIR sent Khurana both the
May 1, 2013 Memorandum and an excerpt of the February 2013 Report discussing Asian-American

37

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 46 of 85

bias. P41.47-66; T5.7:9-12:14. Khurana is not an expert in statistical analysis and is not more qualified
than OIR to conduct statistical modeling. T7.48:12-49:23, 69:16-70:7. Khurana took no steps to
determine why OIR was investigating Asian-American bias or what an “appropriate” statistical analysis
would show. T7.70:8-15.
188. FF 188 is inaccurate, incomplete, and misleading. The cited evidence does not support
Harvard’s assertion that OIR analyzed “only a few” factors. All of OIR’s work was accurate, complete,
and reliable. Supra ¶¶ 161, 179-180. OIR and Fitzsimmons knew that the personal rating penalizes
Asian Americans, supra ¶¶ 161, 163, 165, 169-170, 176-177; SFFA-FF 118-120, and that there is a
statistically significant negative correlation between being Asian American and being admitted, supra
¶¶ 179-180, 182-184, SFFA-FF 118, 123-126.
189. FF 189 is inaccurate, incomplete, and misleading. SFFA demonstrated that Harvard’s
discrimination against Asian Americans led to the rejection of hundreds of applicants. SFFA-FF 62.
SFFA also identified a litany of Asian-American applicants who were victims of racial stereotyping.
SFFA-FF 110-111. SFFA presented the files of applicants who had been victimized by Harvard’s racial
stereotyping to highlight the discrimination that Asian Americans pervasively face. T9.136-142.
190. FF 190 is incomplete and misleading. Harvard does not dispute that SFFA’s standing
members are Asian American, applied to Harvard, and were rejected. Doc. 324 at 14; Doc. 566 at 1922. SFFA is not required to offer evidence concerning its members. Infra ¶¶ 275-276.
191. FF 191 is inaccurate. SFFA-FF 109.
192. FF 192 is inaccurate. There are numerous examples of stereotyping in the docket binder.
SFFA-FF 108. These descriptions reflect stereotypes of Asian Americans and far outnumber those
used with applicants of other races. SFFA-FF 94-95, 107-108.
193. FF 193 is incorrect, incomplete, and misleading. OCR’s review did find evidence of racial
stereotyping against Asian Americans. Supra ¶ 155. And SFFA did not engage in a “cursory review”

38

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 47 of 85

of Harvard’s admissions system. SFFA closely reviewed the limited (and partially cherry-picked)
sample of application files that Harvard was ordered to produce and found evidence of discrimination
against Asian Americans. Supra ¶ 189.
194-198. FF 194-198 are inaccurate. Harvard’s newly minted theory for its disparate treatment
of Asian Americans in Sparse Country is not supported by any trial testimony. No Harvard witness
ever gave this explanation or claimed that the lower threshold for Asian Americans in Sparse Country
was an oversight. T1.143:14-153:8. Dean Fitzsimmons vigorously defended this disparate treatment
at trial. SFFA-FF 106. P50 indicates that Harvard imposed a lower recruiting threshold for whites in
Sparse Country for the Class of 2016. See P50.4. Regardless, there is still a “consistent . . . pattern of
differential recruitment-letter thresholds between White and Asian-American applicants.” FF 198.
Harvard admits it imposed a higher standard on Asian-Americans than whites in Sparse Country and
it has never claimed that it was a mistake. FF 195-196; SFFA-FF 106. This disparate treatment reflects
a desire to have fewer Asian-American applicants. SFFA-FF 105-106.
199-201. FF 199-201 are inaccurate. Admissions officers’ testimony that they never employed
quotas, ceilings, floors, or targets based on race or ethnicity was not credible. As was their testimony
that the admissions process is “free of discrimination and bias” and that race is “never a negative
factor.” Race is a negative factor for Asian Americans. SFFA-FF 21-133, 134-140. The testimony of
Harvard’s witnesses was not consistent, credible, or unrebutted. SFFA-FF 129-133.
202. FF 202 is incomplete and misleading. Harvard has employed race-neutral “practices” to
pursue diversity, but it has never employed race-neutral “alternatives” in order to achieve student body
diversity. SFFA-FF 153.
203-204. FF 203-204 are inaccurate, incomplete and misleading. Before this lawsuit, Harvard
had never examined the availability of race-neutral alternatives. SFFA-FF 153; T14.214:3-6. The Ryan
Committee was the first committee Harvard ever formed to analyze the availability of race-neutral

39

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 48 of 85

alternatives, T14.214:3-6, and it failed to conduct any analysis, SFFA-FF 154; see T5.80:12-15. The
Ryan Committee was formed in May 2014, after the launch of Harvard Not Fair. SFFA-FF 154; P340;
D12.1. The committee did not “pause” its work—the committee was “disbanded.” SFFA-FF 154;
T2.70:2-15; T14.217:23-218:1. It was charged with analyzing the availability of race-neutral alternatives
at Harvard College, D12.3, and it was fully capable of doing so, T7.77:12-78:24.
205-206. FF 205-206 are inaccurate, incomplete, and misleading. “The College” did not
“convene[] its own committees.” President Faust, the President of Harvard University, convened the
Khurana Committee, T7.34:12-35.3, and the Smith Committee. T14.220:4-14. She also convened (and
then disbanded) the Ryan Committee. SFFA-FF 154; T14.218:2-6.
207-210. FF 207-210 are inaccurate. It was not “reasonable” to have Fitzsimmons, Khurana,
and Smith alone evaluate the availability of race-neutral alternatives. The Smith Committee’s findings
were never in doubt. The Smith Committee was reviewing the exact issue that Harvard was litigating
(and facing liability for under Title VI). SFFA-FF 155. It is not credible that the Smith Committee
would have made findings that conflicted with Harvard’s experts or undermined Harvard’s litigation
position, especially given the involvement of Harvard’s attorneys. Id. Moreover, Fitzsimmons had
been the Dean of Admissions since 1986, T1.124:5-15, and Smith had been overseeing the Admissions
Office for more than a decade, T7.106:2-22; T2.33:8-10; T14.220:25-221:5, 221:17-20. It is not credible
that they would ever have concluded that they had caused Harvard to violate Title VI. SFFA-FF 155.
211. FF 211 is incomplete and misleading. The Smith Committee refused to identify any goal
for measuring when Harvard would reach a level of minority enrollment sufficient to achieve the
educational benefits of diversity. SFFA-FF 150. The Committee acknowledged that Kahlenberg’s
simulations produced “comparable” racial diversity with no objection that Hispanic admit shares rose
by 4 percentage points while African-American shares declined by a corresponding amount to 10
percent. P316. At trial, however, Dean Smith testified that any race-neutral alternative in which

40

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 49 of 85

Harvard “moved backwards” from the number of African Americans it admitted the prior year was
“a bridge too far.” T7.140:23-141:11.
212. FF 212 is inaccurate. The academic index is an appropriate way to sort applicants, supra
¶ 162, and academic characteristics remain superb under SFFA’s alternatives, infra ¶¶ 228-231, 250.
213. FF 213 is inaccurate and incomplete. Harvard’s lawyers—not the committee—drafted
the Smith Committee’s report. SFFA-FF 155; T7.65-66.
214. FF 214 is inaccurate. The Smith Committee was not “informed” by Professor Card’s
analysis. The committee relied almost entirely on Card’s expert reports; it did not collect data, take
testimony, speak with outside experts, or run simulations. P316; T7.76:25-77:18. It neither disagreed
with any of Professor Card’s findings nor accepted any of Kahlenberg’s findings. P316. The Smith
Committee’s conclusions were never in doubt. Supra ¶¶ 207-210.
215-218. FF 215-218 are incomplete and misleading. SFFA is not proposing that Harvard
discontinue using race without implementing race-neutral alternatives. SFFA-FF 156-158; T6.18:7-11.
SFFA agrees that the Smith Committee found that “the number of African-American and Hispanic
students on campus would decline dramatically” if Harvard eliminated racial preferences and did not
replace them with race-neutral alternatives. P316.8; T13.127:1-128:15.
219-220. FF 219 and 220 are incomplete and misleading. Harvard can achieve the educational
benefits of diversity without eliminating early action. SFFA-FF 157-158; PD32 (Simulations B and D).
221-222. FF 221-222 are incomplete and misleading. Harvard can achieve the educational
benefits of diversity without eliminating the recruited athlete preference. SFFA-FF 157-158; PD32
(Simulations A, C, and D); T6.35:10-22. SFFA never proposed that Harvard eliminate LDC
preferences without also implementing race-neutral alternatives. SFFA-FF 156-158.
223. FF 223 is inaccurate, incomplete, and misleading. LDC preferences do not “serve
important institutional interests.” They overwhelmingly benefit white applicants: 70% of legacy

41

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 50 of 85

students are white, 68% of those on the Dean/Director’s List are white, and 54% of the children of
faculty and staff are white. D746; P634. By contrast, only 4% of legacy students are African American,
3% of those on the Dean/Director’s List are African American, and 1% of the children of faculty and
staff are African American. Id. Giving preferences to these groups decreases racial diversity. Id.;
T6.34:17-35:9. These preferences also overwhelmingly benefit wealthy applicants and decrease
socioeconomic diversity. SFFA-FF 14; 34:17-35:9. The Z-List (or granting “deferred admission”) also
benefits wealthy and well-connected applicants. SFFA-FF 20. There is no empirical evidence that
Harvard would receive fewer donations or that fewer alumni will volunteer their time if Harvard
stopped giving these preferences. T6.23:8-17; T12.36:7-39:9, 68:2-70:23; P316.16-17. It is not credible
that Harvard alumni donate money and volunteer time because they expect to secure benefits for their
children. T3.211:3-8. It also is not credible that individuals will turn down the opportunity to work or
teach at Harvard—one of the premier universities in the world—without the promise of an admissions
preference for their children. T12.39:9-40:11; P316.17.
224. FF 224 is inaccurate. Harvard can achieve the educational benefits of diversity by
eliminating LDC preferences and by increasing socioeconomic preferences. SFFA-FF 157-158.
225-227. FF 225-227 are incomplete and misleading. SFFA is not proposing that Harvard
discontinue the use of standardized tests, T6.134:2-9, and Harvard can achieve the educational benefits
of diversity without doing so, SFFA-FF 157-158.
228-231. FF 228-231 are incorrect, incomplete and misleading. Card’s Simulation 4x (what
Kahlenberg called “Simulation B”) is one of at least four viable race-neutral alternatives that SFFA
identified. PD29; T6.41:3-7, 119:23-25. Under this simulation, African Americans and Hispanics
would make up about the same share of the class (27%) as they do now (28%), while Asian Americans
would rise from 24% of the class to 30% of the class. T6.41:3-43:2; PD25; PD29. Professor Card’s 4x
SES boost would not “overwhelm other considerations in the admissions process.” It is half the size

42

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 51 of 85

of the preference Harvard gives to recruited athletes. T6.41:15-19. Nor would it “depriv[e] Harvard
of the ability to select the most excellent applicants.” Academic characteristics would remain superb,
with high school GPA improving slightly and SAT scores falling only slightly (from the 99th percentile
to the 98th percentile). T6.41:3-43:2; PD29. The decline in Harvard’s ratings is minimal, and should
be considered in context given the substantial increase in socioeconomic diversity. T6.141:4-9; infra
¶ 250. This boost likewise would not “reduce the socioeconomic diversity of African-American and
Hispanic students.” Today, roughly 70% of African Americans and Hispanics are considered
“advantaged.” T6.54:2-56:2. This approach would provide “a much better approximation of the real
world,” id., by causing socioeconomic diversity to improve greatly. The advantaged/disadvantaged
ratio would go from 82%/18% to 48%/52%. T6.41:3-43:2, 54:2-56:2; PD 25; PD29.
232. FF 232 is inaccurate, misleading, and incomplete. Harvard need not discontinue the
athlete preference or the use of standardized tests, supra ¶¶ 221-222, 225-227, and Harvard can achieve
the educational benefits of diversity without doing so, SFFA-FF 157-158.
233-234. FF 233-234 are incomplete and misleading. Although Harvard could do a better job
recruiting low-income applicants, it can achieve student body diversity without improved recruiting.
SFFA-FF 157-158; T6.58:21-59:8.
235. FF 232 is inaccurate, misleading, and incomplete. Harvard can achieve the educational
benefits of diversity without improved recruiting or eliminating the athlete preference, or the use of
standardized tests. SFFA-FF 157-158; supra ¶¶ 219-220, 233-234.
236. FF 236 is inaccurate, incomplete, and misleading. Harvard could expand its partnership
“with organizations that assist underserved high school students,” but it can achieve the educational
benefits of diversity without doing so. SFFA-FF 157-158.

43

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 52 of 85

237. FF 237 is inaccurate, incomplete, and misleading. The cited evidence does not support
Harvard’s assertion that it has “one of the nation’s most generous financial aid programs.” There is
no barrier to increasing the number of non-wealthy applicants that Harvard admits. Supra ¶¶ 25-27.
238. FF 238 is incomplete and misleading. Harvard can increase diversity through increased
admission of community college transfers. Professor Card’s point about the relative qualifications and
diversity of transfers relies upon data on transfers from all “colleges and universities”—not from
community colleges—and is inapposite. DX730 & DD10.124-125; T13.146:24-149:21. In any event,
Harvard can achieve student body diversity without increasing transfer admissions from community
colleges. Supra ¶¶ 219-220.
239. FF 239 is incomplete and misleading. SFFA is not proposing any quota system that
“admit[s] top students by ZIP code or high school.” T6.107:6-108:25. Nor would Harvard need to
implement one to achieve the educational benefits of diversity. SFFA-FF 157-158.
240-241. FF 240-241 are incomplete and misleading. Kahlenberg identified a viable raceneutral alternative (Simulation A) that relied on the College Board neighborhood clusters. PD27;
T6.33:22-40:22, 109:8-12. The decline in Harvard’s ratings would be minimal. Infra ¶ 250. Using the
College Board neighborhood clusters is a race-neutral approach. T6.112:15-113:5. Kahlenberg also
identified three viable race-neutral alternatives (Simulations B, C, and D) that do not rely on the
College Board neighborhood clusters. T6.41:20-42:1, 43:7-44:16, 46:512, 109:8-12; PD32.
242. FF 242 is inaccurate, misleading, and incomplete. Harvard can achieve student body
diversity without eliminating the athlete preference or the use of standardized tests, supra ¶¶ 221-222,
225-227, without increasing its recruiting efforts, supra ¶¶ 233-234, and without using the College
Board neighborhood clusters, supra ¶ 239-241.
243. FF 243 is inaccurate. Harvard did not consider Kahlenberg’s proposals in good faith.
The Smith Committee’s conclusions were never in doubt. Supra ¶¶ 207-201, 214.

44

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 53 of 85

244-245. FF 244-245 are inaccurate, incomplete, and misleading. Kahlenberg does not
support “eliminat[ing] affirmative action in all programs that consider race.” T6.23:18-24:12. He
supports the use of race in admissions when there are no viable race-neutral alternatives available. Id.
It is true that he has long supported using socioeconomic preferences as an alternative to the use of
race. That is because (1) as a matter of “basic fairness,” if someone has “overcome obstacles in life,”
then it is “worth considering that in the application process”; (2) there are immense benefits of having
socioeconomic diversity, where students who have faced economic hardships can educate and learn
from others; (3) “there are costs to using race in deciding who gets ahead,” including “increased
resentment” and “stigma”; and (4) because “poll after poll after poll has suggested that the vast
majority of Americans are uncomfortable or opposed to the idea that race should be a factor in college
admissions,” whereas “large majorities of Americans support the notion of providing a preference to
economically disadvantaged students.” T6.20:19-22:20. Kahlenberg has spent decades studying and
writing about race-neutral alternatives, has served as a consultant to multiple school systems, and has
testified as an expert on issues of desegregation and socioeconomic integration. T6.7:15-12:10.
246. FF 246 is incomplete, inaccurate, and misleading. Although Kahlenberg acknowledged
that Dean Fitzsimmons has expressed a personal commitment to socioeconomic diversity, the
evidence in this case demonstrates that socioeconomic diversity at Harvard is “deeply lacking.”
SFFA-FF 143. To the point, there are 23 times as many wealthy students on campus as poor students.
T6.17:16-22; see also T7.10:9-14:19.
247. FF 247 is incomplete and misleading. Given Kahlenberg’s decades of work in the field,
he was confident that Harvard could “produce considerable racial and ethnic diversity without
resorting to racial preferences.” T6.70:19-23; SFFA-FF 156. But he did not opine “before any
discovery or expert analysis” that Harvard could employ race-neutral alternatives “without

45

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 54 of 85

compromising the excellence of the institution.” His testimony is based on his objective assessment
of the evidence produced in discovery. T6.18:1-11.
248. FF 248 is inaccurate, incomplete, and misleading. SAT scores and GPAs do not have
“limited value in identifying applicants with strong academic potential.” SATs and GPAs (as every
student and parent knows) are “the standard indicators of academic strength or academic preparation
that are widely used in the college admissions process.” T6.31:10-15. They are also the most “objective
factors” and “the standard in the literature on race-neutral alternatives.” T6.52:11-22; supra ¶ 212.
249. FF 249 is inaccurate, incomplete, and misleading. Dean Smith’s testimony conflicts with
his own committee’s findings. Supra ¶ 211. Moreover, he offered no reasoned explanation for why a
decline in the African American share of the class from 14 percent to 10 percent rendered the raceneutral alternatives unworkable. T7.154:24-156:7. He never explained why a 10% African-American
share was acceptable to Harvard as recently as the class of 2016. See D713. The committee’s only
guidepost was that it did not want to “move backwards from where we are today” because this is
“already an issue for us to be dealing with.” T6.140:16-141:11. Nor did the committee address
Kahlenberg’s finding that the use of wealth as a factor could boost African American representation.
P316. SFFA’s proposals increase the Hispanic and Asian-American shares of the class and increase
socioeconomic diversity. PD29, PD31, PD33; SFFA-FF 157-158.
250. FF 250 is inaccurate, incomplete, and misleading. SFFA’s proposals do not entail
“unacceptable declines in the quality of the admitted class.” Academic characteristics would remain
superb. Supra ¶¶ 228-231. That the share with an academic rating of a 1 or a 2 would decline from
76% to 63% (Simulation D) or to 61% (Simulation C) is not an “unacceptable decline.” The proposals
could be adjusted so that Harvard accepted every applicant with an academic rating of 1 and the overall
levels of racial and socioeconomic diversity would not change. The slight decrease in the academic
rating of 2 would have minimal (if any) effect. T6.52:11-54:1.

46

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 55 of 85

251. FF 251 is inaccurate. Harvard can achieve the educational benefits of diversity without
the use of race. SFFA-FF 157-158.
252-253. FF 252-253 are incomplete and misleading. Harvard has not established a date when
its use of race will sunset. SFFA-FF 152. Harvard did not consider SFFA’s proposals in good faith.
The Smith Committee’s conclusions were never in doubt. Supra ¶ 207-210, 214. It is not credible that
SFFA’s race-neutral alternatives—which increase racial and socioeconomic diversity and maintain
Harvard’s academic excellence—will cause enormous “harm . . . to Harvard College’s educational
program, to the student experience on [Harvard’s] campus, and to the institution’s mission and goals.”
254. FF 254 is inaccurate. The current and former students and President Simmons did not
“corroborate the committee’s judgment.” None of them supported the committee’s rejection of the
results of SFFA’s race-neutral alternatives—i.e., that Harvard would be deprived of the educational
benefits of diversity with a class that is more racially and socioeconomically diverse and maintains
academic excellence. SFFA-FF 157-158.
255. FF 255 is inaccurate. Harvard can “achieve its educational and institutional objectives
without considering race as one factor in its admissions process.” SFFA-FF 157-158.
256-259. FF 256-259 are incomplete and misleading. The Court has found that SFFA is a
voluntary membership association with standing to bring this action on behalf of its members. Doc.
324 at 2-3, 6-7, 13, 17; Doc. 566 at 18-22.
RESPONSE TO HARVARD’S PROPOSED CONCLUSIONS OF LAW
A.

Harvard Intentionally Discriminates Against Asian-American Applicants.

260-61. A recipient of federal funds violates Title VI if the same conduct by a state actor
would violate the Equal Protection Clause. Gratz v. Bollinger, 539 U.S. 244, 276 n.23 (2003); CL 294
n.2 (“Harvard assumes . . . that Title VI incorporates the Supreme Court’s equal protection
standard.”). The Equal Protection Clause subjects intentional racial discrimination to strict scrutiny.
When a challenged policy “uses explicit racial classifications,” strict scrutiny is “automatic” because
47

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 56 of 85

“discriminatory intent is presumed.” Anderson ex rel. Dowd v. City of Boston, 375 F.3d 71, 82 (1st Cir.
2004); accord Jana-Rock Const., Inc. v. N.Y. State Dep’t of Econ. Dev., 438 F.3d 195, 204-05 (2d Cir. 2006)
(“When the [policy] expressly classifies persons on the basis of race or national origin, . . . [the]
plaintiff . . . need not make an extrinsic showing of discriminatory animus or a discriminatory effect
to trigger strict scrutiny.”). Harvard explicitly classifies applicants based on race and, accordingly, must
satisfy strict scrutiny. Fisher v. Univ. of Tex. at Austin (Fisher I), 570 U.S. 297, 307-10 (2013).
Harvard fails strict scrutiny if it intentionally discriminates against Asian Americans in favor
of whites. Harvard does not argue that it has a compelling interest in favoring whites over Asian
Americans—a racial minority that has suffered immense discrimination in this country. Nor could it:
“the racial discrimination which it was the object of the Fourteenth Amendment to eliminate” is never
a “permissible state objective.” Loving v. Virginia, 388 U.S. 1, 11 (1967). There is likewise no compelling
interest in racial stereotyping. J.E.B. v. Ala. ex rel. T.B., 511 U.S. 127, 139 n.11 (1994) (“[C]lassifications
that rest on impermissible stereotypes violate the Equal Protection Clause, even when some statistical
support can be conjured up for the generalization.”); City of Cleburne v. Cleburne Living Ctr., 473 U.S.
432, 448 (1985) (Differential treatment based on “mere negative attitudes” about a group violates the
Equal Protection Clause.); Ala. Legislative Black Caucus v. Alabama, 231 F. Supp. 3d 1026, 1107 (M.D.
Ala. 2017) (“[S]trict scrutiny forbids the use of racial stereotypes.”).
Harvard is correct that it violates Title VI if intentional discrimination was “a … motivating
factor” for its actions. Goodman v. Bowdoin Coll., 380 F.3d 33, 43 (1st Cir. 2004) (emphasis added). Thus,
racial discrimination does not need to be Harvard’s “sole[]” purpose, or even its “‘dominant’ or
‘primary’ one.” Vill. of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 265-66 (1977). Harvard
is incorrect, however, that its discrimination must be a product of “racial animus toward” AsianAmerican applicants. Infra ¶¶ 284-285.

48

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 57 of 85

262. Because the parties agree that strict scrutiny governs this case, and because intentional
discrimination violates strict scrutiny, Harvard has the burden to prove that it does not intentionally
discriminate against Asian Americans. Jana-Rock, 438 F.3d at 204-05 (“When the [defendant] expressly
classifies persons on the basis of race or national origin, … [t]he burden of proof shifts, strict scrutiny
applies, and … [the defendant] has the burden.”); e.g., Cooper v. Harris, 137 S. Ct. 1455, 1464 (2017).
Strict scrutiny flips the ordinary burdens in order to “smoke out” the use of “racial prejudice or
stereotype.” Adarand Constructors, Inc. v. Pena, 515 U.S. 200, 225-26 (1995).
Harvard cites three Title VII cases for the proposition that the plaintiff always bears the
burden to prove intentional discrimination. But those cases involved policies that were facially neutral.
Ahern v. Shinseki, 629 F.3d 49, 59 (1st Cir. 2010); Udo v. Tomes, 54 F.3d 9, 13 (1st Cir. 1995); White v.
Vathally, 732 F.2d 1037, 1042 (1st Cir. 1984). Harvard’s admissions program is not facially neutral; it
expressly considers the race of each applicant. SFFA-CL 165; supra ¶¶ 59-60. As Harvard’s amici point
out, Harvard even considers the race of Asian-American applicants. Doc. 623 at 9 ¶ 4, 38 ¶¶ 92-94.
Because Harvard uses racial classifications—which are “pernicious”—it must prove that it uses this
“highly suspect tool” without engaging in “prejudice or stereotype.” Adarand, 515 U.S. at 226. Harvard
cites no case where the defendant classified individuals based on race but the court nevertheless placed
the burden of proof on the plaintiff. That is because none exists. But even if this were a challenge to
a facially neutral policy, SFFA would only need to prove intentional discrimination by a preponderance
of the evidence. SFFA-CL 167. It has carried that burden. SFFA-CL 167-206; infra ¶ 292.
263. SFFA agrees that, once the Court identifies the correct standard of review, the presence
or absence of discriminatory intent is a question of fact. But “the court, not the parties, determines
the correct standard of review.” United States v. Fisher, 624 F.3d 713, 719-20 (5th Cir. 2010).
264. Harvard’s admissions system is not “facially neutral.” Supra ¶ 262. But SFFA agrees that
“Arlington Heights . . . la[ys] out a nonexhaustive list of factors that could constitute circumstantial

49

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 58 of 85

evidence” of intentional discrimination. Jana-Rock, 438 F.3d at 212; Arlington Heights, 429 U.S. at 268
(“The foregoing summary identifies, without purporting to be exhaustive, subjects of proper inquiry
in determining whether racially discriminatory intent existed.”).
265-67. SFFA does not point to “racially disproportionate outcomes” to prove that Harvard
is engaging in a pattern-or-practice of discrimination. SFFA’s regression analysis shows an AsianAmerican penalty—not bare racial disparities. SFFA-CL 38-93; infra ¶¶ 269-270. Courts “accept[]
statistics in the form of multiple-regression analysis to prove statutory violations” in pattern-orpractice cases. McCleskey v. Kemp, 481 U.S. 279, 294 (1987) (citing Bazemore v. Friday, 478 U.S. 385, 40001 (1986)). Pattern-or-practice cases allege intentional discrimination—not disparate impact. Lujan v.
Franklin Cty. Bd. of Educ., 766 F.2d 917, 929 (6th Cir. 1985); Latinos Unidos De Chelsea En Accion
(LUCHA) v. Sec’y of HUD, 799 F.2d 774, 795 n.21 (1st Cir. 1986). They challenge decisions “involv[ing]
a number of relevant variables” that are “to a great extent uniform for all [applicants],” and that “over
time are fairly attributable to the [defendant].” McCleskey, 481 U.S. at 295 n.14. “Therefore, an
unexplained statistical discrepancy can be said to indicate a consistent policy of the decisionmaker.”
Id. at 295 n.15. This is indisputably true of university admissions.
Under the pattern-or-practice rubric, a statistically significant disparity (at the 0.05 p-level) is
sufficient “alone” to prove a prima facie case of intentional discrimination. SFFA-CL 173. (And a
prima facie case that is inadequately rebutted, of course, conclusively proves intentional discrimination.
Castaneda v. Partida, 430 U.S. 482, 501 (1977)). Courts have “rejected the proposition that to establish
intentional discrimination a showing of ‘gross disparities,’ rather than mere statistically significant
disparities, is required.” Berger v. Iron Workers Reinforced Rodmen Local 201, 843 F.2d 1395, 1413 (D.C.
Cir. 1988). While the Supreme Court once used the word “gross,” Hazelwood Sch. Dist. v. United States,
433 U.S. 299, 307 (1977), it did so to describe the “general rule” that a statistical disparity of “two or
three standard deviations” is enough to prove intentional discrimination, id. at 308 n.14 (quoting

50

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 59 of 85

Castaneda, 430 U.S. at 497 n.17). “[T]wo standard deviations” is the mathematical equivalent of
“statistical significance at the .05 level.” Palmer v. Shultz, 815 F.2d 84, 92 (D.C. Cir. 1987). Perhaps
statistical significance would not suffice if the data broadly compared the defendant’s “workforce”
with the “general population.” Segar v. Smith, 738 F.2d 1249, 1278 (D.C. Cir. 1984). But statistical
significance easily suffices when the “statistical evidence is more finely tuned to the relevant … pool.”
Id. There is no more finely tuned pool than the one SFFA uses here—actual applicants to Harvard.
Mister v. Ill. Cent. Gulf R. Co., 832 F.2d 1427, 1435-36 (7th Cir. 1987).
Further, Harvard’s claim that statistical disparities are “rare[ly] sufficient on their own” relies
on cases involving race-neutral policies. Arlington Heights, 429 U.S. at 266 (legislation “neutral on its
face”); McGuire v. Reilly, 386 F.3d 45, 64 (1st Cir. 2004) (“law that is facially neutral”); McCleskey, 481
U.S. at 284 n.3 (facially race-neutral aggravators); Spath v. NCAA, 728 F.2d 25, 28 (1st Cir. 1984)
(“facially neutral” bylaw). While a plaintiff might need a powerful showing to prove through statistics
alone that “a statute, otherwise neutral on its face, [is being] applied so as invidiously to discriminate
on the basis of race,” intentional discrimination “may be proved as well by [a significant statistical
disparity] combined with … racially non-neutral selection procedures.” Washington v. Davis, 426 U.S.
229, 241 (1976). Statistics, in other words, are not the only evidence in such cases: “a selection
procedure that is susceptible of abuse or is not racially neutral supports the presumption of
discrimination raised by the statistical showing.” Castaneda, 430 U.S. at 494. So it is here. Harvard’s
admissions process is not race neutral. Supra ¶ 262. And Harvard’s process also is “susceptible of
abuse” because—as Harvard concedes (at FF 148)—it is “highly subjective.” Castaneda, 430 U.S. at
497; SFFA-FF 98-99; SFFA-CL 171, 174, 180, 190-91.
268. SFFA has never shied away from the fact that statistics are a central feature of this case.
In pattern-or-practice cases, statistics are always “central” and constitute “the core” of the evidence.
SFFA-CL 172; accord Lujan, 766 F.2d at 929 (“numerical disparities” are “the key” in pattern-or-

51

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 60 of 85

practice cases); Robinson v. Metro-N. Commuter R.R. Co., 267 F.3d 147, 158 n.5 (2d Cir. 2001) (“The
heavy reliance on statistical evidence in a pattern-or-practice disparate treatment claim distinguishes
such a claim from an individual disparate treatment claim.”). SFFA’s reliance on statistics to prove
Harvard’s discrimination is a feature—not a bug:
[G]eneral statistical data [proving a pattern of discrimination] . . . is particularly helpful
in the academic context, where the [admission] decision is highly subjective . . . . Often,
there is little direct evidence . . . that there was discrimination in an individual
[admission] decision. Despite such problems of proof, Congress entrusted to the
courts the responsibility of providing a forum for the litigation of claims of
discrimination in universities . . . . [S]tatistical data . . . is an effective method of proof,
and, accordingly, an effective method by which to discharge our responsibility . . . .
Lynn v. Regents of Univ. of Cal., 656 F.2d 1337, 1342 and n.3 (9th Cir. 1981).
While pattern-or-practice cases are typically brought under Title VII, Harvard does not dispute
that SFFA can bring this theory under Title VI. Rightly so. “A pattern and practice claim is not a
freestanding cause of action but merely a method of pro[of],” Aguilar v. ICE, 510 F.3d 1, 16 (1st Cir.
2007), so there is no logical reason why it would be available under Title VII but not other statutes
that prohibit intentional discrimination. See, e.g., United States ex rel. Williams v. City of Brockton, No. 12CV-12193-IT, 2016 WL 7429176, at *3-7 (D. Mass. Dec. 23, 2016) (recognizing a pattern-or-practice
claim under Title VI); Burgett v. Kan. City Bd. of Police Comm’rs, 613 F. App’x 553, 555 (8th Cir. 2015)
(same); Nieves-Márquez v. Puerto Rico, 353 F.3d 108, 126 & nn.20-21 (1st Cir. 2003) (recognizing a
pattern-or-practice claim under the ADA and Rehabilitation Act). Indeed, Title VI and Title VII were
enacted together and use virtually identical language. Due to these “similarities,” courts “frequently
have looked to Title VII in determining rights and procedures available under Title VI.” Smith v. Barton,
914 F.2d 1330, 1336 (9th Cir. 1990); e.g., Bryant v. Indep. Sch. Dist. No. I-38 of Garvin Cty., 334 F.3d 928,
930 n.1 (10th Cir. 2003) (“Courts often use [the] Title VII proof scheme for Title VI claims.”); Blunt
v. Lower Merion Sch. Dist., 767 F.3d 247, 276 n.45 (3d Cir. 2014) (“[T]he general discussion of the
usefulness of statistics … in Title VII cases is instructive [in Title VI cases].”); Lipsett v. Univ. of P.R.,

52

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 61 of 85

864 F.2d 881, 897 (1st Cir. 1988) (“[T]he disparate treatment standard of Title VII applies as well to
claims arising under the equal protection clause.”).
269-270. Professor Arcidiacono’s work reveals statistically significant discrimination against
Asian Americans, and Professor Card’s contrary analysis is not reliable. SFFA-FF 21-93; supra ¶¶ 94151; see generally Doc 624, Keane Amicus Brief. That is likely why some of Harvard’s own amici no
longer dispute that Harvard discriminates against Asian Americans, even though they did before trial.
Compare Doc. 623, with Doc. 471 at 3, 24-27.
271-73. Harvard claims that, even if Professor Arcidiacono’s statistical analysis is reliable, it
does not amount to competent evidence of intentional discrimination. To support this bold assertion,
Harvard makes three sweeping (almost Luddite) criticisms of regression models. It is no exaggeration
to say that, if Harvard’s criticisms were accepted, discrimination law as we know it would cease to
exist. Int’l Bhd. of Teamsters v. United States, 431 U.S. 324, 339 (1977) (“[O]ur cases make it unmistakably
clear that ‘statistical analyses have served and will continue to serve an important role’ in cases in
which the existence of discrimination is a disputed issue.”); Fed. Judicial Ctr., Reference Manual on
Scientific Evidence 306 n.5 (3d ed. 2011) (“Discrimination cases using multiple regression analysis are
legion.”). “There is no presumption that statistical evidence has no useful role to play in [intentional]
discrimination cases—indeed, we are hard pressed to see how anyone could take such a position
consistently with the Supreme Court’s guidance.” Adams v. Ameritech Servs., Inc., 231 F.3d 414, 424 (7th
Cir. 2000). Harvard disagrees (though hypocritically, given that it praises regression analyses whenever
they reach results Harvard likes, SFFA-CL 182).

First, Harvard contends that regression models can never prove causation—only correlation.
CL 272; FF 147. This argument ignores the role of statistical significance. “When statisticians find a
disparity between racial groups to be statistically significant [at the .05 p-level], they are willing to reject
the hypothesis that members of the groups truly had an equal chance of receiving the outcome at

53

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 62 of 85

issue.” Jones v. City of Boston, 752 F.3d 38, 43 (1st Cir. 2014). So are courts. Id. at 47 n.9 (collecting
cases); Fudge v. City of Providence Fire Dep’t, 766 F.2d 650, 658 (1st Cir. 1985) (because regression analysis
is “[w]idely accepted” a finding of statistical significance means “it can then be presumed that [the]
difference … is attributable to discriminatory bias”). In short, “regression analysis is a well recognized
and scientifically valid approach to understanding statistical data, and courts have long permitted
parties to use statistical data to establish causal relationships.” In re Neurontin Mktg. & Sales Practices
Litig., 712 F.3d 21, 42 (1st Cir. 2013).

Second, Harvard argues that regression models can never prove discriminatory intent because
the numbers do not reveal what was actually in the defendant’s head. CL 272-73. This is nothing more
than a criticism of circumstantial evidence of discriminatory intent. But plaintiffs are not “required to
submit direct evidence of discriminatory intent,” since “[t]here will seldom be ‘eyewitness’ testimony
as to the [defendant’s] mental processes.” U.S. Postal Serv. Bd. of Governors v. Aikens, 460 U.S. 711, 71617 (1983). “Circumstantial evidence is not only sufficient,” “but may also be more certain, satisfying
and persuasive than direct evidence.” Desert Palace, Inc. v. Costa, 539 U.S. 90, 100 (2003); SFFA-CL 170.
And statistical evidence is an especially persuasive form of circumstantial evidence. SFFA-CL 171.
Statistics are “[t]he ‘important starting point’ for assessing discriminatory intent” because “people
usually intend the natural consequences of their actions.” Reno v. Bossier Parish Sch. Bd., 520 U.S. 471,
487, 489 (1997). “Statistics showing racial or ethnic imbalance” are particularly “probative” because
such imbalances are “a telltale sign of purposeful discrimination.” Teamsters, 431 U.S. at 339 n.20. “In
cases concerning racial discrimination, ‘statistics often tell much and Courts listen.’” Parham v. Sw. Bell
Tel. Co., 433 F.2d 421, 426 (8th Cir. 1970).

Third, Harvard claims that regression models are unreliable because they “may be biased” by
the absence of omitted variables. CL 272; FF 122, 148. But the notion that a regression analysis is not
reliable unless it “include[s] ‘all measurable variables’” is “plainly incorrect.” Bazemore, 478 U.S. at 400.

54

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 63 of 85

“[T]he law does not require th[is] near-impossible standard,” Hemmings v. Tidyman’s Inc., 285 F.3d 1174,
1188 (9th Cir. 2002), and science counsels against it, FHFA v. Nomura Holding Am., Inc., 104 F. Supp.
3d 441, 503 (S.D.N.Y. 2015) (“‘[P]erforming a multiple regression analysis requires selecting only some
of the multitude of … [potential] predictors because including too many variables can preclude
measurement of the characteristics that are valid predictors.’”). Thus, Harvard cannot disprove SFFA’s
regression model by declaring “simply that many factors go into making up an individual [admissions
decision].” Bazemore, 478 U.S. at 403 n.14. To “defeat the inference of discrimination created by
plaintiffs’ statistics,” the “defendant” must “introduc[e] evidence to support the contention that the
missing factor can explain the disparities as a product of a legitimate, nondiscriminatory selection
criterion.” Palmer, 815 F.2d at 101; accord EEOC v. Gen. Tel. Co. of Nw., Inc., 885 F.2d 575, 583 (9th Cir.
1989) (collecting cases holding that defendants must “produce credible evidence that curing the alleged
flaws [in the plaintiff’s regression model] would also cure the statistical disparity”).
Harvard did not do that. Adding only one of the disputed variables—the personal rating—
would “cure the statistical disparity.” SFFA-FF 68; supra ¶¶ 104, 106. Professor Arcidiacono was
correct to exclude the personal rating because it is not race neutral. SFFA-FF 49-68; supra ¶¶ 132-140.
Absent “clear, affirmative evidence” that a variable is “neutral” and “applied consistently” across racial
groups, “it is the law that [the] variable is to be excluded” from a regression model that seeks to
measure racial discrimination. Trout v. Garrett, 780 F. Supp. 1396, 1412-13 (D.D.C. 1991); accord Segar,
738 F.2d at 1276 (Unless a variable is “‘untainted by discrimination,’” it must be “exclu[ded]” from
the model.); James, 559 F.2d at 332. Harvard does not try to prove that “personal qualities” are actually
unequally distributed across racial groups—that, year after year, African Americans have the best
personal qualities, followed by Hispanics, then whites, and then Asian Americans. SFFA-FF 187.
Thus, Harvard must be imposing these racial disparities, since the law presumes that desirable qualities

55

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 64 of 85

are equally distributed across races unless the defendant can prove otherwise. Segar, 738 F.2d at 127677; De Medina v. Reinhardt, 686 F.2d 997, 1009 & n.7 (D.C. Cir. 1982).
If Harvard means to argue that its admissions process cannot be modeled at all because it uses
unmeasurable subjective criteria, that is wrong too. SFFA-CL 180. Harvard’s argument would mean
that the myriad employment-discrimination cases using regression analysis are wrong, since “almost
all [employment] criteria necessarily have both subjective and objective elements.” Atonio v. Wards Cove
Packing Co., 810 F.2d 1477, 1485 (9th Cir. 1987) (en banc); e.g., Griffin v. Carlin, 755 F.2d 1516, 152526 (11th Cir. 1985) (“promotion process” with “both objective elements … and subjective elements”);
Pegues v. Miss. State Emp’t Serv., 699 F.2d 760, 765, 768-69 (5th Cir. 1983) (referral system that “rel[ied]
on subjective judgments, despite the corralling by objective standards”); EEOC v. Tex. Roadhouse, Inc.,
215 F. Supp. 3d 140, 151, 157-58 (D. Mass. 2016) (hiring process that included criteria such as “happy
and attractive”). Unmeasurable criteria by definition cannot be included in any regression model, yet
those models are still considered reliable. Segar, 738 F.2d at 1276.
Moreover, “[t]he law is clear” that excluding unmeasurable “subjective requirements” from
regression analyses “is entirely proper.” Id. Excluding “an applicant’s essay, the contents of
recommendation[s] . . . and faculty evaluations of a student’s academic or artistic work,” FF 148,
cannot possibly affect the model unless Harvard can prove that they correlate with race, Segar, 738
F.2d at 1276-77. Because Harvard has submitted no evidence that they do correlate with skin color,
these criteria are irrelevant. Indeed, the law penalizes defendants for using non-falsifiable criteria,
which “lend themselves to racially discriminatory abuse.” SFFA-CL 190, 174. Courts presume that
these criteria, if they could somehow be added to the model, would only make the results worse for
the defendant. Id. at 1276; accord Brown v. Gaston Cty. Dyeing Mach. Co., 457 F.2d 1377, 1382-83 (4th
Cir. 1972) (“The lack of objective guidelines … serve to corroborate, not to rebut, the racial bias
pictured by the statistical pattern.”).

56

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 65 of 85

274. Harvard contends that SFFA’s theory of discrimination is contradictory because Harvard
does not appear to impose an admissions penalty on Asian-American ALDCs. “If Harvard intended
to discriminate against Asian-American applicants on the basis of their race,” Harvard contends, “such
discrimination would presumably extend to all Asian-American applicants, not just those who are not
ALDC applicants.” Harvard is wrong both factually and legally.

Factually, Harvard does discriminate against Asian ALDCs. While the small number of AsianAmerican ALDCs counsels against drawing strong statistical conclusions about them, the data clearly
supports SFFA’s position. Like non-ALDCs, Asian-American ALDCs get lower personal ratings than
white ALDCs. SFFA’s theory of discrimination—that Harvard racially stereotypes Asian Americans
on the personal rating—thus holds true for all Asian Americans. Supra ¶ 142. That Asian-American
ALDCs do not appear to be admitted at a lower rate than white ALDCs does not disprove Harvard’s
discrimination. “One familiar aspect of [race] discrimination is the practice, whether conscious or
unconscious, of subjecting [minorities] to higher standards of evaluation than are applied to their
[white] counterparts. [A] court could [thus] conclude[] consistently that [an Asian-American ALDC]
merited [admission] by any standard … and that [an Asian-American non-ALDC] would have been
[admitted] had she been evaluated against the standard that was applied generally to [whites].” Sweeney
v. Bd. of Trustees of Keene State Coll., 604 F.2d 106, 114 (1st Cir. 1979).
Even if Harvard did not discriminate against Asian-American ALDCs, that fact would prove
little. SFFA alleges that Harvard stereotypes Asian Americans on the personal rating to reduce their
presence on campus—not eliminate it entirely. Because Harvard generally prefers ALDCs to nonALDCs and because Asian ALDCs are so rare, Harvard can admit them all without compromising its
goal. SFFA-CL 203. Harvard did the same thing with Jewish applicants in the 1920s. To achieve its
goal of “reduc[ing] the number of Jews,” Harvard discriminated against “‘undesirable’ Jews” from
eastern Europe (who tended to be poorer, non-legacy, and recent immigrants) but not German Jews

57

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 66 of 85

(who tended to be wealthy legacies). Jerome Karabel, The Chosen: The Hidden History of Admission and
Exclusion at Harvard, Yale, and Princeton 98-99 (2005).
Separately, “[i]t is well established that an actor may consider another individual’s race … to
be an asset in some circumstances but a motivation for unlawful discrimination in other
circumstances.” Woods v. City of Greensboro, 855 F.3d 639, 651 (4th Cir. 2017). One well-known aspect
of stereotyping is “the superstar problem,” where “a few superstars are treated well while the bulk of
[the minority group] experience persistent problems.” Joan C. Williams, The Social Psychology of
Stereotyping, 7 Emp. Rts. & Emp. Pol’y J. 401, 434-35 (2003). The few individuals who “break the
stereotypical mold through unstereotypical behavior”—say, an Asian American who is a legacy
(instead of a recent immigrant) or an athlete (instead of a bookworm)—are “perceived as more
[desirable] than similar [whites].” Id. at 418 n.94. But this phenomenon is entirely consistent with illegal
stereotyping, since minorities “who are just below [the superstar] level tend to get disproportionately
lower evaluations.” Deborah L. Rhode, The Subtle Side of Sexism, 16 Colum. J. Gender & L. 613, 620
(2007); Williams, supra, at 418-19.

Legally, Harvard is no less liable if it discriminates against only non-ALDC Asian Americans.
Discrimination on the basis of a protected classification “plus” some other characteristic is still illegal
discrimination. SFFA-CL 204-06. In other words, a defendant who “does not discriminate against the
class … as a whole” but discriminates against only “a subclass” is liable all the same. Chadwick v.
WellPoint, Inc., 561 F.3d 38, 43-44 (1st Cir. 2009). The logic behind this “plus” rubric is simple algebra:
[W]hen one proceeds to cancel out the common characteristics of the two classes
being compared ([e.g.,] married men and married women), as one would do in solving
an algebraic equation, the cancelled-out element proves to be that of married status,
and sex remains the only operative factor in the equation.
Lex K. Larson, Employment Discrimination § 40.04, at 40-12 (2d ed. 1996). To illustrate, take Harvard’s
argument and replace “Asian Americans who are ALDC” with “women who have young children.”
According to Harvard, “[i]f [a defendant] intended to discriminate against [women] on the basis of
58

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 67 of 85

their [sex], such discrimination would presumably extend to all [women], not just those who [have
preschool children].” CL 274. The Supreme Court and First Circuit have rejected this exact argument.
Phillips v. Martin Marietta Corp., 400 U.S. 542, 543-44 (1971); Chadwick, 561 F.3d at 43.
While the “plus” rubric arose under Title VII, the reasoning behind it applies equally to racebased discrimination under Title VI. Discrimination on the basis of sex “plus” something else is illegal
because Title VII requires only that sex be “a motivating factor” for the defendant’s conduct. Chadwick,
561 F.3d at 43 (quoting 42 U.S.C. § 2000e-2(m)); Franchina v. City of Providence, 881 F.3d 32, 53 (1st Cir.
2018). The same is true for race under Title VI. Supra ¶¶ 260-261. Moreover, sex-plus discrimination
violates Title VII because that law prohibits sex “stereotyping.” Chadwick, 561 F.3d at 44 (citing Price
Waterhouse v. Hopkins, 490 U.S. 228, 235 (1989)). The Equal Protection Clause (and thus Title VI)
likewise prohibit racial stereotyping. Supra ¶¶ 260-261. Under the race-plus rubric, then, Harvard is
liable under Title VI even if it does not discriminate against Asian-American ALDCs; its discrimination
against the non-ALDC subclass (i.e., 98% of Asian-American applicants) is sufficient to establish
liability.
275-76. Because SFFA proved that Harvard imposes a statistically significant penalty on Asian
Americans, it demonstrated a prima facie pattern or practice of intentional discrimination. SFFA-CL
173-74. Harvard did not rebut SFFA’s statistical case by proving that the observed disparity is
“insignificant,” Teamsters, 431 U.S. at 360, as Professor Card’s analysis is flawed, SFFA-FF 63-93; supra
¶¶ 94-151. And Harvard did not rebut SFFA’s statistical case by providing a “nondiscriminatory
explanation” for the disparity, Teamsters, 431 U.S. at 360 n.46, as Harvard did not prove that Asian
Americans are actually inferior to whites on the qualities measured by the personal rating,
SFFA-FF 57. While Harvard theoretically could have rebutted SFFA’s statistics with anecdotal or
other non-statistical evidence, that evidence would have to be “sufficient … to permit the trier of fact
to decline to draw the inference of discrimination from the [statistical] proof”—a “high[]” bar. Segar,

59

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 68 of 85

738 F.2d at 1269-70. The only affirmative evidence that Harvard produced was its admissions officers’
testimony. But their flat denials and general assertions of good faith cannot rebut a statistical showing
of discrimination, SFFA-CL 176; Teamsters, 431 U.S. at 343 n.24; Turner, 396 U.S. at 361—especially
because Harvard’s witnesses suffered major credibility problems, SFFA-FF 129-133; supra ¶¶ 133, 199204. In short, SFFA presented a prima facie case and Harvard failed to adequately rebut it, which
means SFFA is entitled to judgment. Teamsters, 431 U.S. at 342-43; Castaneda, 430 U.S. at 501; Wayte v.
United States, 470 U.S. 598, 625 (1985) (Marshall, J., dissenting) (“A prima facie case, of course, is one
that if unrebutted will lead to a finding.”).1
Any suggestion that SFFA cannot prevail on statistics alone—that SFFA must supplement its
statistical case with anecdotal testimony from its members—is incorrect:
[W]hen a plaintiff’s statistical methodology focuses on the appropriate . . . pool and
generates evidence of discrimination at a statistically significant level, no sound policy
reason exists for subjecting the plaintiff to the additional requirement of . . . providing
anecdotal evidence . . . . Such a rule would reflect little more than a superstitious
hostility to statistical proof, a preference for the intuitionistic and individualistic over
the scientific and systemic. It is not difficult to understand that discrimination might
exist even when affected individuals can point to no specific instances of an employer’s
discriminatory conduct. The days of Bull Connor are largely past; discrimination now
works more subtly. Yet its effects are no less pernicious. “In many cases the only
available avenue of proof is the use of racial statistics to uncover clandestine and covert
discrimination.”
Segar, 738 F.2d at 1278-79 (quoting Teamsters, 431 U.S. at 340 n.20). In pattern-or-practice cases,
anecdotal evidence “regarding specific instances of alleged discrimination … simply provides ‘texture’
to the statistics”; it brings the cold numbers to life for a jury. Robinson, 267 F.3d at 168. “Such anecdotal

1 Wisely, Harvard does not argue that the testimony of the two Asian-American student-amici can
rebut SFFA’s statistical showing. An “attempt to combat a prima facie case with evidence about a handful of .
. . decisions is doomed to failure because a prima facie case of class-wide discrimination ‘is not met by
[defendant’s] attempts to parry specific allegations of alleged discrimination.’” Boykin v. Georgia-Pac. Corp., 706
F.2d 1384, 1393 (5th Cir. 1983). “Where the plaintiff is relying on statistical evidence, . . . these witnesses are
about as useful as a football in a game of basketball.” Watson v. Fort Worth Bank & Tr., 798 F.2d 791, 810 n.19
(5th Cir. 1986) (Goldberg, J., dissenting), vacated, 487 U.S. 977 (1988).

60

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 69 of 85

evidence” is not required (especially in a bench trial), since the anecdotes are not even “introduced to
establish that the particular instances of discrimination actually occurred.” Id.; accord Cooper v. Fed.
Reserve Bank of Richmond, 467 U.S. 867, 876 (1984) (“‘[A]t the liability stage of a pattern-or-practice trial
the focus often will not be on individual . . . decisions, but on a pattern of discriminatory
decisionmaking.’”); LUCHA, 799 F.2d at 786 n.21 (“‘[P]attern or practice’ suits . . . focus on
systemwide discrimination rather than on discrimination against an individual . . . applicant.”).
Consider the litigation in Parents Involved. There, an association of parents challenged Seattle’s
race-based high-school assignment policy on behalf of their children. 551 U.S. at 713, 718. No one
demanded testimony or anecdotal accounts from the children. In fact, the children who were standing
members when the case was filed had already aged out and been replaced by completely new children
when the case reached the Supreme Court years later. Br. of Respondent, Parents Involved, 2006 WL
2922956, at *16. But that was not a problem because the association, like SFFA, was challenging a
policy—not individual acts of discrimination against each member. Also, like SFFA, the association
successfully asserted associational standing, 377 F.3d 949, 958 (9th Cir. 2004), a key premise of which
is that “neither the claim asserted nor the relief requested requires the participation of individual
members in the lawsuit.,” Hunt v. Wash. State Apple Advert. Comm’n, 432 U.S. 333, 343 (1977). That is
why SFFA’s members did not need to testify at trial. Requiring them to do so would make them face
the harassment and vitriol that was hurled at Abigail Fisher. See Doc.150-4; cf. 13A Fed. Prac. & Proc.
Juris. § 3531.9.5 n.11 (3d ed.) (associational standing protects individuals who “may fear to be
individually identified with locally unpopular causes” or may be “subject to [retaliation]”).
277. Although SFFA’s statistical evidence is sufficient, SFFA has buttressed it with other
persuasive evidence of discrimination. SFFA-FF 94-133; supra ¶¶ 160-198. All of this evidence must
be considered together “as cumulative.” Segar, 738 F.2d at 1278. Contrary to Harvard’s suggestion,
SFFA does not view its non-statistical evidence as “largely irrelevant.” CL 277 (citing T15.52:4-6).

61

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 70 of 85

What SFFA views as not “particularly important” is the presence or absence of “anecdotal examples”
of discrimination, given SFFA’s powerful “statistical case.” T15.52:1-22; supra ¶¶ 275-276.
278. To prevail, SFFA does not need to catch Harvard making “discriminatory comments”
or engaging in explicit “racial discrimination.” Plaintiffs alleging intentional discrimination are “not
required to come forward with evidence of the ‘smoking gun’ variety.” Resare v. Raytheon Co., 981 F.2d
32, 42 (1st Cir. 1992). The absence of such evidence suggests nothing: “defendants of even minimal
sophistication will neither admit discriminatory animus or leave a paper trail demonstrating it,”
especially not “a notation in a [plaintiff’s] file.” Aman v. Cort Furniture Rental Corp., 85 F.3d 1074, 1082
(3d Cir. 1996); Rosen v. Thornburgh, 928 F.2d 528, 533 (2d Cir. 1991); Thomas v. Eastman Kodak Co., 183
F.3d 38, 58 n.12 (1st Cir. 1999). Even still, SFFA did find evidence of Asian-American discrimination
in Harvard’s files. SFFA-FF 110-111; supra ¶¶ 189-193. “Even if the remarks alone might [be]
insufficient,” they are “certainly relevant and, along with other substantial evidence, create[] a strong
inference of intentional discrimination.” Mangold v. Cal. Pub. Utilities Comm’n, 67 F.3d 1470, 1477 (9th
Cir. 1995).
279. Harvard’s attempts to minimize the discrimination against Asian Americans living in
Sparse Country badly misses the mark. Harvard holds these Asian Americans to a higher standard
than whites, SFFA-FF 106, which Harvard no longer factually disputes, FF 196. Harvard’s newlyminted arguments for why this is not evidence of intentional discrimination lack support and are
contradicted by Fitzsimmons’ testimony. Supra ¶¶ 194-198.
280. SFFA does not contend that Harvard independently violated Title VI by discriminating
against Asian Americans in recruiting. SFFA contends that it helps prove that Harvard discriminates
against Asian Americans in admissions, for at least three reasons. First, the fact that Harvard explicitly
imposed higher standards on Asian Americans in Sparse Country is a particularly stark example of
how it treats Asian Americans vis-à-vis whites; it contradicts Harvard’s assertion that its process is

62

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 71 of 85

“race neutral” between whites and Asian Americans. Second, when asked to explain its discriminatory
recruiting standards, Fitzsimmons reached immediately for an Asian stereotype; he asserted with zero
evidence that Asian Americans are more likely to be recent immigrants to Sparse Country. SFFA-FF
106. That the Dean of Admissions could not resist Asian stereotyping in open court suggests that far
worse Asian stereotyping goes on behind closed doors. Third, Harvard’s discriminatory “recruitment
techniques” sent a “message” to Asian Americans that “surely deterr[ed]” many of them from
“submitting an application.” Teamsters, 431 U.S. at 365-66; see T1.150:25-153:8. That “discrimination
affects the applicant pool in a way that makes the discrimination [in admissions] harder to detect[,]”
since “these non-applicants—victims of discrimination as much as the [rejected] applicants—will
make [Harvard’s admissions] look ‘better’ than it is.” Mister, 832 F.2d at 1436. Thus, Harvard’s
discriminatory recruiting means SFFA’s statistics understate the degree that Harvard discriminates
against Asian Americans. Id.
281. Harvard’s ever-changing position on why Fitzsimmons took no steps in response to
OIR’s findings is reaching the point of absurdity. The work was too preliminary and rudimentary to
be trustworthy—except when it came to touting OIR’s regression analysis showing that Harvard gives
a preference to low-income applicants. SFFA-FF 122-126; supra ¶ 178. OIR’s work confirmed what
Fitzsimmons already knew from OCR’s findings—except OIR (unlike OCR) found that athlete and
legacy preferences could not explain why Asian Americans were faring worse than whites. SFFA-FF
Supra ¶¶ 153-54, 163. OIR’s findings were “circular”—except Harvard concedes that, at most, this
criticism applies to one of several models in one of several studies that OIR shared with Fitzsimmons.
FF 164. Now, it turns out Fitzsimmons (who likes discussing statistics, previously taught the subject,
and had no problem grasping regression analysis about low-income applicants) just did not understand
what he was being told. Except OIR was clear: being Asian American is “negatively associated” with
admission and “negatively correlated with the admission rate.” SFFA-FF 124; supra ¶¶ 178-184, 188.

63

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 72 of 85

282. That Harvard “buried” the OIR reports “demonstrate[s] racial animus.” Lowery v. Circuit
City Stores, Inc., 206 F.3d 431, 439 (4th Cir. 2000). As does Harvard’s awareness (from OIR and OCR)
that its admissions process discriminates against Asian Americans and its decision to do nothing about
it. See, e.g., McReynolds v. Sodexho Marriott Servs., Inc., 349 F. Supp. 2d 1, 18 & n.21 (D.D.C. 2004)
(“statements” and “corporate documents” “indicating an awareness of [the] problem” was “powerful
support to plaintiffs’ statistical showing of significantly low levels of African American promotions”);
Police Officers for Equal Rights v. City of Columbus, 644 F. Supp. 393, 436 (S.D. Ohio 1985) (“Defendants’
knowledge of and refusal to act upon evidence which indicated an underrepresentation of blacks . . .
supports an inference of intentional discrimination”); Stender v. Lucky Stores, Inc., 803 F. Supp. 259,
331-32 (N.D. Cal. 1992) (that the defendant knew women were underrepresented, received a report
on that issue, and implemented no reforms was evidence of discriminatory intent).
While “disparate impact and foreseeable consequences, without more, do not establish a
constitutional violation,” “actions having foreseeable and anticipated disparate impact are relevant
evidence to prove the ultimate fact, forbidden purpose.” Columbus Bd. of Educ. v. Penick, 443 U.S. 449,
464 (1979). A defendant can rebut this inference of discriminatory intent “[w]hen . . . the impact is
essentially an unavoidable consequence of a legislative policy that has in itself always been deemed to
be legitimate”—for example, a policy that gives preferences to veterans even though most are men.
Pers. Adm’r of Mass. v. Feeney, 442 U.S. 256, 279 n.25 (1979). But there is nothing “legitimate” about
limiting the number of Asian Americans admitted to Harvard. That Harvard’s leadership knew its
admissions process imposed a penalty on Asian Americans, but changed nothing, creates “a strong
inference that the adverse effects were desired.” Id. That “an official [at Harvard] with power to take
corrective measures [wa]s ‘deliberately indifferent to known acts’ of discrimination” is sufficient to
make Harvard “liable under Title VI.” United States v. Cty. of Maricopa, 889 F.3d 648, 652-53 (9th Cir.
2018). As is a pattern of discrimination against Asian-American applicants. SFFA-CL 196.

64

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 73 of 85

283. Harvard is hung up on “unconscious bias.” To be clear: SFFA alleges that Harvard is
racially stereotyping Asian Americans. The Equal Protection Clause (and thus Title VI) “forbids” racial
stereotyping. Supra ¶¶ 260-261; Miller v. Johnson, 515 U.S. 900, 928 (1995); see, e.g., id. at 920 (“[W]here
the State assumes from a group of voters’ race that they ‘think alike, . . .’ it engages in racial stereotyping
at odds with equal protection mandates.”); Shaw v. Reno, 509 U.S. 630, 647-48 (1993) (“[A] prosecutor’s
assumption that a black juror may be presumed to be partial simply because he is black . . . violates
the Equal Protection Clause.”). “The Supreme Court has long recognized that unlawful discrimination
can stem from stereotypes.” Thomas, 183 F.3d at 59.
The First Circuit recognizes that racial stereotyping is a “form[] of less conscious bias.” Id. at
42. Illegal stereotyping can be “unwitting or ingrained,” such that the defendant might be “unaware”
that he is acting on a stereotype rather than an accurate observation about the world. Id. at 60. This
does not mean that racial stereotyping is not intentional discrimination; racial stereotyping occurs “in
the context of more nuanced decisions that can be explained based upon reasons other than illicit
bias, which, though perhaps implicit, is no less intentional.” Woods, 855 F.3d at 652 (emphasis added).
“Invidious discrimination steeped in racial stereotyping is no less corrosive of the achievement of
equality than invidious discrimination rooted in other mental states.” Id. at 651.
284-85. Title VI does not require “racial animus,” if Harvard is using that term to mean “ill
will, enmity, or hostility.” SFFA-CL 169 (quoting EEOC v. Joe’s Stone Crab, Inc., 220 F.3d 1263, 1284
(11th Cir. 2000)). The First Circuit’s decision in Goodman did not hold otherwise; it simply used the
phrase “racial animus” as a synonym for “intentional discrimination.” See 380 F.3d at 43. Harvard’s
attempt to tease a draconian rule out of Goodman’s use of these two words is creative, but untenable.
No court has ever held that a law prohibiting intentional discrimination requires the plaintiff to prove
that the defendant harbored ill will toward minorities.

65

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 74 of 85

For example, section 1981, which requires “proof of discriminatory intent,” can be violated
even when there is “no suggestion . . . that the [defendants] held any racial animus.” Goodman v. Lukens
Steel Co., 482 U.S. 656, 664, 668 (1987). That is because intentional discrimination does not turn on
“‘whether, as a subjective matter, [defendants] were favorably disposed toward minorities.’” Id. at 669.
Similarly, Harvard concedes that Title VII’s ban on intentional discrimination does not require “racial
animus.” Thomas, 183 F.3d at 42. That concession is fatal given the similarities between Title VII and
Title VI. Supra ¶ 268. If anything, it is easier to prove intentional discrimination under Title VI (and the
constitutional standard it incorporates) than Title VII. Johnson v. Transp. Agency, Santa Clara Cty., 480
U.S. 616, 628 n.6, 630 n.8 (1987) (Title VII “was not intended to extend as far” as Title VI or the
Constitution due to “‘Congress’ concern that federal courts not impose unwanted obligations on
employers and unions’”); Keller v. Prince George’s Cty., 827 F.2d 952, 963 (4th Cir. 1987) (Title VII is
“more liberal” than “[t]he stricter standards” of the Equal Protection Clause).
Moreover, Title VI cannot require racial animus because it incorporates the Equal Protection
Clause, which prohibits racial stereotyping regardless of the defendant’s motives. For example, Gratz
held that Michigan violated the Constitution and Title VI by engaging in racial stereotyping, even
though the university did so to benefit underrepresented minorities. 539 U.S. at 271, 275. Or consider
racial gerrymandering. Those cases hold that state legislatures violate the Equal Protection Clause by
engaging in a species of “racial stereotyping”—the “offensive and demeaning assumption that voters
of a particular race … will prefer the same candidates at the polls.” Bush v. Vera, 517 U.S. 952, 968
(1996) (opinion of O’Connor, J.); Miller, 515 U.S. at 912. Often, though, the defendants who are found
liable for stereotyping are Democratic legislatures who were trying to increase minorities’ voting power
and comply with the Voting Rights Act. See, e.g., Bush, 517 U.S. at 957; Miller, 515 U.S. at 907-08; Shaw,
509 U.S. at 635-36. No “animus” was present or required in any of those cases.

66

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 75 of 85

286-91. Harvard largely repeats its criticisms of SFFA’s evidence, which are answered above.
Harvard makes three new arguments, but they are all unpersuasive.

First, Harvard expresses disbelief that a “40 member” admissions committee could engage in
racial stereotyping against Asian Americans. But courts routinely hold that legislatures (which often have
far more than 40 people) engage in racial stereotyping and other forms of intentional discrimination.
Supra ¶ 284 (racial gerrymandering); Hunter v. Underwood, 471 U.S. 222, 229 (1985) (Alabama
constitutional convention of 1901 meant to discriminate against African Americans); United States v.
Windsor, 570 U.S. 744, 776 (2013) (342 Representatives, 85 Senators, and President Clinton meant to
discriminate against same-sex couples). The need to prove intent of a multimember body is precisely
why Arlington Heights identified objective indicia of intent that plaintiffs could use.
Regardless, SFFA does not need to prove that all 40 members of Harvard’s admissions
committee acted with “an aligned and common cognitive bias against Asian-Americans.” CL 286. The
Ninth Circuit, in a similar case, explained how discrimination and liability work in this context:
The principal defendant in this case is the University, which has delegated to the faculty
near-total control over hiring. The faculty, first in committee, then as a whole, reviews
applications, chooses the final candidates, and votes on whether to extend any
candidate an offer of employment. The hiring process is therefore not insulated from
the illegitimate biases of faculty members. Indeed, since the faculty is small . . . and
great emphasis is placed on collegiality and consensus decisionmaking, even a single
person’s biases may be relatively influential. . . . As other courts have recognized,
discrimination at any stage of the academic hiring or promotion process may infect
the ultimate employment decision. . . . Accordingly, a plaintiff in a university
discrimination case need not prove intentional discrimination at every stage of the
decisionmaking process . . . . As numerous cases . . . demonstrate, where a university
has delegated employment decisions to a committee and members of that committee
have allegedly engaged in discriminatory treatment, the university is liable.
Lam v. Univ. of Haw., 40 F.3d 1551, 1561-62 (9th Cir. 1994); see, e.g., Roebuck v. Drexel Univ., 852 F.2d
715, 727 (3d Cir. 1988) (attributing discrimination to a university, even though “each successive
evaluator performed a de novo review,” because “at each stage of the process the evaluator had
available and considered the reports and recommendations of each previous evaluator” and so “an

67

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 76 of 85

evaluation at any level, if based on discrimination, . . . allowed discrimination to infect the ultimate
decision”). Harvard is thus liable under Title VI, even if the discrimination originates with only some
members of its admissions committee.

Second, Harvard claims that, unlike OCR, SFFA did not “conduct any rigorous review” of
Harvard’s application files. Harvard-CL 288. That is untrue. SFFA thoroughly examined every file
Harvard was ordered to produce; SFFA also closely reviewed the summary sheets that the Court
ordered Harvard to produce. That review produced ample evidence of intentional discrimination.
SFFA-FF 110-111; supra ¶ 189. And, at trial, SFFA highlighted the file of an Asian-American applicant
who been subjected to discrimination to give context to its claims. Id. Harvard’s real complaint is that
SFFA will not relent and allow anecdotal evidence to be the focus of this case. But, again, the contents
of individual application files are largely irrelevant; pattern-or-practice cases challenge policies (not
individual admission decisions, supra ¶¶ 275-276) and SFFA does not need “smoking gun” evidence
to prove Harvard’s liability, supra ¶ 278.

Third, Harvard faults SFFA for not introducing “expert testimony” on “cognitive bias.” CL
289. But SFFA introduced extensive testimony on Asian-American stereotyping. SFFA-FF 94-98, 104.
And this Court does not need expert testimony to conclude that Asian Americans are subject to wellknown racial stereotypes. Chin v. Runnels, 343 F. Supp. 2d 891, 907 (N.D. Cal. 2004) (“[T]his Court
takes note that . . . Asian–Americans have been particularly vulnerable to stereotyping and exclusion
when subjective selection criteria center on ‘leadership’ and ‘people skills,’” “in part because they are
widely stereotyped as ‘passive,’ and ‘unassertive,’ as well as ‘more equipped for technical than peopleoriented work.’” (citing law-review articles)); Chadwick, 561 F.3d at 44. Nor does this Court need expert
testimony to recognize that all people, including Harvard employees, are vulnerable to stereotypical
thinking. Chin, 343 F. Supp. 2d at 906 (“A number of courts have recognized that subjective decisionmaking allows for subtle biases or unconscious stereotyping to affect selection processes.” (citing cases

68

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 77 of 85

and social-science research)); Kimble v. Wis. Dep’t of Workforce Dev., 690 F. Supp. 2d 765, 776 (E.D. Wis.
2010) (“Individuals draw lines and create categories based in part on race, gender and ethnicity, and
the stereotypes they create can bias how they process and interpret information and how they judge
other people.” (citing law-review articles)). SFFA is not aware of any cases that refuse to hold a
defendant liable for racial stereotyping because the plaintiff did not submit “expert testimony,” and
Harvard cites none.
292. For the reasons above, SFFA prevails under strict scrutiny and the pattern-or-practice
framework. But even if Arlington Heights controls, SFFA has proven its case. As noted above, racial
disparities are an “‘important starting point’ for assessing discriminatory intent” even when they are
not stark enough to alone prove intentional discrimination. Reno, 520 U.S. at 489 (quoting Arlington
Heights, 429 U.S. at 266). Professor Arcidiacono and OIR both found that Harvard’s admissions
process “bears more heavily on [Asian Americans],” Arlington Heights, 429 U.S. at 266; SFFA-FF 2437. The “degree” of that effect is stark, and Harvard’s “justification” for the disparity is unpersuasive.
Dowd, 375 F.3d at 83. “The historical background” of Harvard’s admissions system, moreover, stems
from “invidious” discrimination against Jews, Arlington Heights, 429 U.S. at 267; SFFA-FF 96. There
are also “contemporary statements” and testimony that reflect stereotyping. Arlington Heights, 429 U.S.
at 268; SFFA-FF 108-112. Harvard’s abrupt change of its reading procedures on the eve of trial, SFFAFF 113-114, reflects a “[d]eparture[] from the normal procedural sequence.” Arlington Heights, 429 U.S.
at 267. And, there is wealth of additional evidence that confirms that Harvard intentionally
discriminates against Asian-Americans applicants. SFFA-FF 94-133.
B.

Harvard’s Use of Race Does Not Comply with Precedent.
i.

Harvard is not entitled to any deference.

293-302. The Supreme Court gives universities “some” deference on only one issue: whether
“a diverse student body would serve [their] educational goals.” Fisher I, 570 U.S. at 311. But Harvard
receives no deference on the issue SFFA is actually contesting—whether Harvard is pursuing “the
69

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 78 of 85

benefits of a student body diversity that encompasses a broad array of qualifications and characteristics
of which racial or ethnic origin is but a single though important element.” Id. at 314-15. “Strict scrutiny
does not permit a court to accept a school’s assertion that its admissions process uses race in a
permissible way.” Id. at 312. The Court also has “clarified” that universities receive “no deference” on
whether their use of race is “narrowly tailored.” Fisher v. Univ. of Tex. at Austin (Fisher II), 136 S. Ct.
2198, 2208 (2016) (citing Fisher I, 570 U.S. at 311-12).
ii.

Harvard failed to consider or implement workable race-neutral
alternatives.

303-04. Harvard concedes that it must prove “workable” race-neutral alternatives would not
promote “the educational benefits of diversity . . . ‘about as well and at tolerable administrative
expense.’” Fisher I, 570 U.S. at 312. Harvard also must prove that it gave “serious, good faith
consideration” to those alternatives. Id. This Court may “not defer” to Harvard on either issue. Id.
305.

Narrow tailoring does not require a university to “abandon” its “reputation for

excellence” by making a “dramatic sacrifice of . . . academic quality”—for example, moving to a
“lottery system” or “decreasing the emphasis for all applicants on . . . GPA and [test] scores.” Grutter
v. Bollinger, 539 U.S. 306, 339-40 (2003). But narrow tailoring can require less dramatic sacrifices; after
all, a race-neutral alternative need only work “about as well.” Fisher I, 570 U.S. at 312. Similarly, narrow
tailoring does not require a university to “sacrifice all other aspects of diversity” by “privileging one
characteristic above all others”—for example, “exclusive” focus on class rank. Fisher II, 136 S. Ct. at
2213-14. But narrow tailoring can require a university to adopt other measures that do not “preclude
the university from conducting the individualized assessments necessary to assemble a student body
that is not just racially diverse, but diverse along all . . . qualities.” Grutter, 539 U.S. at 340. Harvard
does not have a compelling interest in remaining a bastion of privilege.
306. Harvard has never given serious, good-faith consideration to race-neutral alternatives.
Harvard now concedes that it gave no consideration until 2017, decades after it started using race, CL
70

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 79 of 85

308-10, even though the law has long required such consideration. E.g., City of Richmond v. J.A. Croson
Co., 488 U.S. 469, 507 (1989) (citation omitted). Although Harvard ultimately relented and convened
the Smith Committee in 2017, this long-overdue review by three committed partisans who were staring
down this lawsuit was not serious or in good faith. SFFA-CL 230.
307. That Harvard implemented various race-neutral efforts at the same time it used racebased admissions does not mean it considered race-neutral alternatives. Race-neutral alternatives are
just that—alternatives to use of race. Parents Involved, 551 U.S. at 735 (that the defendant simultaneously
pursued “its goals . . . through means other than the racial classifications” is not “evidence that it
considered alternatives”). Harvard had the duty to study what the world would look like if it did not
use race at all. Wessmann v. Gittens, 160 F.3d 790, 799-800 (1st Cir. 1998) (examining whether “strict[ly]”
neutral admissions would differ significantly in consequence from race-conscious admissions); Johnson
v. Bd. of Regents of Univ. of Ga., 263 F.3d 1234, 1259 (11th Cir. 2001) (examining whether the university
genuinely considered “substituting wholly race-neutral measures” for race-based admissions). Harvard
admits it did not study that question until 2017. Nor, of course, did it consider race-neutral alternatives
“before” turning to race. Fisher I, 570 U.S. at 312; Fisher II, 136 S. Ct. at 2211.2
308. Because Harvard concedes that it did not consider race-neutral alternatives until the
Smith Committee, this Court should hold that Harvard’s admissions process did not comply with Title
VI before 2017. While SFFA does not seek forward-looking relief based on Harvard’s past violations,
the Court should formally recognize Harvard’s noncompliance for at least two reasons. First, this

2 Harvard states that it “eliminated and then reinstated Early Action” in order to “increas[e] diversity.”
CL 307 (citing FF 219). Tellingly, Harvard uses the word “diversity” in this sentence to mean “the matriculation
rate of African-American and Hispanic students.” FF 219. This Freudian slip reflects Harvard’s genuine view—
that “diversity” primarily means the number of underrepresented racial minorities on its campus. In any event,
that Harvard “has considered ways to increase the number of [racial minorities] at the university” is obviously
not “evidence that [it] has sought by race-neutral means to increase the broader diversity of the student body”
or that it “has genuinely considered substituting wholly race-neutral measures for the current race-based
admissions.” Johnson, 263 F.3d at 1259.

71

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 80 of 85

noncompliance means that Harvard was violating the law during Bakke, Grutter, and Fisher—putting
the lie to its repeated insistence that those decisions somehow blessed its admissions process. Second,
Harvard’s long refusal to consider race-neutral alternatives, and its grudging decision to do so only
after it was sued, is powerful evidence that its consideration in 2017 was not serious or in good faith.
Santiago-Ramos v. Centennial P.R. Wireless Corp., 217 F.3d 46, 56 (1st Cir. 2000); Aiken v. City of Memphis,
37 F.3d 1155, 1164 (6th Cir. 1994).
309-11. The Smith Committee was not a serious or good-faith consideration of race-neutral
alternatives. SFFA-FF 153-155; supra ¶¶ 207-210. (Harvard does not even try to argue that the Ryan
Committee did, since it performed little work and was disbanded before reaching any conclusions.)
Harvard did not study race-neutral alternatives (and thus violated Title VI) for at least forty years—
from the time it filed its amicus brief in Bakke until the Smith Committee was convened in the middle
of this litigation. Because Harvard “settled into a continuing practice … violative of [the] laws,” it
should be extremely difficult for Harvard to prove that the Smith Committee acted seriously and in
good faith; “[i]t is the duty of the courts to beware of efforts to defeat injunctive relief by protestations
of repentance and reform, especially when abandonment seems timed to anticipate suit.” United States
v. W. T. Grant Co., 345 U.S. 629, 632 n.5 (1953); accord United States v. Mass. Mar. Acad., 762 F.2d 142,
157-58 (1st Cir. 1985); Chaffin v. Kan. State Fair Bd., 348 F.3d 850, 865 (10th Cir. 2003). Smith and
Fitzsimmons were never going to conclude that they had caused Harvard to violate Title VI by using
race when viable race-neutral alternative were available. Supra ¶¶ 207-210. If this Court has any doubts,
it must resolve them in favor of SFFA; it cannot “‘presume the University acted in good faith’ and
place on [SFFA] the burden of rebutting that presumption.” Fisher I, 570 U.S. at 312.
312-15. Harvard has not proven that it lacks available and workable race-neutral alternatives.
Harvard could achieve student body diversity without using race by eliminating LDC preferences and
increasing socioeconomic preferences. SFFA-FF 157-158; supra ¶¶ 219-255. Harvard does not contest

72

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 81 of 85

that this is a race-neutral approach. And Harvard does not argue that it would impose “[in]tolerable
administrative expense,” Fisher I, 570 U.S. at 312—for example, by increasing Harvard’s financial-aid
expenditures, supra ¶¶ 25-27.
Harvard instead complains that Kahlenberg’s approach would decrease the number of African
Americans from 14% to 10%. CL 314; FF 249. But Harvard does not have a compelling interest in
maintaining a particular percentage of African Americans. Fisher I, 570 U.S. at 311. Harvard must
prove that, if it decreases African-American enrollment to 10% (while increasing the enrollment of
Hispanics, Asian Americans, and non-wealthy students), Harvard could not “about as well” achieve
the “educational benefits of diversity.” Id. at 312. Yet Harvard has not “concretely demonstrated” that
a 4% change in the share of African-American enrollment is “significant in any [educational] way, such
as students’ capacity and willingness to learn.” Wessmann, 160 F.3d at 799-800. It presented no studies,
expert testimony, or anything other than “broad generalizations by a few witnesses” and “rank
speculation.” Id. at 800. Nor did Harvard “attempt to defend” the Smith Committee’s concern with
African-American isolation by proving that there is a material difference between 10% and 14% in
terms of “avoidance of racial isolation.” Parents Involved, 551 U.S. at 727. In fact, African-American
enrollment was 10% in 2016. Supra ¶ 249. Yet Harvard did not express concern—let alone sound the
alarm that it was no longer achieving the educational benefits of diversity.
Even if a 4% drop in African-American enrollment would meaningfully decrease student-body
diversity in a vacuum, Harvard never explains why that drop would not be offset by the massive gains
in racial, geographic, and socioeconomic diversity. Harvard has said not one word about these
“positive” effects. Fisher II, 136 S. Ct. at 2215. “Given . . . the potential dangers of stereotyping, [this
Court] cannot allow generalities emanating from the subjective judgments of [Harvard] officials to
dictate whether a particular percentage of a particular racial or ethnic group is sufficient or insufficient
for individual students to avoid isolation and express ideas.” Wessmann, 160 F.3d at 800. Harvard’s

73

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 82 of 85

other argument is that SFFA’s approach would decrease the academic quality of its freshmen classes.
CL 314; FF 250. This argument is meritless. SFFA-FF 158; supra ¶¶ 228-231, 250.
iii.

Harvard’s use of race violates Grutter.

316. Harvard’s use of race violates several of the limits that Grutter set. Harvard does not
seriously consider diversity in all its forms. SFFA-CL 215-16. Harvard uses race as more than a plus
factor. SFFA-CL 217-20. And Harvard’s use of race is neither limited in logic, SFFA-CL 221-23, nor
in time, SFFA-CL 224-25. Harvard’s amici say out loud what Harvard believes in private: that
universities should always consider race because it is “a fundamental part of [an applicant’s] identity.”
Doc. 626 at 17; Doc. 626 at 55-57 n.30 (criticizing race “blindness”); Doc. 623 at 30 (“Race-blind
admissions is an act of erasure” because “[t]o try to not see my race is to try to not see me”). Harvard
is entitled to that belief, but it cannot act on it while continuing to accept federal funds. Because the
“‘ultimate goal’” of Title VI and the Constitution is “‘eliminating entirely’” consideration of “‘such
irrelevant factors as a human being’s race,’” Harvard cannot reject a “colorblind mentality” without
rejecting “‘Our Constitution’” and the basic “‘civil rights’” principle that “‘all citizens are equal before
the law.’” Parents Involved, 551 U.S. at 730 & n.14.
317-19. The Supreme Court has never evaluated whether Harvard’s admissions system meets
strict scrutiny in practice. SFFA-CL 162. At most, Bakke and Grutter endorsed the theoretical idea of
Harvard’s admissions process, as self-described by Harvard in an untested attachment to its amicus
brief. As this case has revealed, however, Harvard was actively violating the law during Bakke, Grutter,
and Fisher. SFFA-FF 153; supra ¶¶ 203-204. Harvard cannot use precedent as a safe harbor. In fact,
such open defiance of those decisions reveals that not even Harvard believes in them.
320-21. As Harvard concedes, it has the burden to prove that it uses race only as a “plus”
factor and that it treats applicants as individuals. Harvard is wrong to suggest that “individualized”
review is sufficient to prove the legality of its process. Harvard also must prove, among other things,

74

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 83 of 85

that its use of race has a logical and durational endpoint and that it considered and correctly rejected
race-neutral alternatives. SFFA-CL 221-29.
322-23. Harvard’s witnesses did not credibly testify that race is “just one factor among many”
for African-American and Hispanic applicants. And Professor Card confirmed that the preference is
far more than a “plus” factor. SFFA-FF 146; supra ¶ 94. Race is the defining feature for most AfricanAmerican and Hispanic applicants. SFFA-FF 145-150; supra ¶¶ 92-95.
iv.

Harvard engages in racial balancing.

324-325. SFFA agrees that Harvard must prove it does not engage in racial balancing. While
Harvard can pay “some attention to the numbers,” Grutter, 539 U.S. at 336, that attention must be
“tied to . . . [a] pedagogic concept of the level of diversity needed to obtain the asserted educational
benefits.” Parents Involved, 551 U.S. at 726. In other words, Harvard must “work[] forward from some
demonstration of the level of diversity that provides the purported benefits.” Id. at 729. SFFA’s
evidence shows that Harvard is not just paying some attention to numbers—Harvard is engaged in
racial balancing. SFFA-FF 134-140; supra ¶¶ 79-91. The racial composition of Harvard’s admitted class
has not varied meaningfully from year to year. SFFA-FF 140; supra ¶ 79.
C.

SFFA Has Standing.

326-30. Harvard repeats its challenge to SFFA’s standing solely “for the purposes of potential
further review.” CL 328 n.3. This Court correctly held that Harvard’s “indicia of membership” test
does not apply to SFFA and that SFFA (as Harvard conceded) has associational standing under the
Hunt factors. Doc. 324 at 8-17; Doc. 566 at 21-22. This Court also correctly held that SFFA would
satisfy the indicia-of-membership test anyway. Doc. 324 at 15-17.

75

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 84 of 85

Dated: January 23, 2019

Respectfully submitted,
/s/ William S. Consovoy
wsc
William S. Consovoy
Thomas R. McCarthy
J. Michael Connolly
Cameron T. Norris
CONSOVOY MCCARTHY PARK PLLC
3033 Wilson Boulevard, Suite 700
Arlington, Virginia 22201
703.243.9423
[e-mail]
[e-mail]
[e-mail]
[e-mail]

Adam K. Mortara
Scott J. McBride
Krista J. Perry
BARTLIT BECK LLP
54 West Hubbard Street, Suite 300
Chicago, IL 60654
312.494.4400
[e-mail]
[e-mail]
[e-mail]
John M. Hughes
Katherine L.I. Hacker
Meg E. Fasulo
BARTLIT BECK LLP
1801 Wewatta Street, Suite 1200
Denver, CO 80202
303.592.3100
[e-mail]
[e-mail]
[e-mail]
Paul M. Sanford BBO #566318
BURNS & LEVINSON LLP
One Citizens Plaza, Suite 1100
Providence, RI 02903
617.345.3000
[e-mail]

Patrick Strawbridge BBO #678274
CONSOVOY MCCARTHY PARK PLLC
Ten Post Office Square
8th Floor South PMB #706
Boston, MA 02109
617.227.0548
[e-mail]
Michael H. Park
CONSOVOY MCCARTHY PARK PLLC
745 Fifth Avenue, Suite 500
New York, NY 10151
212.247.8006
[e-mail]

76

Case 1:14-cv-14176-ADB Document 627 Filed 01/23/19 Page 85 of 85

CERTIFICATE OF SERVICE
I hereby certify that this document filed through the CM/ECF system will be electronically
sent to all counsel of record via the CM/ECF system.
/s/ William S. Consovoy
William S. Consovoy

77

wsc