Introduction

Social science has informed the U.S. Supreme Court’s findings and rulings in a number of decisions affecting racial equality and equal opportunity.¹  But the evidence provided to the Supreme Court is not always sound, and when it is not, claims of questionable empirical provenance can easily crowd the marketplace of ideas.²  One area in which social science has an important role to play is where doctrine intersects with real world consequences, as it does in the pending Fisher v. University of Texas (Fisher II) case.  In the first Fisher v. University of Texas (Fisher I) case, the Supreme Court outlined considerations that should guide trial courts in deciding the constitutionality of university affirmative action programs and, in particular, the viability of race-neutral alternatives.  The Fisher I Court stated, “The reviewing court must ultimately be satisfied that no workable race-neutral alternatives would produce the educational benefits of diversity.”³  The Court noted that while narrow tailoring “does not require exhaustion of every conceivable race-neutral alternative” the lower courts are required to “examine with care, and not defer to, a university’s ‘serious, good faith consideration of workable race-neutral alternatives.’”⁴  The Fisher I Court declared that a university’s consideration of race should be deemed impermissible if the reviewing court were to find that “a nonracial approach . . . could promote the substantial interest about as well and at tolerable administrative expense.”⁵  The Court concluded that strict scrutiny can be neither “strict in theory, but fatal in fact” nor “strict in theory but feeble in fact.”⁶

In response to these directives from the Court in Fisher I, on remand the Fifth Circuit panel soberly concluded, “Put simply, this record shows that UT Austin implemented every race-neutral effort that its detractors now insist must be exhausted prior to adopting a race-conscious admissions program—in addition to an automatic admissions plan not required under Grutter that admits over 80% of the student body with no facial use of race at all.”⁷  The Fifth Circuit was likewise not persuaded by Ms. Fisher’s arguments that socioeconomic disadvantage should replace consideration of race in University of Texas, Austin’s (UT’s) holistic admissions program; instead the court was persuaded by UT’s citations to the social science research indicating that the utility of socioeconomic disadvantage as a means of yielding racial diversity is significantly limited.⁸   Moreover, the Fifth Circuit conceded that in the context of the Fisher litigation it was relatively “ill-equipped to sort out race, class, and socioeconomic structures” and that wishing that race no longer matters “does not make it so.”⁹

How the Court rules in Fisher II will be revealed soon enough—with Justice Scalia’s passing and Justice Kagan’s recusal, the most plausible scenarios have Justice Kennedy casting the decisive swing vote in a 4-3 decision (or 3-1-3 decision akin to Justice Powell’s opinion in Regents of the University of California v. Bakke),¹⁰ or perhaps because of the problems around plaintiff’s standing and/or the University’s Eleventh Amendment immunity, the Court will dismiss this latest certiorari petition as improvidently granted.¹¹  Regardless of the exact outcome in Fisher II, the social science questions around percent plans and class-based race-neutral alternatives are important in the post-Fisher constitutional landscape that U.S. colleges and universities must operate within for the foreseeable future.  Accordingly, within the space constraints of this Essay, I attempt to summarize and synthesize the contemporary social science and policy scholarship relevant to the questions of race-neutral alternatives.  As will be demonstrated below, the consensus conclusion of social science researchers is that percent plan and class-based alternatives at the undergraduate level are neither substantively effective nor cost-efficient substitutes for race-conscious admission programs at leading American universities.¹²

The question I address in this Essay is whether, when the goal is to achieve racial and ethnic diversity in the student body, class can substitute for race in university admissions.  This question and associated value implications are analytically distinct from the important social policy justifications supporting socioeconomic diversity for its own sake.  Percentage plans and affirmative action bans are discussed together in this Essay because of frequent overlap between the two in situations the courts have been called on to address.¹³  The social science questions I address in this Essay are also separate from doctrinal questions about whether percentage plans and class-based efforts might have constitutional vulnerabilities of their own.¹⁴

For reasons of organization and economy, in this Essay I draw a pragmatic distinction between primary sources that are the principal focus of discussion herein and additional sources that round out the social science literature responsive to particular questions.¹⁵  The primary works discussed in this Essay are largely made up of peer-reviewed social science pieces published within the last half-dozen years or so.  Some additional works are included in this primary category because they are repeatedly cited in various Fisher I and II amicus briefs and deserve comment or criticism for that reason alone, particularly when the notable absence of scholarly peer-review may have contributed to sloppy and/or misleading claims.  The additional works are noted in this Essay at various points to reinforce certain themes and to provide readers with a sense of the weight of social science authority.  Full citations for these additional works are provided in Appendix A.

I.  Studies With a Nationwide Focus

Because it is the broader implications of Fisher I and II for selective U.S. colleges and universities that has greater policy relevance, I begin with social science studies that focus on socioeconomic or percent plan simulations on a nationwide level.  These studies show with a fair degree of consistency that class-based approaches and percent plans are not workable race-neutral alternatives, and the exceptions to the general rule occur under the most aggressive and unrealistic modeling assumptions.  One important 2015 study by Reardon et al. used agent-based modeling to analyze a nationally representative data set of 10,000 students graduating from high school and applying to colleges in 2004.¹⁶  A strength of this approach is that the authors capture a significant amount of the dynamism of the application, admissions, and enrollment systems in the United States.  Thus, the approach can take into account how outcomes at some colleges impact outcomes at competing and less selective institutions and the fact that more affluent students apply and choose to enroll at selective universities at higher rates than low-income students.¹⁷  Reardon et al. also performed multiyear simulations with the same data to get a sense of the equilibria that would eventually “burn-in” after long-term information about students’ patterns of choices and colleges’ decisions became available.¹⁸  This research team then performed a series of robust simulations on the top 10 percent of institutions in terms of selectivity and tested combinations of varying degrees (none, moderate, or strong) of both race-conscious programs and socioeconomic¹⁹ affirmative action programs.

The key findings by Reardon et al. are summarized in Table 1 below.  Perhaps their most important result is that, if top colleges used “strong” class-based consideration without race-based consideration, the percentage of African Americans and Latinos in the student body would be less than half as great (10.2 percent) compared to a strong race-conscious program without class-based consideration (26.0 percent).  In fact, the strong class-based simulation yielded almost a third fewer Black and Latino students (10.2 percent) than the “moderate” race-conscious simulation (14.9 percent) and the moderate class-based simulation resulted in half as many Black and Latino students (7.5 percent) as the moderate race-based simulation (14.9 percent).

Table 1:  Composition of Top Colleges: Various Affirmative Action Scenarios, Reardon et al. 2015 (2004 Data)²⁰

Sigal Alon’s recent book complements the aforementioned study in key respects, for she too uses a robust set of admission simulations to test whether class-based affirmative action can effectively substitute for race-conscious admissions at America’s “most competitive” and “highly competitive” colleges.²¹  Professor Alon’s representative data set is older (it includes students entering higher education in 1995–96), but her data allows for an unusually robust analysis of socioeconomic alternatives to race-conscious affirmative action.²²  Wealth is an important factor to analyze, since a number of scholars have suggested that accumulated family wealth is a better proxy than family income for true socioeconomic and educational disadvantage, especially for African Americans who have encountered longstanding discrimination in housing markets.²³  Alon is able to account for wealth by using data from students’ federal financial aid applications, including expected family contribution data that incorporates wealth, assets, and other benefits, adjusted for family size.²⁴  One asset of Alon’s work is that she stretches empirical modeling of class-based affirmative action beyond the standard fare that would be typically used to measure disadvantage, while, at the same time, not beyond the realm of what is possible for a committed and motivated university to operationalize.²⁵

To provide further description of her various models, Alon’s “economic” simulation is based on expected family contribution, her socioeconomic status (“SES”) model is expanded to include parental education levels, her “structural” model gives a plus factor for residing in a poor neighborhood or attending a high-poverty high school, and finally her “multidimensional” model blends the earlier SES and structural models such that students receiving a plus factor have registered socioeconomic disadvantage on multiple fronts (for example, low expected parental contribution and first-generation college student and  attended a high-poverty high school).  Despite the robustness of her SES and wealth measures, all of Alon’s economic simulations show declines in African American and Latino students’ enrollment compared to current enrollment levels at elite U.S. colleges and universities.  Alon concludes:

In sum, the student bodies of elite colleges would be substantially less diverse racially and ethnically under all types of class-based affirmative action relative to current race-based policy.  Admissions that focus on structural disadvantages generate more racial and ethnic diversity than those that target individual socioeconomic status.²⁶

Results from Alon’s simulations are summarized below in Table 2.

Table 2: Composition of Elite U.S. Universities:
Various Class-Based Affirmative Action Simulations,
Alon 2015 (1995 Data)²⁷

Alon’s results, displayed above, are essentially an exercise in empirical matching: She replaced race-based admits with counterparts under various socioeconomic-based criteria.  Later in her book, Alon performs an additional set of simulations in which she discontinues all consideration of legacy and donor preferences, athletic preferences, race-conscious admissions, etc., and then fills the bottom quarter of the class at elite colleges with class-based admits.²⁸  Under such a scenario, SES preferences yield almost the same share of underrepresented minorities as is currently the case at these elite schools (15 percent versus 16 percent).²⁹  But Alon readily acknowledges this latter scenario is much less realistic because it is extremely unlikely that elite colleges would be willing to absorb the necessary fiscal consequences of ending all alumni and athletic preferences while simultaneously tripling their funding of need-based financial aid.³⁰  In Part III of this Essay, I expand on this theme of the substantial budgetary constraints that would make it very difficult for leading U.S. private and public universities to adopt transformative change with respect to expanding admission and enrollment opportunities for low-income students.  For now, there is a second important caveat (one not expressly discussed by Alon) that limits the generalizability of Alon’s last finding: Even if one entertains heroic assumptions about overcoming the budgetary limitations of ending legacy and athletic preferences, the results are still limited to largely elite private universities and would have quite limited carryover application to America’s most highly selective public universities.³¹

The analyses by Reardon et al. and Alon (above in Tables 1–2) are consistent with other contemporary studies.  For example, using robust modeling in an attempt to simulate a nationwide top 10 percent class rank admissions program at selective colleges, Howell found declines in African American and Latino enrollments even under somewhat optimistic conditions.³²  Similarly, Espenshade and Radford modeled class-based alternatives at a group of elite private universities and found that socioeconomic status was not a suitable substitute for the inclusion of race within holistic admissions.³³  Carnevale and Strohl, in work published in 2010 and 2013, found that class-based affirmative action is not an effective substitute for race-conscious programs.³⁴  As have Xiang and Rubin (one of the country’s leading statistical methodologists) made similar findings in a national simulation of U.S. law schools.  Xiang and Rubin estimate that substituting class-based affirmative action would significantly erode African American enrollments at leading law schools without helping graduation and bar passage rates.³⁵  In addition, modeling simulations by Long show the upper-bound limits and inefficiencies of using proxies for race, including proxies that most courts would not deem race-neutral.³⁶

The studies discussed above are consistent with a body of earlier research that finds class-based programs are not an effective substitute for race-conscious affirmative action (see Appendix A for the following: Fryer et al. 2008; Long 2007; Krueger et al. 2006; Bowen et al. 2005³⁷; Carnevale and Rose 2004; Long 2004; Bowen and Bok, 1998; Cancian 1998; Kane 1998).  The crux of the problem is that, as noted in many of the earlier studies, although there is a meaningful positive correlation between race/ethnicity and socioeconomic status, the correlation is not so strong that class can effectively substitute for race.  Moreover, programs that can be designed to take class into consideration will tend to be swamped by the numerically far larger number of disadvantaged white and, to a lesser extent, Asian American students before substantial numbers of typically more disadvantaged Black and Latino students can be admitted at leading American universities.

I conclude this Part with one recent study by Carnevale, Rose, and Strohl³⁸ that, in some respects, is a bit of an outlier compared to the research discussed above, though it is cited in the Fisher II amicus briefs of Richard Kahlenberg and Richard Sander, who advocate replacing race-based with class-based affirmative action.  Carnevale et al. rely on a data set of high school seniors graduating in 2004 to model 193 top colleges and universities that, combined, represent a freshman class with 250,000 seats.³⁹  The authors’ models include socioeconomic-based admissions in different combinations with other factors like performance in the top tenth of one’s high school.⁴⁰  The richness of Carnevale, Rose and Strohl’s measures of socioeconomic disadvantage—including, among other factors,  neighborhood education and poverty levels, wealth, and relative opportunities to take advanced placement courses⁴¹—represents both a strength and a weakness.  Carnevale et al. conclude in this study that “‘race-blind’ and ‘race-conscious’ (giving an added boost to underserved minorities) forms of affirmative action can substitute for the use of ‘race alone’ in college admissions.”⁴²

But there are several important shortcomings in this study by Carnevale et al.  First, unlike the Reardon study discussed earlier, Carnevale et al. rely upon the unrealistic assumption that all qualified disadvantaged students would apply to and choose to enroll at America’s top colleges.⁴³  This assumption runs contrary to a large body of real world data and scholarship documenting that economically disadvantaged students, especially Latinos, exhibit far more bounded application and enrollment choice patterns, sometimes referred to as “under-matching”.⁴⁴  This phenomena was observed in Texas after the Ten Percent Plan admissions guarantee.⁴⁵  Carnevale et al.’s assumption likely results in an overstatement of differences between the status quo and their enrollment models.⁴⁶  Second and relatedly, Carnevale et al.’s models, including their “pure merit” baseline,⁴⁷ assume away the existence of athletic preferences, legacy preferences, etc., and therefore face the same feasibility problems as the less realistic set of Alon’s “reform” simulations described above.  Third,  Carnevale et al. hypothesize a version of a nationwide Ten Percent Plan, which would be impossible to operationalize.⁴⁸  Finally and perhaps most importantly, Carnevale, Rose, and Strohl assume that colleges in tandem with the federal government⁴⁹ will commit the financial resources necessary to allow for a radical increase in enrollment opportunities for low-income students at leading American colleges and universities (see discussion above, and below in Part III).  The takeaway, given these mutually-reinforcing limitations, is that the usefulness of the Carnevale et al. study is limited to learning from the comparisons between their different simulation models, and not from comparing their simulations with the current reality of American higher education.⁵⁰  Still, it is Carnevale et al.’s simulation that combines consideration of race with multiple other factors including SES that yields the highest proportion of African Americans and Latinos  compared to their other simulations.⁵¹

In summary, recent empirical studies focusing on national simulations—including Reardon et al., 2015; Alon, 2015; Xiang and Rubin, 2015; Howell, 2010; and Espenshade and Radford, 2009—consistently find that socioeconomic status is not a satisfactory race-neutral alternative to race-conscious measures at selective U.S. higher education institutions.  The exceptions to this pattern, such as Carnevale et al., 2014, deploy less realistic modeling assumptions and therefore do not seriously call into question the conclusion that class-based approaches are not a workable substitute for race-conscious efforts at the undergraduate level.

II.  Studies of Particular States and Universities

In Grutter v. Bollinger,⁵² the Supreme Court noted that “Universities in California, Florida, and Washington State, where racial preferences in admissions are prohibited by state law, are currently engaged in experimenting with a wide variety of alternative approaches.  Universities in other States can and should draw on the most promising aspects of these race-neutral alternatives as they develop.”⁵³  Today, there is a large body of scholarship on the impact of affirmative action bans and related race-neutral efforts in higher education.  Outcomes in Texas universities after the Hopwood v. Texas⁵⁴ decision and enactment of the Ten Percent Plan have been studied more exhaustively than plans in any other state, including scores of papers and articles generated through a decade-long project at Princeton and a number of pieces coauthored by top social scientists.  There is also a smaller but still sizeable body of research analyzing various aspects of California’s experience under a state law affirmative action ban.  There are fewer studies of Washington (such as, Brown and Hirschman, 2006) and Michigan (such as, Kidder 2013), and, due to data limitations, there is little research on Florida beyond descriptive statistics.⁵⁵  So, in this Part, I focus mainly on Texas and California.

In order to contextualize the studies discussed immediately below, it is important to recognize that Texas and California have undergone substantial demographic transformations since affirmative action was initially banned twenty years ago.  In 1996—the year of the Hopwood ruling in Texas and Proposition 209 in California—Latinos made up 29 percent of the high school graduating class in Texas and 30 percent in California.⁵⁶  By 2014, Latinos made up 46.8 percent of the high school graduates in Texas and 48.4 percent of public high school graduates in California,⁵⁷ which was accompanied by substantial corresponding declines in the proportion of white high school graduates in these two states.

A second foundational point is that, to the extent that the Texas Ten Percent Plan admissions guarantee has some positive effects on racial diversity, this is so by virtue of the persistence of a high degree of neighborhood and K-12 school segregation (Tienda & Niu, 2006; Adams 2001).  This fact alone does not condemn the Ten Percent Plan—even the authors of the Ten Percent Plan were “dismayed” by calls “to use percentage plans as a rationale for discarding affirmative action”⁵⁸—but it signifies a conflict with our democratic and social policy ideals if the nominally race-neutral Ten Percent Plan has some success largely because half of Latinos and 40 percent of African Americans in Texas attend K-12 schools made up of 90–100 percent minority students.⁵⁹  The Fifth Circuit recognized this feature as a weakness of the Ten Percent Plan.⁶⁰

In examining the question of race-neutral alternatives at UT Austin and other selective Texas institutions, the briefs and earlier judicial opinions in Fisher tend to rely on descriptive admission and enrollment statistics rather than more robust studies and statistical analyses, perhaps because there never was a district court trial in Fisher.  There are, however, scores of academic papers that take a more sophisticated look at various aspects of the Texas Ten Percent Plan,⁶¹ so the review below can only highlight some key points emerging from this large literature.

Harris and Tienda’s 2012 study focusing on Latinos found that their application rates to UT Austin and Texas A&M fell after Hopwood banned affirmative action and under the Ten Percent Plan “owing to rapid growth in the number of high school graduates, their disadvantage in percent of applicants relative to whites grew over time.”⁶²  Admission rates for Latinos were also reduced during the Ten Percent Plan years.⁶³  A 2010 study by Harris and Tienda found, after controlling for many factors, that, even several years after the Texas Ten Percent Plan, African American and Latino enrollments at UT Austin and Texas A&M had not been restored to the levels existing before the affirmative action ban in Texas.⁶⁴  While the Ten Percent Plan appears to have benefits such as broadening the pool of Texas high schools sending students to UT Austin,⁶⁵ the benefits of geographic diversity have limited relevance to the question of whether race-neutral alternatives can substitute for race.

So what is the big picture regarding the efficacy of the Texas Ten Percent Plan?  One revealing finding from Horn and Flores’ 2012 study was that, over a twelve-year span, 60 percent of white students eligible for the admissions guarantee enrolled at one of the two public flagships (UT Austin or Texas A&M at College Station), compared to only 45 percent of eligible Latinos and 30 percent of eligible African Americans.⁶⁶  Later, in their 2015 review, Flores and Horn concluded as follows:

Analyses of data over the last 12 years suggest two stories related to percent plans effectiveness, at least in Texas.  First, the soaring numbers of the non-White population, particularly Latinos, among college-eligible students may be leading to inaccurate praise of not only the percent plan but other related state college enrollment policies as well.  More accurate assessments of percent plan effectiveness that account for these demographic realities indicate that Latinos are less likely to go to college despite their heavily increasing share of the high school graduate population.  The second story of the percent plan in Texas is where eligible students choose to go to college.  As noted, the data suggest that underrepresented students who are percent-plan-eligible are more likely to enroll in a nonselective . . . institution, a choice likely influenced by concentrated disadvantage both economically and in the high schools attended.⁶⁷

Experience with the Texas Ten Percent Plan yields a couple of other conclusions worth mentioning.  The first is that numerous studies document that financial aid such as UT Austin’s Longhorn scholars program and Texas A&M’s Century Scholars program are critical in prompting Black and Latino students to enroll,⁶⁸ which underscores the importance of looking at the higher financial aid costs of race-neutral and class-based alternatives (discussed below in Part III of this Essay).

A second point is that admitting more and more of the freshman class based on one factor alone (high school rank) may be showing signs of strain in terms of minority students’ later academic outcomes at the leading public universities in Texas.⁶⁹  The share of UT Austin’s in-state freshmen admitted via the Ten Percent Plan ballooned from 41 percent in 1998 to 81 percent in 2008, the year Abigail Fisher applied.⁷⁰  At UT Austin, there was a six point black-white gap in six-year graduation rates among the first several Ten Percent Plan cohorts (freshmen entering in 1998–2001) and a nine point Latino-white gap.  Yet for UT Austin’s 2005–08 freshmen classes, the Black-white gap in graduation rates ballooned to fifteen points and the Latino-white gap grew to thirteen points.⁷¹  At Texas A&M—which has selectivity on par with UT Austin but did not use race as a plus factor after Grutter and which tends to struggle with low proportions of African Americans—there was an eighteen point black-white graduation rate gap in 2005–08 and a ten point Latino-white gap.⁷²  These statistics do not necessarily reflect causation, but they are consistent with a point UT Austin and its amici made in defending the race-conscious element of the holistic admissions plan at issue in Fisher (including during Fisher II oral argument in counsel’s response to the late Justice Scalia’s question about mismatch): Holistic admissions allows the university to admit more minority students possessing indicators of likely college success.⁷³  These data provide one more reason to caution judges and policymakers against reflexively dismissing the overwhelming consensus of amicus briefs filed in Fisher I and II by American universities, higher education associations, leading corporations, and the military, emphasizing the importance of holistic full-file applicant review inclusive of the consideration of race along the lines of what the Court’s majority endorsed in Grutter.⁷⁴  As an additional example from one of the national studies discussed earlier, Alon found that the beneficiaries of race-conscious admissions at the mostly-private elite universities she studied had notably higher graduation rates (73 percent) than did the SES-disadvantaged students with similar credentials who enrolled at less selective schools (65 percent).⁷⁵

Turning from Texas to California and the impact of the Proposition 209 ballot initiative in the Golden State, in a recent policy paper for Educational Testing Service (ETS), Patricia Gándara and I performed a wide-ranging review of the University of California system (UC).  We examined UC’s sponsored K-12 academic preparation programs, UC’s freshman and transfer undergraduate admissions outcomes and policies, and UC’s aggressive need-based undergraduate financial aid policy, as well as looking at UC professional schools.  We concluded that UC’s efforts experimenting with a myriad of robust race-neutral admissions policies show that they do not offer a panacea.  These alternatives are highly costly, inefficient, and produce discouraging results, especially at the most highly selective UC campuses.⁷⁶  UC’s President and Chancellors reached similar conclusions in their amicus briefs in Fisher I and II and these themes were amplified in briefs by the State of California and UC student organizations.

Chart 1 below displays African Americans’ share of freshman admission offers for the UC system, UC Berkeley, and UCLA (enrollment figures are similar).  Proposition 209, which took effect in 1998, was accompanied by significant drops in African American admissions across UC.  At Berkeley and UCLA, there were dramatic declines of 55 percent in admission offers to African Americans between the pre-209 baseline of 1995 and 1998.  The low point for UC Berkeley was 2004, when only 2.8 percent of freshman offers went to African Americans.  At UCLA, the post-209 low point was reached in 2006, when UCLA had the fewest African American freshmen of any year since Nixon was in the White House.  In 2007, UCLA responded with a more holistic but still race-neutral approach to comprehensive review of applicant files.  This made a modestly positive contribution at UCLA, but it did not bring African American enrollments even close to pre-209 levels.

Chart 1: African Americans as a Percentage of California Public High School Graduates & UC’s In-State Freshmen Admits, 1995 to 2014⁷⁷

African American admissions (and enrollments) at UC Berkeley and UCLA have typically hovered under 4 percent in recent years, far below the 7 percent figure in 1995.  American Indian admissions remain far below pre-209 levels.  For Latinos at UC Berkeley and UCLA, it was only in 2014 that admission levels came back to the pre-209 baseline of 1995, and that was because Latino public high school graduates soared from 30 percent to almost half of California’s total graduates during that span.  These basic patterns of declines in UC admissions and enrollments at highly selective UC campuses are confirmed in a number of studies.⁷⁸

A couple of other notable studies are Gaertner and Hart’s 2013 study of class-based affirmative action at the University of Colorado-Boulder (CU-Boulder) and Kahlenberg and Potter’s 2012 report for the Century Foundation, which are cited multiple times in the Fisher Supreme Court and Fifth Circuit briefs.  Plaintiff Abigail Fisher, for example, references Gaertner and Hart for her argument to the Fifth Circuit that “by intensifying its focus on socioeconomic and related factors that are already part of UT’s race-neutral admissions system, the university could secure all the diversity gains achieved through racial preferences, advance other valuable interests, and avoid the constitutional and societal costs that racial discrimination imposes.”⁷⁹

The main problem with the Gaertner and Hart study⁸⁰—which found that class-based consideration could be a decent substitute for race-based considerations at CU-Boulder—is not with the methodological design and details but with its lack of generalizability.  The authors concede that most of the “research on class-based policies has shown these policies to be poor substitutes for race-conscious admissions” and that their findings are an outlier compared to the existing research literature.⁸¹  They also note that CU Boulder is not a very selective school (84 percent freshman acceptance rate), that it employed relatively weak consideration of race before moving to race-neutral alternatives, and it has never had a high proportion of minorities in its student body.  Thus, in the Gaertner and Hart study, African Americans, Latinos and Native Americans combined were only 13 percent of CU-Boulder freshmen, which is about half of their proportion among Colorado high school graduates.⁸²  Moreover, CU-Boulder is perennially among the lowest of the sixty U.S. members of the elite Association of American Universities (AAU) in terms of African American enrollment at about 1.6 percent (only Caltech and UC San Diego are comparably low).⁸³  Finally, among the AAU public universities, CU-Boulder has among the lowest combinations of per pupil state and institutional gift aid (part of the state disinvestment theme discussed in Part III below).

In light of these background factors associated with CU-Boulder’s relatively low baseline diversity, its low baseline need-based financial aid, and its lower selectivity level, the Gaertner and Hart study is not a sound basis for drawing any conclusions about the workability of class-based approaches at most leading American universities.  Tellingly, after Colorado voters rejected an anti-affirmative action ballot initiative in 2008, CU-Boulder decided the most effective approach was to adopt a “class-plus-race” system rather than replacing race-sensitive admissions.⁸⁴

Perhaps the best known and most cited piece of empirical advocacy for class rather than race-based admissions is Kahlenberg and Potter’s “A Better Affirmative Action” report.  But compared to the peer-reviewed research discussed earlier in this Essay, this non-peer reviewed report is plagued by misleading claims.  Moreover, as the NAACP Legal Defense Fund and other civil rights groups have pointed out, the Kahlenberg and Potter report presents a false choice between race-conscious and class-conscious approaches.⁸⁵  Kahlenberg and Potter’s central empirical claim, one made with an eye toward policy impact, is captured in this summary of how they see their study’s findings:

The evidence presented in . . . this paper suggests that it is possible to produce a critical mass of African American and Latino students in leading universities without resorting to racial preferences per se.  Indeed, at seven of the ten universities examined using race-neutral plans—UT Austin, Texas A&M, the University of Washington, the University of Florida, the University of Georgia, the University of Nebraska, and the University of Arizona—the representation of African Americans and Latinos met or exceeded the levels achieved when the universities had used racial preferences.⁸⁶

Here are many of the reasons Kahlenberg and Potter’s claims are inaccurate and/or misleading:

• At Texas A&M, even Kahlenberg and Potter’s own graph shows that African American freshmen enrollments never returned to 1995 (pre-Hopwood) levels during the entire 1998–2010 period without affirmative ⁸⁷ Instead, in the text the authors are able to make this claim because they chose to use 1996 as their baseline year, yet it is well-documented that the March 1996 Hopwood ruling had an immediate chilling effect on admissions and yields for the 1996 incoming cohort even though the ban was not in effect for a full cycle until 1997 (see Appendix A regarding Finnell 1998; Dickson 2006; MALDEF et al., 2005).
• Regarding Texas A&M and Latinos, using 1995 as the baseline, it took thirteen years before Latino enrollments returned to pre-Hopwood levels in ⁸⁸ During that time, Latino high school graduates in Texas rose considerably, from about 27 to 37 percent of all graduates.  The studies by Flores and Horn discussed earlier in this Part document the shrinkage in opportunities at Texas A&M for underrepresented minority students.
• At UT Austin when race-conscious plus factors were not considered in admissions from 1997–2004, the university struggled to maintain minority enrollments, with African Americans representing only 3 or 4 percent of the freshman class, compared to 5 percent in ⁸⁹ It was only in the era after Grutter when UT Austin could combine the Ten Percent Plan and race-sensitive holistic admissions that African American enrollments met or exceeded the pre-Hopwood baseline for several consecutive years beginning in 2005.⁹⁰
• At UT Austin, it was not until 2003 that Latino freshman enrollments met and exceeded the pre-Hopwood 1995 baseline of 15 percent, and, by that time, Latino public high school graduates swelled above a third of the state total in ⁹¹ As noted earlier in this Essay, the steady rise in Latino high school graduates is not a basis for attributing university enrollment successes to the Texas Ten Percent Plan.⁹²
• Regarding the University of Florida (UF), the authors acknowledge but do not fully appreciate the import of the fact that the Governor’s “One Florida” initiative prohibited affirmative action in admissions but did not restrict race-conscious financial aid and other recruitment ⁹³ Marin and Lee, who studied in depth what happened in Florida, saw these factors as crucial at UF: “At UF, while they have lost race-conscious affirmative action in admissions, they heavily use race-targeted programs to address minority recruitment and practically ignore the Talented 20 Program . . . . It is the race-targeted recruitment, aid, and support programs that are critical at UF.”⁹⁴ Notably, the UF’s recruitment-based diversity strategies seem to have floundered in recent years for reasons that may relate to the recession.  In 2007, UF enrolled a freshman class that was 14 percent African American , but by 2013 this had plummeted by more than half, to 6.2 percent of freshmen⁹⁵ and UF freshmen enrollments did not rebound in either 2014 (5.5 percent) or 2015 (6.1 percent).⁹⁶  These recent freshmen figures are indeed much worse for African Americans than the period between 1995 and 2000, prior to implementation of the “One Florida” and “Talented 20” initiatives banning affirmative action.  Within the UF undergraduate student body, there were about 1200 fewer African Americans in 2013 compared to 2009.⁹⁷  There are similar problems with Kahlenberg and Potter’s characterizations about Florida State University (mentioned in the text of their report, but not the quote above).⁹⁸
• Regarding the University of Arizona (UA), Kahlenberg and Potter acknowledge that “[i]t is unclear to what extent the ban changed undergraduate admissions at the state’s public universities”⁹⁹ but they nonetheless included Arizona as one of the seven success They based their claim on a form the UA submitted for the College Board common data set.¹⁰⁰  The College Board form, however, was prepared as a matter of routine by an associate analyst in the institutional research office and the analyst’s “check yes” response is contradicted by contemporaneous public statements by UA President Shelton that the University of Arizona did not use race in undergraduate admissions before the Proposition 107 ban.¹⁰¹
• At the University of Nebraska, the authors’ charts show that, in 2004, the year before the university ended consideration of race, African Americans were about 5 percent of freshmen and 5 percent of the state high school graduates.¹⁰² But in the years after race was no longer considered, African American freshmen enrollments fluttered between 2.1 percent and 2.9 percent, even as the share of high school graduates rose to 8 percent by 2010.

In light of all of the above, Kahlenberg and Potter’s “seven of the ten universities” claim is an empty, meaningless statistic that clutters the marketplace of ideas with respect to the discourse about race-neutral alternatives in U.S. higher education.  Moreover, in evaluating the empirical case that Kahlenberg and Potter seek to make for class rather than race-sensitive admissions, it is important not to lose sight of the three universities that, even they, do not count as success stories: UC Berkeley, UCLA, and the University of Michigan.  It is no accident that these are three of America’s most selective (and most prestigious) public universities, and are all located in states with large underrepresented minority populations.  At UC Berkeley, UCLA and the University of Michigan, African American enrollments were approximately 7 percent of the freshman class before affirmative action bans and are considerably below those levels today despite many robust and costly race-neutral efforts.

In summary, in this Part of the Essay I show that recent empirical studies focusing on particular states and universities generally find that percentage plan and socioeconomic disadvantage are not workable substitutes for consideration of race in college admissions—including the studies of Texas by Harris and Tienda (2010, 2012), Flores and Horn (2012, 2015) and Long (2010), and studies of California by Kidder and Gándara (2015) and Kurlaender et al. (2015).  There are distinguishable exceptions to this consensus, as the Gaertner and Hart findings (2013) are not generalizable given the unusual facts of CU-Boulder’s admissions.  Moreover, the pollyannaish claims about class-based alternatives in the Kahlenberg and Potter report (2012) were not peer-reviewed and do not stand up when subjected to careful scrutiny.

III.  The High Cost of Race-Neutral Financial Aid

In Fisher I, the Supreme Court declared that race cannot be considered when a “nonracial approach . . . could promote the substantial interest about as well and at tolerable administrative expense.”¹⁰³  One limitation with the studies summarized earlier is that even sophisticated admissions and enrollment simulations do not account for the large augmentations in financial aid that would be required for class-based considerations in admissions to effect meaningful change in U.S. higher education enrollment patterns.¹⁰⁴  In short, class-based affirmative action programs are substantially more expensive than race-conscious ones.¹⁰⁵  In this Part, I review scholarship and descriptive statistics regarding the financial consequences of class-based admission approaches at both leading private and public universities.

While elite private universities have more per capita resources and endowments than leading public universities, the competitive institutional market dynamics that result in private college tuition-setting and financial aid equilibria would make it difficult to truly move the needle with large-scale class-based affirmative action programs for undergraduates.  As the President of Vassar recently stated (rather bluntly) in response to a New York Times inquiry, “Talented, low-income kids are out there . . . . But the problem for schools is when you admit one of those kids, you forgo $50,000 a year that you could use for other things.”¹⁰⁶

William Bowen and his colleagues estimated, using 1995 data, that a moderate shift to a socioeconomic-based plus factor in admissions at highly selective private institutions, would necessitate an increase in grant aid of about 12 percent.¹⁰⁷  How much would such an investment cost today at elite private U.S. universities?  Among America’s leading research-intensive universities there are thirty-four private universities where low-income students represent fewer than a quarter of the student body (average of 15 percent Pell Grant students).  Undergraduates at these institutions currently receive a grand total of $4.67 billion in grant aid (2013 data), with roughly 95 percent of aid coming from the institution’s own funds.¹⁰⁸  A 12 percent increase or redistribution of grant aid at these private universities would cost on the order of $561 million.

Below, Chart 2 displays cost estimates if each of these thirty-four private universities increased (or shifted) 12 percent of grant aid to support a modest class-based admissions plan in the manner described by Bowen et al.  The price tag varies with the size of a school’s student body and existing gift aid commitments, so applying the 12 percent figure across-the-board will overshoot or undershoot the mark in specific cases compared to more nuanced estimates.  For example, the University of Southern California (USC) already enrolls more low-income students (23 percent) than the other universities in this group, whereas Washington University in St. Louis enrolls the fewest low-income students (6 percent).  For these and other reasons, the estimates in Chart 2 are only intended to provide an impressionistic sense of scale.  It must be emphasized that these are likely lower-middle estimates compared to some of the more aggressive socioeconomic enrollment models discussed above (especially Carnevale et al. 2014).  And these estimates do not include other indirect costs that may be significant, such as pre-college and campus academic and social support programs in order to prevent graduation rates from flagging.

Chart 2: Annual Cost of Adding/Shifting 12 Percent in Gift Aid to Support Class-Based Admissions at Elite Private Universities (2013 data)¹⁰⁹

At the private institutions displayed in Chart 2, the average annual cost of adding 12 percent more gift aid to support class-based admissions would be about $16.5 million per university annually.  At Harvard, a 12 percent increase in grant aid would cost $21 million, an amount equal to all of Harvard’s annual federally-funded research expenditures in the social sciences.¹¹⁰  At a smaller elite university like Tufts, the cost would be about $6.5 million, but viewed in the context of Tufts’s modest undergraduate population and tuition resource base of only 5000 students, this too would pose a considerable fiscal burden.  Though not displayed in Chart 2, Bowen et al. likewise estimated that private liberal arts colleges would need to increase grant aid by about 12 percent to support modest class-based admissions policies,¹¹¹ so highly selective liberal arts colleges like Williams, Oberlin, and Pomona would, on a per capita basis, likely face roughly similar fiscal challenges.

Turning to leading public universities, the prospect of paying for substantially enlarged financial aid for low-income students is even more daunting.  In fact, Reardon et al. conclude it “would be very costly—infeasibly costly—for most public universities.”¹¹²  In the current environment there are, to borrow a phrase from the Supreme Court’s race jurisprudence, “built-in headwinds”¹¹³ pushing against substantial increases in need-based financial aid for low-income students at leading public universities.  In per capita inflation-adjusted terms, the states invested 20 percent less in higher education in 2014 compared to 2008 when the recession struck, and thirty-one states cut funding to higher education by more than 20 percent.¹¹⁴  State-level disinvestment in public universities, moreover, is a deep and long-term trend that was underway well before the great recession.  For example, in California, the state’s funding to the University of California between 1991 and 2015 progressed at less than half the rate of inflation.¹¹⁵

State disinvestment has exerted strong pressure on U.S. public universities to raise tuition,¹¹⁶ and long-term state disinvestment trends prompting higher tuition and/or states’ allocation priorities toward “merit” aid (and away from need-based aid) have adversely effected enrollment opportunities for underrepresented minority and low-income students at public universities.¹¹⁷  Moreover, state-level disinvestment has incentivized public universities to aggressively recruit and enroll more and more out-of-state and international students as an offsetting strategy for generating revenue.¹¹⁸

The University of Michigan at Ann Arbor (UM) is a paradigmatic example of these budget-driven enrollment trends.  UM operates under a state law affirmative action ban passed in 2006, and while it has employed numerous race-neutral alternatives in an effort to lessen the decline in racial diversity,¹¹⁹ the University of Michigan has not substantially boosted its enrollment of low-income undergraduates. Only 16 percent of UM undergraduates receive Pell Grants, lower than at most top publics.¹²⁰  For budgetary reasons, UM has increased the proportion of out-of-state undergraduates it serves so that today over two-fifths of Michigan’s undergraduates are domestic out-of-state students who must and can pay twice as much tuition ($40,000 versus $20,000).  The school’s dependency on out-of-state students for revenue is reflected in the fact that although 55 percent of UM-Ann Arbor’s undergraduates are in-state residents, the grant aid contribution made by the State of Michigan is extremely meager: $660,000, compared to the $37.2 million of its own funds that UM commits to grant aid for new students.¹²¹

Michigan represents a natural experiment disconfirming the suggestion that leading public universities can or will significantly enhance class-based financial aid and admissions in tandem as a race-neutral alternative.  In this vein, the University of California represents a natural experiment of a different sort, pointing to the upper limits of what elite public universities and state lawmakers are willing to sustain financially (also note that UC’s policy redirecting substantial tuition to need-based aid is longstanding and pre-dated Proposition 209).  In assessing the situation at the University of California in comparison to peer institutions, Professor Gándara and I conclude in a recent paper:

The vast majority of highly selective American universities and state lawmakers do not have the stomach to financially tolerate the UC’s level of institutional commitment to low-income students.  Among the 60 elite U.S. institutions in the Association of American Universities (AAU) there are six UC campuses (UC Berkeley, UCLA, UC San Diego, UC Davis, UC Irvine, UC Santa Barbara).  Ranking all 60 AAUs based on the proportion of undergraduates who are Pell Grant recipients in 2013 (i.e., low-income), the UC campuses rank 1st, 2nd, 3rd, 4th, 5th, and 8th.  Among the public universities in the AAU, only Arizona and Michigan have per capita institutional gift aid on par with UC campuses.  Moreover, state gift aid is another important component, and students at the UC campuses in the AAU receive several times more per capita state gift aid as do the students at the other 28 public AAU universities . . . . In light of all of the above factors in combination, UC effectively represents an upper-bound limit on commitment to class-based alternatives to affirmative action at highly selective American universities.¹²²

Conclusion

The weight of social science research, as outlined in this Essay, supports the conclusion that socioeconomic status is not an effective alternative to race-conscious measures with respect to undergraduate diversity at selective colleges and universities in the United States.  In addition, the very high cost of socioeconomic-based approaches (due to the combination of increasing financial aid commitments and foregoing tuition revenues) is difficult to reconcile with the U.S. Supreme Court’s consideration of “tolerable administrative expense.”  In the process of reviewing the empirical studies and financial data discussed above, I’ve emphasized that increasing socioeconomic diversity can be supported for social policy reasons separate from questions about race-neutral alternatives—in other words race and class need not be presented as a false choice.  Regarding this last point, much more important than anything I can say about the topic is what American colleges and universities choose to do in the real world.  Recently the American Council on Education (ACE) administered a first-of-its-kind national survey of U.S. undergraduate admissions and enrollment management officials at 338 four-year colleges and universities (combined enrollment at these schools was 2.7 million students).¹²³  The ACE admissions survey found that, among institutions that employ race-conscious measures, 74 percent consider socioeconomic disadvantage in the admissions process, but at institutions that do not consider race, only 27 percent look at socioeconomic disadvantage in admissions.¹²⁴  The false choice of race or class is also belied by recent research indicating that colleges and universities with higher levels of race and socioeconomic diversity can uniquely leverage educational benefits of diversity for their students.¹²⁵

 

Appendix A:  Additional Sources Noted in This Essay

Michelle Adams, Isn’t It Ironic? The Central Paradox at the Heart of “Percentage Plans”, 62 [small-caps]Ohio State L.J.[end-small-caps] 1729 (2001).

Susan K. Brown & Charles Hirschman, The End of Affirmative Action in Washington State and Its Impact on the Transition From High School to College, 79 [small-caps]Soc. Educ.[end-small-caps] 106 (2006).

Maria Cancian, Race-Based Versus Class-Based Affirmative Action in College Admissions, 17 [small-caps]J. Pol’y Analysis & Mgmt.[end-small-caps] 94 (1998).

David Card & Jesse Rothstein, Racial Segregation and the Black-White Test Score Gap, 91 [small-caps]J. Pub. Econ.[end-small-caps] 2158, 2160 (2007).

Anthony P. Carnevale & Stephen J. Rose, Socioeconomic Status, Race/Ethnicity, and Selective College Admissions, in [small-caps]America’s Untapped Resource: Low-Income Students in Higher Education[end-small-caps] (Richard D. Kahlenberg ed., 2004).

Anthony P. Carnevale & Jeff Strohl, How Increasing College Access Is Increasing Inequality, and What to Do About It, in [small-caps]Rewarding Strivers[end-small-caps] (Richard D. Kahlenberg ed., 2010).

[small-caps]Dalton Conley, Being black, Living in the Red: Race, Wealth, and Social Policy in America[end-small-caps] (1999).

Dalton Conley, The Why, What, and How of Class-Based Admissions Policy, in [small-caps]The Future of Affirmative Action: New Paths to Higher Education Diversity After Fisher v. University of Texas[end-small-caps] 203 (Richard D. Kahlenberg ed., 2014).

Lisa M. Dickson, Does Ending Affirmative Action in College Admissions Lower the Percent of Minority Students Applying to College?, 25 [small-caps]Econ. Educ. Rev.[end-small-caps] 109 (2006).

Susanna Finnell, The Hopwood Chill: How the Court Derailed Diversity Efforts at Texas A&M, in [small-caps]Chilling Admissions: The Affirmative Action Crisis and the Search for Alternatives[end-small-caps] 71 (Gary Orfield & Edward Miller eds., 1998).

Roland G. Fryer et al., An Economic Analysis of Color-Blind Affirmative Action, 24 [small-caps]J. Law & Econ. Org.[end-small-caps] 319 (2008).

Harry J. Holzer & David Neumark, Affirmative Action: What Do We Know?, 25 [small-caps]J. Pol’y Analysis & Mgmt.[end-small-caps] 463 (2006).

[small-caps]Richard D. Kahlenberg, The Remedy: Class, Race, and Affirmative Action[end-small-caps] (1996).

Richard D. Kahlenberg, Class-Based Affirmative Action, 84 [small-caps]Cal. L. Rev.[end-small-caps] 1037 (1996).

Thomas J. Kane, Misconceptions in the Debate Over Affirmative Action in College Admissions, in [small-caps]Chilling Admissions: The Affirmative Action Crisis and the Search for Alternatives [end-small-caps]17 (Gary Orfield & Edward Miller eds., 1998).

William C. Kidder, Policy Brief, Restructuring Higher Education Opportunity?: African American Degree Attainment After Michigan’s Ban on Affirmative Action (Aug. 2013) (unpublished manuscript), http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2318523 [https://perma.cc/BX5P-8H5Z].

Alan Krueger et al., Race, Income, and College in 25 Years: Evaluating Justice O’Connor’s Conjecture, 8 [small-caps]Am. L. & Econ. Rev.[end-small-caps] 282 (2006).

Mark C. Long, Race and College Admissions: An Alternative to Affirmative Action?, 86 [small-caps]Rev. Econ. & Stat.[end-small-caps] 1020, 1031–32 (2004).

Mark C. Long, Affirmative Action and Its Alternatives in Public Universities: What Do We Know?, 67[small-caps] Pub. Admin. Rev.[end-small-caps] 315 (2007).

Mark C. Long & Marta Tienda, Winners and Losers: Changes in Texas University Admissions Post-Hopwood, 30 [small-caps]Educ. Eval. & Pol’y Analysis[end-small-caps] 255 (2008).

Mex. American Legal Def. & Educ. Fund et al., Blend It, Don’t End It: Affirmative Action and the Texas Ten Percent Plan After Grutter and Gratz, 8 [small-caps]Harv. Latino L. Rev.[end-small-caps] 33 (2005).

[small-caps]Thomas M. Shapiro & Melvin L. Oliver, Black Wealth/White Wealth: A New Perspective on Racial Inequality[end-small-caps] (rev. ed. 2006).

Marta Tienda & Sunny Xinchun Niu, Capitalizing on Segregation, Pretending Neutrality: College Admissions and the Texas Top 10% Law, 8 [small-caps]Am. L. & Econ. Rev.[end-small-caps] 312 (2006).

 

[1].        Liliana M. Garces, Balancing Liberty and Equality: Justice Kennedy’s Decisive Vote in Fisher v. University of Texas, Part II, 44 [small-caps]Educ. Researcher[end-small-caps] 442 (2015).

[2].        See, e.g., [small-caps]Think Tank Research Quality[end-small-caps] 311–15 (Kevin G. Welner et al. eds., 2010); Rachel F. Moran, What Counts as Knowledge? A Reflection on Race, Social Science, and the Law, 44 [small-caps]L. & Soc’y Rev.[end-small-caps] 515 (2010).

[3].        Fisher v. Univ. of Tex. at Austin (Fisher I), 133 S. Ct. 2411, 2420 (2013).

[4].        Id. (quoting Grutter v. Bollinger, 539 U.S. 306 (2003)).

[5].        Id. at 2420 (quoting Wygant v. Jackson Bd. of Educ., 476 U.S. 267 (1986)).

[6].        Id. at 2421.

[7].        Fisher v. Univ. of Tex. at Austin (Fisher II), 758 F.3d 633, 649 (5th Cir. 2014), cert. granted, 135 S. Ct. 2888 (2015).

[8].        Id. at 656.

[9].        Id. at 657.  Here the Fifth Circuit is wrestling with the jurisprudential and sociological aspects of the “class not race” argument, themes which are the central focus of the Essay by Cheryl Harris in this symposium issue of the UCLA Law Review Discourse.  Cheryl I. Harris, Fisher’s Foibles, From Race and Class to Class Not Race, 64 [small-caps]UCLA L. Rev. Disc.[end-small-caps] (forthcoming 2016); see also Samuel R. Bagenstos, On Class-Not-Race (U. of Mich. Pub. Law & Legal Theory Research Paper Series, Paper No. 390, 2014), http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2404355.

[10].     438 U.S. 265 (1978).

[11].     The federal jurisdiction problems in the Fisher case are noted in University of Texas, Austin’s (UT Austin’s) various U.S. Supreme Court and lower court briefs.  See also Mario L. Barnes et al., Judging Opportunity Lost: Assessing the Viability of Race-Based Affirmative Action After Fisher v. University of Texas, 62 [small-caps]UCLA L. Rev.[end-small-caps] 272, 286–88 (2015).  Granting review only requires the vote of four Supreme Court justices, and if one assumes that the late Justice Scalia voted in favor of review, then it is possible that as few as three of the justices presently on the Court wanted to grant review in Fisher II.  Traditionally, the Court ends up with two or three “DIG” (dismissed as improvidently granted) cases per term.  See generally Michael E. Solimine & Rafael Gely, The Supreme Court and the DIG: An Empirical and Institutional Analysis, 2005 [small-caps]Wis. L. Rev.[end-small-caps] 1421 (2005).

[12].     The theme of “mismatch”—which came up during the Fisher II oral argument and in the amicus briefs—is beyond the scope of this Essay, but I have published on that topic elsewhere and Richard Lempert has an essay devoted to that topic in this symposium issue.  See William C. Kidder & Angela Onwuachi-Willig, Still Hazy After All These Years: The Data and Theory Behind “Mismatch”, 92 [small-caps]Tex. L. Rev.[end-small-caps] 895 (2014) (book review); Richard Lempert, Mismatch and Science Desistance: Failed Arguments Against Affirmative Action, 64 [small-caps]UCLA L. Rev. Disc.[end-small-caps] (forthcoming 2016).  Likewise, affirmative action at the professional or graduate school level is not something I have space to address in this Essay.

[13].     Peter Hinrichs, The Effects of Affirmative Action Bans on College Enrollment, Educational Attainment, and the Demographic Composition of Universities, 94 [small-caps]Rev. Econ. & Stat.[end-small-caps] 712, 715 (2012).

[14].     Fisher I, 133 S. Ct. at 2433 (Ginsburg, J., dissenting) (“I have said before and reiterate here that only an ostrich could regard the supposedly neutral alternatives as race unconscious.”); Brian Fitzpatrick, Is the Future of Affirmative Action Race Neutral? (Vanderbilt Pub. Law Research Paper No. 14-11, 2014), http://papers.ssrn.com/sol3/papers.cfm?abstract_id= 2426656.

[15].     I found this to be the best way to balance succinctness with social science norms, which place a premium on comprehensiveness, rigor, and the cumulative nature of evidence.

[16].     [small-caps]Sean F. Reardon et al., Can Socioeconomic Status Substitute for Race in Affirmative Action College Admissions Policies? Evidence From a Simulation Model[end-small-caps] (2015), http://www.ets.org/Media/Research/pdf/reardon_white_paper.pdf [https://perma.cc/PF3P-9V5C].  Their data set is called the “Education Longitudinal Study 2002” because it began when these students were high school 10th graders.  Id. at 27, 34.

[17].     Id. at 4–5, 17, 20, apps. B, C at 26–33.

[18].     Id. at 11.

[19].     Reardon et al. employed a common composite index for socioeconomic status.  Id. at 35 (“The family resource dimension is based on the ELS SES index, a composite measure of mother’s and father’s education, mother’s and father’s occupation, and family income.”).

[20].     Table 1 is a simplified adaptation of Reardon et al.’s Figure 2, id. at 12.

[21].     [small-caps]Sigal Alon, Race, Class and Affirmative Action[end-small-caps] 185–87 (2015) (ebook).  The “most competitive” and “highly competitive” categories are defined in the Barron’s college guide, a publication that a number of social scientists rely upon as a means of clustering colleges based on freshman selectivity.

[22].     Alon relies on the Beginning Postsecondary Students Longitudinal Study, which links to federal financial aid data in the national Postsecondary Student Aid Study.  Id. at 301 n.1.

[23].     [small-caps]Richard D. Kahlenberg & Halley Potter, The Century Found., A Better Affirmative Action: State Universities That Created Alternatives to Racial Preferences[end-small-caps] 18 (2012), https://www.tcf.org/assets/downloads/tcf-abaa.pdf [https://perma.cc/GRS3-XC4Z].  Additional works on black-white wealth disparities are included in Appendix A of this Essay.  See generally, e.g., [small-caps]Dalton Conley, Being Black, Living in the Red: Race, Wealth, And Social Policy in America[end-small-caps] (1999); Dalton Conley, The Why, What, and How of Class-Based Admissions Policy, in [small-caps]The Future of Affirmative Action: New Paths to Higher Education Diversity After Fisher v. University of Texas[end-small-caps] 203 (Richard D. Kahlenberg ed., 2014), https://tcf.org/assets/downloads/16_The-Why-What-and-How-of-Class-Based-Admissions-Policy.pdf; [small-caps]Thomas M. Shapiro & Melvin L. Oliver, Black Wealth/White Wealth: A New Perspective on Racial Inequality[end-small-caps] (rev. ed. 2006).

[24].     [small-caps]Alon[end-small-caps], supra note 21, at 182–84.  This financial aid form is called the Free Application for Federal Student Aid (FAFSA) and is widely used, not only for federal aid eligibility, but also for state and institutional aid sources.

[25].     For example, currently the wealth and other financial data from the FAFSA form is not available in time for freshman admission decisions at many colleges and universities.  The timing of the FAFSA deadline would have to be moved up to coincide with the admissions cycle—which in turn would require analysis of whether moving up the FAFSA deadline itself would have an adverse impact over time (and if so, how much?) on disadvantaged students, especially underrepresented minorities and those from low-income backgrounds.

[26].     Id. at 195.

[27].     Table 2 is a simplified adaptation (with permission from the publisher, the Russell Sage Foundation of New York, NY) of [small-caps]Alon[end-small-caps], supra note 21, at 194 fig.8.3.

[28].     Id. at 268–70.

[29].     Id. at 270 fig.11.1.

[30].     Id. at 270–72.  One illustrative example of elite university commitment to intercollegiate athletics is Yale University, where 15 percent of all undergraduates participate in athletics.  While Yale and other Ivy League schools have a special status in Division I which allows them to not provide athletic scholarships for student-athletes, Yale still expends over $38 million annually on its athletics program.  The Equity in Athletics Data Analysis Cutting Tool, [small-caps]Office of Postsecondary Educ., U.S. Dep’t of Educ.[end-small-caps], http://ope.ed.gov/athletics [https://perma.cc/C7T4-PRWQ].  Absent athletic preferences, Yale would not be able to compete in the Ivy League of NCAA Division I, which many Yale leaders would likely regard as harmful not only for student-athletes but to the overall student experience and to alumni relations.  Moreover, there are collective action barriers that would disincentivize Yale or any other similar university from defecting from Ivy League athletics and dropping down to a level such as NCAA Division III in which  athletic preferences would be far less consequential in admissions.

[31].     The institutions listed in Alon’s “elite college” category, derived from the Barron’s college guide, include seventy-two private institutions and ten public institutions, three of which  are U.S. military academies,  a distinctive niche apart from public flagships.  [small-caps]Alon[end-small-caps], supra note 21, at 296–97 n. 22.

My conclusion about the public/private distinction follows from three additional observations about how the enrollment profiles of elite privates and publics differ in verifiable ways:

1. Alumni and donor preferences are orders of magnitude more prevalent at elite private universities as compared to elite public universities (most selective publics do not have alumni preferences);
2. Athletic preferences at Ivy League and similarly elite private colleges are not necessarily helpful to underrepresented minorities, whereas at leading public universities with Division I athletics, the cessation of all athletic preferences would have a distinctly deleterious net impact on the overall proportion of African American enrolled Among 2005–08 freshman cohorts, African Americans represented only 5–6 percent of all scholarship athletes at Harvard, Yale, and Penn, compared to 19–27 percent at University of California (UC) Berkeley, UC Los Angeles (UCLA), and UT Austin, though the same is not generally true for Latinos; and
3. Athletics has a larger relative influence on the total admitted class at many elite privates compared to elite publics because these smaller elite schools must enroll high proportions of student-athletes in order to mount athletic For example, student-athletes make up 15 percent of Yale’s undergraduate student body versus only 1.5 percent of the study body at a large flagship university like UT Austin.  By implication, cessation of plus factors for athletes at elite privates (where they tend to benefit white students) has a beneficial practical impact for minority students’ opportunities at elite privates as compared to selective publics.

As an implication of the points above, at elite U.S. public institutions the combination of far smaller (and, at many schools, non-existent) alumni preferences and the higher negative racial dividend of ending athletic considerations means that the positive diversity benefits of Alon’s Figure 11.1 “reform” simulation would not obtain with respect to elite public flagship universities with few exceptions.  The athletics statistics cited above are from GSR Homepage, [small-caps]Nat’l Collegiate Athletics Ass’n.[end-small-caps] http://web1.ncaa.org/GSRSearch/exec/homePage [http://perma.cc/SZS3-7M72] [hereinafter NCAA], and The Equity in Athletics Data Analysis Cutting Tool, supra note 30.

[32].     Jessica S. Howell, Assessing the Impact of Eliminating Affirmative Action in Higher Education, 28 [small-caps]J. Labor Econ.[end-small-caps] 113, 116, 150–52 (2010).

[33].     [small-caps]Thomas J. Espenshade & Alexandria Walton Radford, No Longer Separate, Not Yet Equal: Race and Class in Elite College Admission and Campus Life[end-small-caps] 361–64 (2009).

[34].     Anthony P. Carnevale & Jeff Strohl, How Increasing College Access Is Increasing Inequality, and What to Do About It, in [small-caps]Rewarding Strivers[end-small-caps] 71, 165 (Richard D. Kahlenberg ed., 2010); [small-caps]Anthony P. Carnevale & Jeff Strohl, Separate and Unequal: How Higher Education Reinforces the Intergenerational Reproduction of White Racial Privilege[end-small-caps] 37 (2013), https://cew.georgetown.edu/wp-content/uploads/2014/11/SeparateUnequal.FR_.pdf [https://perma.cc/4Q7C-KP2Y].

[35].     Alice Xiang & Donald B. Rubin, Assessing the Potential Impact of a Nationwide Class-Based Affirmative Action System, 30 [small-caps]Stat. Sci.[end-small-caps] 297 (2015).  This study relies on somewhat dated data: the LSAC Bar Passage Study (1990–91 law school applicants graduating and taking the bar in 1994 and 1995) that has been extensively analyzed by numerous authors.  Yet, a newer equivalent to the Bar Passage Study has not yet been made available to academic researchers.

[36].     See generally [small-caps]Mark Long, The Promise and Peril for Universities Using Correlates of Race in Admissions in Response to the Grutter and Fisher Decisions[end-small-caps] (2015), http://www.ets.org/Media/Research/pdf/long_white_paper.pdf [https://perma.cc/F6ML-X2FU]; Mark C. Long, Is There a “Workable” Race-Neutral Alternative to Affirmative Action in College Admissions?, 34 [small-caps]J. Pol’y Analysis & Mgmt.[end-small-caps] 162 (2014).

[37].     This work is referenced in the Financial Aid Part (Part III) rather than the Appendix.

[38].     Anthony P. Carnevale et al., Achieving Racial and Economic Diversity With Race-Blind Admissions Policy, in [small-caps]The Future of Affirmative Action: New Paths to Higher Education Diversity After Fisher v. University of Texas[end-small-caps] 187 (Richard D. Kahlenberg ed., 2014).

[39].     Id. at 188, 193.

[40].     Id. at 191–93.

[41].     Id. at 193–94.

[42].     Id. at 188.

[43].     Id.  To be clear, the converse assumption (that there would be no change in elite university applicant and yield rate behavior under a class-based affirmative action system) is also flawed.  See [small-caps]Alon[end-small-caps], supra note 21, at 186.  The question is what constitutes reasonably bounded counterfactual simulations.

[44].     See [small-caps]Alon[end-small-caps], supra note 21, at 184–85; [small-caps]William G. Bowen et al., Crossing the Finish Line: Completing College at America’s Public Universities[end-small-caps] 106–08, 208–16 (2009); [small-caps]Reardon et al.[end-small-caps], supra note 16, at 5; Caroline M. Hoxby & Christopher Avery, The Missing “One-Offs”: The Hidden Supply of High-Achieving Low Income Students (Nat’l Bureau of Econ. Research, Working Paper No. 18586, 2012), http://www.nber.org/papers/w18586.pdf [https://perma.cc/94ZR-CJXY].

[45].     Catherine Horn & Stella M. Flores, When Policy Opportunity Is Not Enough: College Access and Enrollment Patterns Among Texas Percent Plan Eligible Students, 3 J. Applied Res. on Child. 1, 16–17 (2012); Mark C. Long & Marta Tienda, Winners and Losers: Changes in Texas University Admissions Post-Hopwood, 30 [small-caps]Educ. Evaluation & Pol’y Analysis[end-small-caps] 255, 266–67 (2008).

[46].     One indicator of the lack of realism undergirding this assumption is that Asian Americans in the “status quo” condition represent 15 percent of freshmen at top colleges, Carnevale et al., supra note 38, at 192 tbl.15.1, but under all of the Carnevale et al. simulations Asian Americans end up significantly worse off numerically speaking, regardless of how vastly different these simulations are from each other and including simulations based solely on SAT scores.  Such results are facially implausible, and are contrary to Professor Alon’s simulation results for Asian Americans, [small-caps]Alon[end-small-caps], supra note 21, at 283 tbl.A8.2, and experience of low-income Asian Americans in Texas and California.  In two of the other Essays in this symposium volume, Kim West-Faulcon and Nancy Leong explore ways in which Asian Americans are positioned in misleading and questionable ways by legal and political opponents of affirmative action.  Kimberly West-Faulcon, Obscuring “Asian Penalty” With Illusions of “Black Bonus”, 64 [small-caps]UCLA L. Rev. Disc.[end-small-caps] (forthcoming 2016); Nancy Leong, The Misuse of Asian Americans in the Affirmative Action Debate, 64 [small-caps]UCLA L. Rev. Disc.[end-small-caps] 90 (2016).

[47].     Their definition of “pure merit” ranks students exclusively based on SAT/ACT scores.  Carnevale et al., supra note 38, at 188.

[48].     In different ways, these problems are flagged in the Fisher II amicus briefs submitted by Amherst and thirty-seven other private colleges and by 821 social science researchers.

[49].     Carnevale et al., supra note 38, at 202.

[50].     In this context, Carnevale et al. reached the following findings, ranked from the model with the lowest to highest racial diversity (percentages of African Americans and Latinos are listed in parentheses): (a) admission solely by SAT/ACT scores (1 percent, 4 percent); (b) SAT/ACT scores with a socioeconomic status (SES) plus factor (3 percent, 10 percent); (c) SAT/ACT scores plus guaranteed admission if one’s test scores were in the top 10 percent of one’s high school (6 percent, 11 percent); (d) SAT/ACT scores plus guaranteed admission if one’s test scores were in the top 10 percent of one’s high school, and a SES plus factor (9 percent, 14 percent); and (e) guaranteed admission if one’s test scores were in the top 10 percent of one’s high school, a plus factor for SES and a plus factor for race (14 percent, 18 percent). Id. at 192–99.  The authors caution that some of their more robust SES and class rank simulations could have the net result of lowering graduation rates.  Id. at 201.

[51].     Id. at 199 tbl.15.7.

[52].     539 U.S. 306 (2003).

[53].     Id. at 342.

[54].     78 F.3d 932 (5th Cir. 1996).

[55].     [small-caps]Stella M. Flores & Catherine L. Horn, Texas Top Ten Percent Plan: How It Works, What Are Its Limits, and Recommendations to Consider[end-small-caps] 12 (2015), http://www.ets.org/Media/Research/pdf/flores_white_paper.pdf [http://perma.cc/ 3NW5-ZGE8].

[56].     [small-caps]Catherine L. Horn & Stella M. Flores, The Civil Rights Project, Percent Plans in College Admissions: A Comparative Analysis of Three States’ Experiences[end-small-caps] 29 tbl.5 (2003), http://civilrightsproject.ucla.edu/research/college-access/admissions/percent-plans-in-college-admissions-a-comparative-analysis-of-three-states2019-experiences/horn-percent-plans-2003.pdf.

[57].     [small-caps]William C. Kidder & Patricia Gándara, Two Decades After the Affirmative Action Ban: Evaluating the University of California’s Race-Neutral Efforts[end-small-caps] 17 fig.6 (2015), http://www.ets.org/Media/Research/pdf/kidder_paper.pdf [https://perma.cc/RHL5-A8YN]; [small-caps]Univ. of Tex. at Austin, The University of Texas at Austin Accountability Report[end-small-caps] 3 (2016), http://www.txhighereddata.org/Interactive/Accountability/UNIV_Complete_PDF.cfm?FICE=003658 [https://perma.cc/QBV2-WZJM] (measure #6).

[58].     Brief of the Authors of the Texas Ten Percent Plan as Amicus Curiae in Support of Respondents at 8, Gratz v. Bollinger, 539 U.S. 244 (2003) (No. 02-516), http://diversity.umich.edu/admissions/legal/gra_amicus-ussc/um/10percent-gra.pdf [https://perma.cc/23HC-WNUR].

[59].     Fisher II, 758 F.3d 633, 651 (5th Cir. 2014), cert. granted, 135 S. Ct. 2888 (2015) (citing [small-caps]Gary Orfield et al., Civil Rights Project, E Pluribus . . . Separation: Deepening Double Segregation for More Students[end-small-caps] 46, 50 (2012)).

[60].     Id.

[61].     A great many of these papers are associated with the Texas Higher Education Opportunity Project, available at http://theop.princeton.edu/publications [https://perma.cc/58S7-Q2GR].

[62].     Angel L. Harris & Marta Tienda, Hispanics in Higher Education and the Texas Top Ten Percent Law, 4 [small-caps]Race & Soc. Probs.[end-small-caps] 57, 60–61, 65 (2012).

[63].     Id. at 62–63, 65.

[64].     See generally Angel Harris & Marta Tienda, Minority Higher Education Pipeline: Consequences of Changes in College Admissions Policy in Texas, 627 [small-caps]Annals Am. Acad. Pol. & Soc. Sci.[end-small-caps] 60 (2010).

[65].     Mark C. Long et al., Policy Transparency and College Enrollment: Did the Texas Top Ten Percent Law Broaden Access to the Public Flagships?, 627 [small-caps]Annals Am. Acad. Pol. & Soc. Sci.[end-small-caps] 82, 101 (2010).  Note that similar geographic diversity benefits did not obtain for Texas A&M at College Station, possibly because A&M’s statewide presence via, for example, agricultural extension programs, meant that A&M’s feeder high schools were already less concentrated than UT Austin’s.  Id.

[66].     Horn & Flores, supra note 45, at 1, 15–16.

[67].     [small-caps]Flores & Horn[end-small-caps], supra note 55, at 17 (citations omitted).

[68].     Marta Tienda, Lessons About College Admissions From the Texas Top 10 Percent Law, in [small-caps]Equal Opportunity in Higher Education: The Past and Future of California’s Proposition[end-small-caps] 209, 195 (Eric Grodsky & Michal Kurlaender eds., 2010); Harris & Tienda, supra note 62, at 66.

[69].     Cf. Eric Furstenberg, Academic Outcomes and the Texas Top Ten Percent Law, 627 [small-caps]Annals Am. Acad. Pol. & Soc. Sci.[end-small-caps] 167 (2010).  High school grades remain the best single predictor of college Grade Point Average (GPA), both at UT Austin and more generally.  Tienda, supra note 68.  So, to be clear, I am not arguing for greater reliance on SAT scores—only that even an effective predictor like high school GPA naturally has limitations when relied upon to the exclusion of all other pieces of information in an applicant’s file, including indicia such as academic motivation, “grit,” and leadership potential.  Cf. Steven B. Robbins et al., Do Psychosocial and Study Skill Factors Predict College Outcomes? A Meta-Analysis, 130 [small-caps]Psychol. Bull.[end-small-caps] 261 (2004) (documenting incremental predictive validity of noncognitive factors).

[70].     [small-caps]The Univ. of Tex. at Austin Office of Admissions, Implementation and Results of the Texas Automatic Admissions Law Report No.[end-small-caps] 12, 9 tbl.2B (2009), http://www.utexas.edu/student/admissions/research/admission_reports.html [https://perma.cc/6L7E-A6MG].

[71].     Gaps widened because white graduation rates increased significantly, while Black and Latino graduation rates at UT Austin stagnated.  Data presented are averages for four freshman cohorts using federal six-year graduation rates to smooth annual fluctuations.  Data available at NCAA, supra note 31; see also [small-caps]Univ. of Tex. at Austin, Freshman-Cohort Graduation Rates[end-small-caps] (2008), https://web1.ncaa.org/app_data/inst2015/703.pdf [https://perma.cc/S45G-3J36]; [small-caps]Univ. of Tex. at Austin, Freshman-Cohort Graduation Rates[end-small-caps] (2001), https://web1.ncaa.org/app_data/inst2008/703.pdf [https://perma.cc/MLR3-NG6Z].

[72].     [small-caps]Tex. A&M Univ., Coll. Station, Freshman-Cohort Graduation Rates[end-small-caps] (2008), https://web1.ncaa.org/app_data/inst2015/697.pdf [https://perma.cc/2XEV-XSUH].

[73].     See Transcript of Oral Argument at 67–68, Fisher v. University of Texas at Austin, 135 S. Ct. 2888 (2015) (No. 14-981) (“If you look at the academic performance of holistic minority admits versus the top 10 percent admits, over time, they–they fare better.”).

[74].     My discussion of racial gaps in graduation rates at UT Austin implicates one strand of the larger “diversity within diversity” issue at the heart of the Fisher case.  Devon Carbado’s Essay with Turetsky and Purdie-Vaughns in this symposium issue explores these themes in greater depth, including how affluent African Americans encounter significant race-related barriers in American society.  Devon W. Carbado et al., Privileged or Mismatched: The Lose-Lose Position of African Americans in the Affirmative Action Debate, 64 [small-caps]UCLA L. Rev. Disc.[end-small-caps] (forthcoming 2016); see also Devon W. Carbado, Intraracial Diversity, 60 [small-caps]UCLA L. Rev.[end-small-caps] 1130 (2013); Elise Boddie, Commentary on Fisher: The Importance of Diversity Within Diversity, [small-caps]SCOTUSblog[end-small-caps] (Oct. 11, 2012, 10:50 AM), http://www.scotusblog.com/2012/10/commentary-on-fisher-the-importance-of-diversity-within-diversity/ [https://perma.cc/PAS8-QLLF].

[75].     [small-caps]Alon[end-small-caps], supra note 21, at 240 fig.10.3 (documenting a gap of 77 percent versus 72 percent in graduation rates when including transfers).  The point of this statistic is to highlight the uphill difficulty of managing large SES-based admission regimes without threatening graduation rates.  There are limitations—such as selection bias among those with similar credentials and if, counterfactually, more low-SES students had enrolled at elite universities with sufficient financial support, they too may have enjoyed higher graduation rates—that caution against making even stronger inferences based upon these data alone.

[76].     See [small-caps]Kidder & Gándara[end-small-caps], supra note 57, passim.

[77].     Reprinted from [small-caps]Kidder & Gándara[end-small-caps], supra note 57, at 16 fig.5 (with permission) (using data from UC Office of the President and the California Departments of Finance and Education).

[78].     [small-caps]Equal Opportunity in Higher Education: The Past and Future of California’s Proposition [end-small-caps]209 (Eric Grodsky & Michal Kurlaender eds., 2010); Hinrichs, supra note 13, at 720 figs.1 & 2; Michal Kurlaender et al., Access and Diversity at the University of California in the Post-Affirmative Action Era, in [small-caps]Affirmative Action and Racial Equity: Considering the Fisher Case to Forge the Path Ahead[end-small-caps] (Uma M. Jayakumar & Liliana M. Garces Eds., 2015).

[79].     Plaintiff-Appellant’s Supplemental Brief at 38, Fisher v. Univ. of Tex. at Austin, 758 F.3d 633 (2014) (No. 09-50822), 2013 WL 5603455. Likewise, a recent law journal article by Ngov, optimistically concluding that socioeconomic status is a workable race-neutral alternative, overgeneralizes from the Gaertner & Hart study and places undue reliance on Kahlenberg’s works.  See Eang L. Ngov, Following Fisher: Narrowly Tailoring Affirmative Action, 64 [small-caps]Cath. U. L. Rev.[end-small-caps] 1 passim (2015).

[80].     Matthew N. Gaertner & Melissa Hart, Considering Class: College Access & Diversity, 7[small-caps] Harv. L. & Pol’y Rev.[end-small-caps] 367 (2013).

[81].     Id. at 377, 378, 393.

[82].     [small-caps]W. Interstate Comm’n for Higher Educ., Knocking at the College Door: Projections of High School Graduates—Colorado[end-small-caps] (2013), http://www.wiche.edu/info/knocking-8th/profiles/co.pdf [https://perma.cc/3CEN-KMH8].

[83].     [small-caps]Kidder & Gándara[end-small-caps], supra note 57, at 50 fig.A2.

[84].     Gaertner & Hart, supra note 80, at 396–98.

[85].     [small-caps]NAACP Legal Def. & Educ. Fund et al., Response to the Century Foundation’s Report on Affirmative Action in Higher Education[end-small-caps] (2012), http://www.naacpldf.org/files/case_issue/Century-Response.pdf [https://perma.cc/NLF7-ZCE8].

[86].     [small-caps]Kahlenberg & Potter[end-small-caps], supra note 23, at 12.

[87].     Id. at 31 fig.TX-TAMU-1.  Texas A&M chose not to restart race-conscious admissions after Grutter, unlike UT Austin.

[88].     Id. at 31.

[89].     Id. at 29.

[90].     See also Fisher II, 758 F.3d 633, 655 (2014).

[91].     [small-caps]Kahlenberg & Potter[end-small-caps], supra note 23 at 29–30.

[92].     [small-caps]Flores & Horn[end-small-caps], supra note 55, at 11 (“[I]gnoring the dramatic changes in the high school graduate population gives the appearance of substantially restoring access for students of color to levels before the percent plan, when, in reality, for a much larger population and share of students of color, it has actually declined.”).

[93].     Horn & Flores, supra note 45, at 57 (“[B]oth the University of Florida and Florida State University have continued to make use of race-conscious financial aid opportunity where permitted.  After the implementation of One Florida, the University of Florida began additional policies to supplement the race-conscious policies that remained in place.  Florida State University focused on aggressive affirmative recruiting and enrollment management strongly targeted by race.”).

[94].     [small-caps]Patricia Marin & Edgar K. Lee, Appearance and Reality in the Sunshine State: The Talented 20 Program in Florida[end-small-caps] 34 (2003), http://civilrightsproject.ucla.edu/research/college-access/admissions/appearance-and-reality-in-the-sunshine-state-the-talented-20-program-in-florida/marine-appearnace-reality-sunshine-2003.pdf [https://perma.cc/J7FJ-TR3Q].

[95].     Jeff Schweers, Black Enrollment at UF Takes a Hit, [small-caps]Gainesville Sun[end-small-caps] (Jan. 31, 2015, 5:20 PM), http://www.gainesville.com/article/20150131/ARTICLES/150139930?p=1&tc=pg [https://perma.cc/2CXT-N2GT].

[96].     Office of Inst. Planning and Research, Common Data Set, [small-caps]Univ. of Fla.[end-small-caps], http://ir.aa.ufl.edu/common-data-set [https://perma.cc/PVQ7-RXHA].

[97].     Id.  On the broader statewide impact of the “One Florida” plan, see also Neely Tucker, Jeb Bush Got His Way, and Then He Got a Mess, [small-caps]Wash. Post [end-small-caps](Jan. 7, 2016), http://www.washingtonpost.com/sf/national/2016/01/07/decidersbush/ [https://perma.cc/7BAB-WMM4] (“Sixteen years later, One Florida has had mixed results for the state’s minorities. African Americans were about 18 percent of the freshman class at all state universities and colleges before the program, roughly in line with their percentage of the overall population.  Today, African Americans account for 21 percent of the state’s population, but just 13 percent of the state’s freshman collegiate class.”).

[98].     Florida State’s success relates strongly to the Center for Academic Retention and Enhancement (CARE) program that in 2000 consolidated earlier retention programs related to the State’s desegregation plan.  [small-caps]Marin & Lee[end-small-caps], supra note 94, at 36.  Florida State’s CARE program combines a summer bridge program, an alternative admissions program plus financial scholarship support.  Partly by focusing recruiting in high-minority areas of South Florida, today out of the 400 disadvantaged students admitted to CARE five out of six are underrepresented minorities (51 percent African American and 32 percent Latino) and these students receive need-based scholarships covering their tuition and fees (and now other expenses too).  Byron Dobson, FSU Awarded $2.2 Million to Expand Programs, [small-caps]Tallahassee Democrat[end-small-caps] (Sept. 27, 2015, 11:35 PM), http://www.tallahassee.com/story/news/local/fsu-news/2015/09/27/fsu-awarded-grants-first-generation-students/72800506 [https://perma.cc/KTT9-6YVN].  In other words, the specifics of this Florida State University (FSU) program would appear to be vulnerable if all race-conscious efforts were prohibited in Florida.

[99].     [small-caps]Kahlenberg & Potter[end-small-caps], supra note 23, at 59.

[100].   Id. at 59, 69 n.283 (citing Common Data Set, [small-caps]Univ. of Ariz. Office of Inst. Research & Planning Support[end-small-caps], http://oirps.arizona.edu/CommonDataSet.asp [https://perma.cc/GRS3-XC4Z]).

[101].   Vishal Ganesan & Anna Swenson, Prop 107 Ends Affirmative Action in Arizona, [small-caps]College Fix[end-small-caps] (Nov. 8, 2010), http://www.thecollegefix.com/post/4572 [https://perma.cc/6U5R-KTQX] (“Shelton also denied that the factors outlawed by Proposition 107 have had any affect on admissions considerations.  ‘Indeed, we have never used race or gender as a criterion for admission to the U of A.’” (quoting Shelton)); Maria Leon, University of Arizona Chief Worried About End of Affirmative Action, [small-caps]Latin Am. Herald Trib.[end-small-caps] (Nov. 6, 2010),  http://www.laht.com/article.asp?ArticleId=376015&CategoryId=12395 [https://perma.cc/PD4T-HSLJ] (quoting president Shelton: “In UA we never used race or gender as a part of admission criteria.”).

[102].   [small-caps]Kahlenberg & Potter[end-small-caps], supra note 23, at 57–58

[103].   Fisher I, 133 S. Ct. 2411, 2420 (2013) (quoting Wygant v. Jackson Bd. of Educ., 476 U.S. 267, 280 n.6 (1986)).  Though Wygant is a public employment case, the quotation about “tolerable administrative expense” was originally anchored to the university admissions setting.  Id. at 280 n.6 (quoting Kent Greenawalt, Judicial Scrutiny of “Benign” Racial Preference in Law School Admissions, 75 [small-caps]Colum. L. Rev.[end-small-caps] 559, 578–579 (1975)).

[104].   See e.g., [small-caps]Reardon et al.[end-small-caps], supra note 16, at 21 (“The models presented in this paper do not address issues of cost or financial aid.”).

[105].   See [small-caps]Alon[end-small-caps], supra note 21, 199–201; [small-caps]Kidder & Gándara[end-small-caps], supra note 57, at 27–28.

[106].   [small-caps]Alon[end-small-caps], supra note 21, at 200.  Nor are private universities unaware that, on average, in future years low-income students are apt to make smaller philanthropic contributions to the university than alumni from affluent backgrounds.  Id. at 201; [small-caps]William G. Bowen et al., Equity and Excellence in American Higher Education[end-small-caps] 181–82 (2005).

[107].   Id. at 181.

[108].   This is a subset of the private doctoral-granting universities with a Carnegie Classification of “very high” research activity.

[109].   I use federal Integrated Postsecondary Education Data System (IPEDS) data for the estimates in this section.  IPEDS, [small-caps]Nat’l Ctr. Educ. Stat.[end-small-caps], http://nces.ed.gov/ipeds/datacenter/Data.aspx [https://perma.cc/PN9A-3LYR].  At these private universities, federal grant aid (which is income-based) is only 3.8 percent of total gift aid, and since Pell grant dollars would go up as well if more low-income students were admitted, I did not take out federal grant aid in these calculations.  The totals also include even smaller amounts of state grant aid (for example, a low-income student at the University of Southern California (USC) receiving a need-based Cal Grant), which varies from state to state.  These estimates are for all undergraduates, not simply the freshman class.

[110].   Higher Education Research and Development Survey–Data: Harvard University, [small-caps]Nat’l Sci. Found.[end-small-caps], http://ncsesdata.nsf.gov/profiles/site?method=report&fice=2155&id=h2 [https://perma.cc/R7QB-FPXS] (listing Harvard as having $20.3 million in research & development expenditures from the federal government in the Social Sciences in 2014).

[111].   [small-caps]Bowen et al.[end-small-caps], supra note 106, at 181.

[112].   [small-caps]Reardon et al.[end-small-caps], supra note 16, at 19.

[113].   Griggs v. Duke Power Co., 401 U.S. 424, 432 (1971).

[114].   [small-caps]Michael Mitchell & Michael Leachman, Ctr. on Budget & Policy Priorities, Years of Cuts Threaten to Put College Out of Reach for More Students[end-small-caps] 3 (2015), http://www.cbpp.org/sites/default/files/atoms/files/5-13-15sfp.pdf [https://perma.cc./3LAL-WTD7].

[115].   William Kidder, Race, Class, Access and Accountability: The Higher Ed Act 50, Presentation at the UC Irvine School of Law’s Symposium on Higher Education Access (Oct. 2015) (on file with author).  Inflation estimates come from the Bureau of Labor Statistics/CPI.  Contrast UC funding with California’s funding for prisons/corrections, which over the same period increased at 3.5 times the rate of inflation.  Id.

[116].   Id. at 10.

[117].   Ozan Jaquette et al., Tuition Rich, Mission Poor: Nonresident Enrollment Growth and the Socioeconomic and Racial Composition of Public Research Universities, [small-caps]J. Higher Educ.[end-small-caps] (forthcoming), http://web.missouri.edu/~cursb/research/tuition-rich-mission-poor-jaquette-curs-posselt-10-24-2014.pdf [https://perma.cc/2635-MYZF]; [small-caps]State Merit Scholarship Programs and Racial Inequality [end-small-caps](Donald E. Heller & Patricia Marin eds., 2004); Robert K. Toutkoushian & M. Najeeb Shafiq, A Conceptual Analysis of State Support for Higher Education: Appropriations Versus Need-Based Financial Aid, 51 [small-caps]Res. Higher Educ.[end-small-caps] 40, 41 (2010); Drew Allen & Gregory C. Wolniak, Exploring the Effects of Tuition Increases on Racial/Ethnic Diversity of Public Colleges and Universities (2015), http://www.aera.net/Newsroom/RecentAERAResearch/ExploringtheEffectsofRelativeTuitionIncreasesontheRacialEthnicCompositionofPublicColleges/tabid/15915/Default.aspx (follow “Read the full paper” hyperlink) [https://perma.cc/G2SM-QMTK] (paper presented at the AERA Annual Meeting).

[118].   Ozan Jaquette & Bradley R. Curs, Creating the Out-of-State University: Do Public Universities Increase Nonresident Freshman Enrollment in Response to Declining State Appropriations?, 56 [small-caps]Res. Higher Educ.[end-small-caps] 535 (2015); [small-caps]Kidder & Gándara[end-small-caps], supra note 57, at 27; Gary Orfield, Fisher v. University of Texas at Austin II: Are There Workable Alternatives to Affirmative Action? What the Research Shows (2015) (on file with author) (unpublished AERA research summary).

[119].   The negative impact on racial diversity of Michigan’s affirmative action ban is described in Schuette v. Coal. to Defend Affirmative Action, 133 S. Ct. 1623, 1678–79 (2013) (J. Sotomayor, dissenting).

[120].   College Navigator, University of Michigan—Ann Arbor, [small-caps]Nat’l Ctr. Educ. Stat.[end-small-caps], https://nces.ed.gov/collegenavigator/?q=University+of+Michigan&s=all&id=170976 [https://perma.cc/G5ZZ-JVWE] (containing financial aid and enrollment data).

[121].   Id.

[122].   [small-caps]Kidder & Gándara[end-small-caps], supra note 57, at 27.

[123].   [small-caps]Lorelle L. Espinosa et al., Race, Class, and College Access: Achieving Diversity in a Shifting Legal Landscape[end-small-caps] 10 (2015).

[124].   Id. at 30.

[125].   Julie J. Park et al., Does Socioeconomic Diversity Make a Difference? Examining the Effects of Racial and Socioeconomic Diversity on the Campus Climate for Diversity, 50 [small-caps]Am. Educ. Res. J.[end-small-caps] 466, 486–87 (2012) (“Importantly, our institution-level findings highlight how institutions with higher levels of both racial and socioeconomic diversity (via CCI) [cross-class interactions] contribute uniquely to student engagement in diversity-related activities and CRI.”).