Gratz v. Bollinger and Grutter v. Bollinger
Supporting Gratz, the National Association of Scholars as amicus curiae (pp 5-19, 21-22): The University's research (the Gurin Report) offers no proof that increases in student racial diversity yield educational benefits. Other scholarship using Gurinís primary database shows no correlation between student racial diversity and educational benefits. (The complete brief is online here.)
Comments by Curtis Crawford,* indented in italics.
THE UNIVERSITY BEARS A HEAVY BURDEN TO DEMONSTRATE
The Fourteenth Amendment to the Constitution provides that "No State shall . . . deny to any person within its jurisdiction the equal protection of the laws." U.S. CONST. amend. XIV ß 1. Because the "rights created by the first section of the Fourteenth Amendment are, by its terms, guaranteed to the individual," Shelley v. Kraemer, 334 U.S. 1, 22 (1948), "a [stateís] racial classification causes Ďfundamental injuryí to the Ďindividual rights of a person,í" Shaw v. Hunt, 517 U.S. 899, 908 (1996) (internal citation omitted).
Accordingly, this Courtís precedents set a high hurdle for any state actor that wishes to apply a race-based classification: the state actor must demonstrate that its racial classification advances a "compelling governmental interest[ ]" and that its use of race is "narrowly tailored" to meet that interest. Adarand Constructors, Inc. v. Pena, 515 U.S. 200, 227 (1995); see also City of Richmond v. J.A. Croson Co., 488 U.S. 469, 493-94 (1989) (plurality opinion) (same).
Mere recital of a compelling governmental interest is not enough; the state must provide a "strong basis in [-6-] evidence for its conclusion" that its use of race advances a compelling governmental interest. Wygant v. Jackson Bd. of Educ., 476 U.S. 267, 277 (1986) (plurality opinion); Croson, 488 U.S. at 500. The Gurin Report provides no such evidence: it does not show that student racial diversity (much less that achieved by racial preferences) creates any educational benefits. Because the Gurin Report is irrelevant to this litigation, it cannot help the University meet its burden.
II. THE GURIN REPORT DOES NOT SHOW
The district court relied on the Gurin Report in finding that student racial diversity results in educational benefits: "Gurin reports that . . . Ď[s]tudents who experienced the most racial and ethnic diversity in classroom settings and in informal interactions with peers showed the greatest engagement in active thinking processes, growth in intellectual engagement and motivation, and growth in intellectual and academic skills.í" Pet. App. 23a (quoting Gurin Rep., at 5).
Gurinís report does not support her claims. The Report purports to measure the relationship between student-body racial diversity and educational quality at the University of Michigan. See Gurin Rep., at 3-4. It does nothing of the sort.
Gurin did not compare her alleged educational benefits with the number or proportion of Asian, African-American, Hispanic-American, Native American, and white students at the University of Michigan or any other institution. Instead, she relied on studentsí answers to questions about whether they had enrolled in ethnic studies courses, had discussed racial issues, or had close friends of a different race. Her chosen measurement (or variable, to use [-7-] the statisticianís phrase) was not the extent of racial diversity in the studentís university community but how much that student talked (or was talked to) about race and racial issues, in class and out.
As for the outcome Ė the educational benefits said to result from this talking about race Ė it is hard to find any evidence in Professor Gurinís study of the "growth in intellectual and academic skills" that she claims. She did not report effects on grades, standardized test scores, or graduate-school admission rates. Instead, she looked to studentsí answers to questions such as whether they believe "causes of behavior often form [a] chain that goes back in time" or how important they think it is to "[w]rite original works" or whether they are satisfied with themselves. Because Gurin did not even look at student-body racial diversity and did not measure improvements in educational outcomes, her study says nothing about whether the latter follows from the former. Her study is therefore irrelevant to this case.
If Gurin wished to study the effects of racial diversity in college student bodies, her first step should have been to gather information on the number and proportion of Asian, African-American, Hispanic-American, Native American, and white students in a number of different universities or at different times. She had before her, in one of the three databases she used for her study, information that would have enabled her to compare the extent of such diversity at a large number of colleges and universities. 4 She chose not to use [-8-] this information to measure student racial diversity. Instead, she seized upon student answers to questions about ethnic studies, racial dialogue workshops, and interracial experiences, and named those answers "diversity experiences," even though they are unrelated to actual racial diversity in the student body. The Report asked whether a student:
1. enrolled in the Intergroup Relations, Community, and Conflict Program ("IGRCCP") ethnic studies course; 5
2. "[e]nrolled in an ethnic studies course" and "discussed racial issues";
3. "[e]nrolled in an ethnic studies course" and "attend[ed] a racial/cultural awareness workshop";
4. "[h]ad a course that had [an] important impact on [his or her] views of racial/ethnic diversity and [-9-] multiculturalism," and participated in racial "dialogue groups" at the University of Michigan;
5. "[h]ad a course that had [an] important impact on [his or her] views of racial/ethnic diversity and multiculturalism," and was involved "with groups and activities [at the University of the Michigan] reflecting other cultural/ethnic backgrounds";
6. "[h]ad a course that had [an] important impact on [his or her] views of racial/ethnic diversity and multiculturalism," and attended "multiethnic [themed] campus events";
8. "[e]nrolled in an ethnic studies course" and had "close friends in college who were of [a different] race";
9. "[h]ad a course that had [an] important impact on [his or her] views of racial/ethnic diversity and multiculturalism," "[h]ad meaningful, and honest discussions about race and ethnic relations" with other-race students, and "[s]hared personal feelings and problems"; 10. "[h]ad a course that had [an] important impact on [his or her] views of racial/ethnic diversity and multiculturalism," and, of his or her "six closest friends at Michigan," how many were of a different race.6
[-10-] The fact that Gurin labeled these measurements as "classroom diversity" and "informal interactional diversity" experiences, does not make them relevant to this litigation. They tell us nothing about actual student racial diversity at the University of Michigan or elsewhere. *The first six - enrolling in an ethnic-studies course, discussing race, participating in workshops or dialogue groups, and attending cultural activities or multi-ethnic events -do not require the presence on campus of any students of another race.*
As for the last four questions, affirmative answers presuppose the presence of some students of another race on campus. But *the Gurin Report does not indicate how many such students must be present to achieve the effects reported or what races must be represented or in what proportions.* Therefore, affirmative answers to those questions say nothing about the particular form of racial diversity that is fostered by the University. **Nor do affirmative answers indicate that the other-race students whose presence is implied will not be there absent racial preferences.** As Gurin must concede, some of her claimed benefits are attributable to the presence of students who receive no preferences from the University (e.g., Asians, whites, Arabs).7 Even though these claimed [-11-] benefits are independent of racial preferences, the University erroneously credits them to its admissions programs. See G.E. Zuriff, Is Racial and Ethnic Diversity Educationally Beneficial?, WORLD & I, Aug. 2002, at 271, 273 (discussing this error).
Perhaps recognizing that her "diversity" measurements are irrelevant to the issue in this litigation, Gurin attempted to save them by claiming that, because student racial diversity correlates with her ten measurements and her ten measurements correlate with educational benefits, student racial diversity must correlate with educational benefits. See Gurin Rep., at 30-33. 8 Gurin could not possibly mean to argue that everything that correlates with her "diversity experiences" improves education at the University. For example, two of Gurinís "diversity experiences" include taking an ethnic studies course coupled with participating in a racial-dialogue group or workshop. See id. App. C, at 13, 18-19. Many schools, including the University of Michigan, offer such dialogue groups and workshops as a "remedy" for a hostile racial climate.9 According to Gurinís reasoning, because a hostile racial climate correlates with her measurements, it must improve [-12-] education at the University. Not only does Gurinís reasoning violate the rules of logic; it violates the rules of statistics as well. See NAS Study, at 82 (providing a mathematical and statistical refutation of Gurinís claims).10
If Gurin had wished to measure whether student racial diversity produces educational benefits, she should have actually compared the two, not used other, irrelevant measurements. This would not have been difficult; Gurinís primary database includes this information, yet she chose not to use it. The reason why is not hard to find: Using the same database as Gurin, Alexander Astin, Director of the Higher Education Research Institute at UCLA, compared student-body racial diversity and educational benefits and found no relationship between them. See infra Part IV.
The district court recited some of NASís basic criticisms of the Gurin Report without responding to them. See Pet. App. 27a. It essayed an answer to just one of the points made by NAS (restated here at p. 10 & n.7 above), that the Gurin Report gives us no idea how many minority students are needed to achieve any beneficial educational results or whether racial preferences are needed to provide the requisite diversity. This argument, the court said, goes [-13-] only to "whether the means employed to achieve [the compelling governmental interest of educational benefits] are narrowly tailored." Id. Not so. Unless we know how diverse, and in what way diverse, a student body must be to yield the desired educational benefits, we cannot know that the racial preferences chosen to achieve those benefits will survive the strict scrutiny that compelling-interest analysis demands.
Gurinís "Learning" and "Democracy Outcomes"
The Gurin Report purports to "confirm[ ] that racial diversity and student involvement in activities related to diversity have a direct and strong effect on learning." Gurin Rep., at 9. But Gurin could not possibly have reached any conclusions on academic outcomes, for the simple reason that she never measured them.
Gurin rejected accepted, traditional measurements of academic outcomes in favor of her own, subjective, ones. Even though her database includes data on such traditional measurements of academic achievement as self-reported grades, graduation rates, admissions to graduate school, and performance on seven standardized tests, see ALEXANDER W. ASTIN, WHAT MATTERS IN COLLEGE? 188-99, 218-20 (1993), Gurin chose to include only one of those measurements Ė self-reported grades (Gurin Rep. App. C, at 14) Ė which she reported has no "consistent relationship[ ]" with student outcomes. Gurin Rep., at 38. 11 Thus, Gurin found no [-14-] consistent link between her purported measurements for racial diversity and the only even arguably objective 12 measurement of academic achievement in her study. *Gurin instead created eleven of her own, subjective measurements of studentsí political and social attitudes and self-esteem.* Although these measurements are unrelated to academic outcomes, Gurin labeled them "learning outcomes."
These measurements include:
1. whether students "[e]njoy analyzing reasons for behavior," "[p]refer simple rather than complex explanations," and "[t]ake peopleís behavior at face value";
2. whether they "[t]hink about the influence of society on other people" and on themselves, and believe "[c]auses of behavior often form [a] chain that goes back in time";
3. how much they valued "[g]eneral knowledge," and "problem-solving," thinking, writing, and foreign language skills;
4. how personally important it was to them to "[w]rite original works (poems, novels, short stories, etc.)" and "[c]reate artistic works (painting, sculptures, decorating, etc.)"; [-15-]
5. whether they were intellectually satisfied with their education at the University;
6. whether they expected to attend graduate school;
7. whether they wanted to attend graduate school;
8. whether they thought they had a greater "[d]rive to achieve" and intellectual "[s]elf-confidence" than the average person their age;
10. whether they thought they were more prepared for graduate school than when they entered college;
11. *whether they had more "[g]eneral knowledge" and better "problem-solving," thinking, writing, and foreign language skills than when they entered college.* 13
Gurin mislabeled her measurements as "[s]ocial historical thinking," "[c]omplex thinking," and "[i]ntellectual engagement variables," even though, overall, they tell us nothing about studentsí thinking abilities or intellectual skills.
In measurements (1), (2), (3), and (4), students were asked a little about what they think about, but mostly about what kind of thinking or other intellectual activity they "enjoy" or "value." They certainly did not ask whether students are capable of or engage in Gurinís "complex [-16-] thinking" or even her "social historical thinking." Measurement (5) asked students if they enjoyed their time at the University and found classes satisfying, not whether they did well. Measurements (6) and (7) asked students whether they wish to attend graduate school, not whether and where they were actually admitted. That is a separate measurement in Gurinís database Ė a measurement that Gurin excluded. See ASTIN, supra, at 197-98. And measurements (8), (9), (10), and (11) merely assessed studentsí self-esteem. The fact that students believe they are better writers and "problem solvers" than they were four years earlier does not make it so. Such self-assessments are inherently suspect. See Justin Kruger & David Dunning, Unskilled and Unaware of It: How Difficulties in Recognizing Oneís Own Incompetence Lead to Inflated Self-Assessments, 77 J. PERSONALITY & SOC. PSYCHOL. 1121, 1123-24 (1999) (finding that those who were most confident of their abilities were often the least able).
*Common sense indicates that these are not valid measurements of educational benefits.* Nowhere in society Ė not in graduate school admissions, college rankings, job recruitment Ė do we measure a studentís academic success by asking him how much he personally values creating artistic works or whether he enjoys guessing the reasons for peopleís behavior. Very few parents would be likely to accept a transcript that reported not grades, but their childís self-ratings of his abilities and drive to achieve. This Court likewise should not do so.
In addition to her "learning outcomes," Gurin provided over a dozen "democracy outcome" measurements Ė measurements of studentsí political and social attitudes and citizenship. These are, on their face, irrelevant; Gurin did not even claim that they measure academic outcomes. Instead they measure how effective the Universityís courses are in changing studentsí political and social views.
[-17-] These measurements range from the extraordinarily bland and uncontroversial (e.g., whether the respondent considers othersí perspectives,14 volunteers in his community,15 or values "racial understanding"16), to the political (e.g., whether the respondent "think[s] about the amount of power people in different segments of society have"17). Gurin went so far as to define a positive outcome as a belief that the Universityís diversity programs are beneficial, and a negative outcome as a response that these programs "foster[ ] more intergroup division than understanding" and lead to tense, uncomfortable discussions on "ethnic, racial, and gender issues."18 These measurements tell us more about Gurinís political views than they do about education at the University.
Gurin attempted to save her "democracy outcomes" by linking them to education at the University. Education, she explained, "equip[s] students for meaningful participation" in a democracy by making them "better able to understand and consider multiple perspectives, deal with the conflicts that different perspectives sometimes create, and appreciate the common values and integrative forces that harness differences in pursuit of the common good. Students can best develop a capacity to understand the ideas and feelings of others in an environment characterized by diverse others . . . ." Gurin Rep., at 5-6; accord id. at 29.
But racial diversity is not necessary to any of these outcomes. Students can listen to "multiple perspectives" and [-18-] learn to be considerate of others with or without a racially diverse student body. The diversity Gurin describes as good for democracy Ė diversity of viewpoint Ė is very different from the diversity of skin color that the University seeks to achieve.
The Gurin Report does not even show a significant relationship between studentsí "diversity experiences" and the educational benefits she purported to measure. This is all the more remarkable because, as discussed later at Part III, Gurin violated basic statistical rules to strengthen such a relationship. In the end, Gurinís own numbers indicate that changes in her "diversity" measurements have either no effect on students 19 or an effect so tiny as to be unnoticeable.
For the vast majority of her measurements, Gurin reported that even extremely large increases in her "diversity" measurements account for less than 1 percent of studentsí "learning" outcomes. Put differently, Gurin found that more than 99 percent of a studentís "learning" is determined by factors other than her "diversity" measurements.20 See Gurin Rep. App. D, tbls.; see also NAS Study, at 80. Only in a handful of cases could Gurin report effects as high as 3 or 4 percent, still exceedingly weak [-19-] findings. Gurinís "democracy" measurements fared only slightly better. For the vast majority of her measurements, Gurin could account for less than 4 percent of studentsí "democracy" outcomes. See Gurin Rep. App. D, tbls.; see also NAS Study, at 80-81.
The Gurin Report is therefore irrelevant to this litigation for two independent reasons: first, because it does not compare student racial diversity to accepted measurements of academic outcomes and, second, because it finds that "diversity experiences" have no effect or only a minuscule effect on students.
* * * *
If Gurin had wished to measure the effects of student-body racial diversity on academic outcomes, she should have done so. As we have said, this would not have been difficult; Gurinís dataset includes accepted, traditional measurements of both. *Although Gurin provides no explanation for why she chose not to use these measurements,* a look at Alexander Astinís book What Matters in College? may explain.
Using the principal database that Gurin uses Ė the one that records racial diversity and educational outcomes at 184 colleges and universities Ė Astin compared percentages of Asian-Americans, African-Americans, and Hispanic-Americans at those colleges and universities with self-reported college grades, college graduation rates, admissions to graduate schools, and performance on seven standardized tests. See ALEXANDER W. ASTIN, WHAT MATTERS IN COLLEGE? 62, 188-90 (grades), 191-93 (graduation rates), 197-98 (admissions to graduate school), 199-220 (standardized tests). *Astin Ė who is a supporter of racial preferences Ė reported that academic "outcomes are generally not affected" by student-body racial diversity, and that any effects "are very weak and indirect." Id. at 362. From this Astin concluded: "The values, attitudes, self-concept, and socioeconomic status of the peer group are much more important determinants of how the individual student will develop than are the peer groupís abilities, religious [-22-] orientation, or racial composition."* Id. at 363 (emphasis added).
Gurin must have been aware of Astinís conclusion, for she cited Astinís book four times in her Report. Gurin Rep. App. B, at 24 (listing studies she reviewed for this litigation); Gurin Rep. App. B, at 14, 18 (listing her review of the literature). *Yet Gurin never addressed Astinís contradictory findings in her study.*
Since his 1993 study, Astin has reaffirmed these findings. In an interview that appeared in the Chronicle of Higher Education in 2001, Astin stated again that the correlation between student-body racial diversity and beneficial academic outcomes "is yet to be convincingly demonstrated," and that "[t]he research still needs to be done that would demonstrate that link." See Peter Schmidt, Debating the Benefits of Affirmative Action, THE CHRONICLE OF HIGHER EDUCATION, May 18, 2001, at A25 (interview with Alexander Astin).
Despite Astinís repeated statements that neither he nor any other researcher has demonstrated a link between student-body racial diversity and academic outcomes, the district court cited Astinís work as "solid evidence" of this link. Pet. App. 24a-25a. But Astinís own statistical findings, published conclusions, and most recent statement on the matter belie reliance on any link between student racial diversity and educational benefits.
4. Gurin used three datasets. The Cooperative Institute Research Program ("CIRP") database records the number of "Asian," "black," "Mexican-American/ Chicano," "Puerto Rican American," "American Indian," and "white" students at 184 colleges, see Robert Lerner & Althea K. Nagai, A Critique of the Expert Report of Patricia Gurin in Gratz v. Bollinger, at 24, available at ceousa.org, and the percentage of the student body that is non-white, see NAS Study, at 88. Gurin performed four separate analyses of the CIRP data, and included approximately sixty CIRP measurements, but she never used the CIRP measurements of student-body racial diversity. See Gurin Rep. App. C, at 6. The other two data sets Ė the Michigan Student Survey ("MSS") and Intergroup Relations, Community, and Conflict Program ("IGRCCP") Ė include no such data. They cover one class of students at the University of Michigan. Because racial composition is a constant in these databases, Gurin could not use them to measure changes in racial diversity in the student body. See Gurin Rep. App. C, at 7, 9. [return to text]
5. Gurin describes this as a course on "the history of group experiences in the United States, a contemporary analysis of group inequalities in the economic, educational, and political arenas, and an analysis of political issues and policies (such as immigration, bilingual education, affirmative action, sexual harassment, Middle East peace initiatives) . . . ." Gurin Rep. App. C, at 9-10. [return to text]
6. Measurement (1) is drawn form the IGRCCP database. See Gurin Rep. App. C, at 21. Measurements (2), (3), (7), and (8) are drawn from the CIRP databse. See Gurin Rep. App. C, at 6-7, 13. Measurements (4), (5), (6), (9), and (10) are drawn from the MSS database. See Gurin Rep. App. C, at 9, 18-19. [return to text]
7. Although Gurin did not reveal what percentage of her outcomes results from the presence of Asian and white students Ė students who receive no racial preference Ė her Report estimates that the percentage is quite large. The Universityís white students, who "come from the most segregated backgrounds and hence have the most to learn from the racial/ethnic diversity they find at Michigan," Gurin Rep. App. E, at 2, report that most of their interaction with other-race students is with Asians, Gurin Rep. App. E, at 3. For "students of color," Gurin attributed their high number of "interracial relationships [to] the predominance of white students on the Michigan campus." Gurin Rep. App. E, at 3-4. Since a large percentage of Gurinís results are thus completely independent of the Universityís racial preferences, the district court erred in relying on these results in assessing the effects of the Universityís racial preferences. [return to text]
8. As noted above, one of Gurinís databases includes information on racial diversity of the student body. (The Gurin Report calls this "structural diversity." See Gurin Rep., at 30.) But, as also noted above, Gurin never used this information on student racial diversity in her study, see id. App. C, at 6; she only noted that it sometimes correlates with her ten "diversity" measurements, see id. at 31-32. [return to text]
9. See, e.g., DINESH DíSOUZA, ILLIBERAL EDUCATION: THE POLITICS OF RACE AND SEX ON CAMPUS 138-41 (1991) (reporting that the University of Michigan instituted racial sensitivity workshops in response to racist incidents at the undergraduate radio station).[return to text]
10. The NAS Study explains that "the argument is unsound, because it can be shown mathematically that if variables A and B are positively correlated, and variables B and C are positively correlated, it is possible that A and C are negatively correlated. To be sure, one can deduce that A and C are positively correlated if one knows that the correlations between A and B and between B and C are very high (near 1). But Gurinís correlations are less (usually much less) than .25, which isnít nearly large enough . . . logically or mathematically." NAS Study, at 82. Wood and Sherman have recently issued a study analyzing the data from the same database Gurin used. Their analysis demonstrates that the indirect correlations Gurin reported are indeed too small for Gurin to claim her "diversity" measurements correlate with her "education" outcomes. See Thomas E. Wood & Malcolm J. Sherman, Supplement to Race and Higher Education, available at nas.org.rhe2/html. [return to text]
11. Gurin looked for 24 possible relationships between her "diversity" measurements and grades. For only 6 of these did Gurin report results; for the other 18, Gurin found no relationship. Moreover, these 6 factors indicate that her "diversity" measurements have almost no impact on grades; even very large increases in these measurements lead to extremely small changes in grades. See Gurin Rep. App. D tbl. D1, at 2, row 1. Finally, Gurinís results are inconsistent with one another, or with any theory of the beneficial effects of student racial diversity adopted by the district court: Gurin found that increases in "diversity courses" correlate with African-Americans reporting lower grades and with Hispanics and whites reporting higher grades. Gurin Rep., at 38. [return to text]
12. The fact that the grades she used are self-reported adds an element of subjectivity to that measurement as well. [return to text]
13. Measurements (3), (4), (7), (8), (9), (10), and (11) are drawn from the CIRP database. See Gurin Rep. App. C, at 14-16. Measurements (1), (2), (5), and (6) are drawn from the MSS and IGRCCP databases. See Gurin Rep. App. C, at 19, 21. [return to text]
14. Gurin Rep. App. C, at 20 (MSS), 21 (IGRCCP). [return to text]
15. Gurin Rep. App. C, at 16 (CIRP). [return to text]
16. Gurin Rep. App. C, at 17 (CIRP).[return to text]
17. Gurin Rep. App. C, at 21-22 (IGRCCP). [return to text]
18. Gurin Rep. App. C, at 20 (MSS). [return to text]
19. Phrased in econometric terms, Gurin found no "statistically significant" results. This means that whatever numbers she found were completely unreliable Ė they were likely due to random chance, and if she repeated her experiment she would just as likely find no result. [return to text]
20. In econometric terms, Gurin found an r2 value of 1% or less. This means that Gurinís model can explain only 1% (or less) of a studentís "learning." [return to text]