Diversity, Race as Proxy, and Religion as Proxy



Cite as Eugene Volokh, Diversity, Race as Proxy, and Religion as Proxy, 43 UCLA L. Rev. 2059 (1996).



I.        The Appeal of Diversity
II.        Heightened Scrutiny
III.       The Diversity Puzzle
IV.        Race and Religion
V.        The Trouble with Diversity

The Appeal of Diversity

       The diversity justification for race-consciousness has an undeniable appeal.  Diversity of "experiences, outlooks, and ideas" 1 is a worthy goal, in the academic context and in many others, and it's true that one's experiences, outlooks, and ideas correlate in some measure with one's race.  The degree of this correlation is uncertain, but its existence is not.  A reasonable, unprejudiced decisionmaker could certainly conclude that race can sometimes be a useful proxy for this intellectual diversity.

       Diversity is particularly appealing because of what it is not.  It is not based on theories of racial responsibility.  It is not based on a vision of group rights, or on a theory that proportional racial representation is an end in itself. 2  It is not based on controversial views of compensation for past discrimination.  It does not require a different level of justification for programs that disadvantage whites than for programs that disadvantage minorities.  It does not even require a social consensus about the magnitude of present discrimination.

       All these other suggested justifications for race-consciousness, I hasten to say, have their supporters, and these supporters have made reasonable arguments in their favor.  I generally disagree with these arguments, but I will not try to rebut them here.  I only suggest that these other justifications may in some respects be a harder sell than diversity.

       Diversity is appealing because it's forward-looking; it ascribes no guilt, calls for no arguments about compensation.  It seems to ask simply for rational, unbigoted judgment: thoughtful professionals evaluating the whole person, including the regrettably but undeniably important factor of race, in a quest to achieve an eminently legitimate, even uncontroversial, goal.

       The trouble, though, is that a key part of antidiscrimination law requires that people avoid looking at the whole person.  Antidiscrimination law tells us that some attributes of a person -- for instance, the person's race, sex, and religion -- must be ignored, even if they do correlate with abilities or attitudes.  The Equal Protection Clause, the Civil Rights Act, and similar laws generally require decisionmakers to close their eyes to those attributes.

       In this Article, I focus on this tension between the "whole person" model of the diversity argument and the "close your eyes" model of antidiscrimination law, and ask two basic questions:

       1.       If it's proper in the diversity admissions (or hiring) context to use race as a proxy for a person's experiences, outlooks, and ideas, why is it improper to use race as a proxy in other contexts?

       2.       If it's proper to consider an applicant's race for diversity purposes, what would stop universities from also considering the applicant's religion, an attribute that is at least as good a predictor of a person's attitudes and experiences?

       I believe these questions have no satisfactory answers.  The diversity rationale, standing alone, cannot be reconciled with the normal bars on the use of race and religion as proxy.  If we value these bars, as I think we should, the diversity rationale ought not be accepted.  Supporters of race-conscious programs could still argue for such programs on other grounds, but the diversity rationale, despite its popularity, cannot carry the day.

Heightened Scrutiny

       If the Equal Protection Clause simply called for reasonable, unbigoted decisionmaking in pursuit of a legitimate goal, diversity-based race-consciousness would pass with flying colors.  But the Equal Protection Clause demands more.  A huge chunk of equal protection law (and antidiscrimination law more generally) is aimed precisely at barring the use of reasonable, unbigoted judgments that race is a valid proxy for experiences, outlooks, or ideas.

       Imagine an employer who prefers to hire whites rather than blacks, because of his statistically supported judgment that whites are more likely than blacks to have a certain set of experiences or attitudes that are useful for the job.  Imagine also a prosecutor (or a defense attorney) who prefers to have blacks rather than whites on a jury because of his statistically supported judgment that blacks are more likely than whites to have a certain set of experiences or attitudes that will lead them to favor his client. 3  Or imagine the same in the sex classification context. 4

       If I were prosecuting a black defendant accused of resisting an arrest by white police officers, or defending a white police officer accused of beating a black suspect, I wouldn't have to be irrational or bigoted to conclude that black jurors would, on average, sympathize less with my side than white jurors.  In fact, the more I knew about the shameful history of racist violence by police officers, the stronger my preference would be.

       Likewise, say that, as some have plausibly asserted, black students are more likely to have experiences that provide a useful perspective on police discretion issues in Fourth Amendment law.  One could equally suggest that these experiences might, as a general matter, make black applicants more suited for jobs as, say, investigators of charges of police abuse.  Conversely, one could argue that the experiences might -- as a general rule -- make black applicants less suited for jobs as zealous defenders of the police against charges of such abuse.

       The traditional reaction to these examples is that, even when race is correlated with a relevant job characteristic, for example, zeal in defending police officers, one should just look at that characteristic and not use race as a proxy.  But the diversity rationale would suggest that it is permissible to use race as a proxy for experiences, outlooks, and ideas.  What makes the hiring examples different?

       Similarly, just as there are practical benefits to diversity, there may be practical benefits to uniformity.  In some situations, teams may work better if their members share the same experiences, outlooks, and ideas.  Uniformity has its costs, but reasonable, unprejudiced people can conclude that it can sometimes also have its benefits. 5  But I take it that the Equal Protection Clause bars a government employer from preferring his fellow whites -- or, for that matter, his fellow blacks or Jews or Cubans -- because of eminently plausible theories that shared race or ethnicity is correlated with the teamwork needed to get the job done.

       One of the great tasks of antidiscrimination law over the past thirty years has been to persuade people that they ought not use race and sex as proxies, even when race and sex are statistically plausible proxies.  It's hard enough to persuade people to give up their irrational prejudices; it's harder still to persuade them to give up their rational ones.

       The diversity rationale, though, openly embraces the use of such proxies.  Perhaps this embrace is wise, or at least a necessary evil, and the actions of my hypothetical lawyer or employer are not.  But the embrace may well lead people -- both the government decisionmakers who are directly bound by equal protection jurisprudence and the private people to whom the jurisprudence is meant to send a normative signal -- to ask "What makes my use of race as a proxy, in choosing employees or jurors or business partners or fellow club members, horribly wrong but your use wonderfully right?" 6

The Diversity Puzzle

       This, then, is the diversity puzzle:  How does considering race because of its correlation to experiences, outlooks, and ideas differ when the goal is educational diversity from when the goal is winning a trial or doing the job more effectively?  If race-consciousness is good in the diversity context, how can we still argue that it's bad in the other contexts?

       Quotas vs. Factors:  The distinction cannot be that quotas are bad but considering race as one of many factors is good. 7  Imagine a government employer who says:  "I don't automatically reject blacks (or Jews) -- nothing of the sort! But I have studies that show that blacks (Jews) tend to have a different set of experiences, outlooks, and ideas than other groups.  So I consider all I know about the applicant, his objective qualifications as well as what I've learned about his personal experiences, outlooks, and ideas, and use race (ethnicity) as just one of many factors." 8  Alternatively, imagine a lawyer who says the same about his peremptory challenges in jury selection.  Surely this sort of decisionmaking is unconstitutional, even though it doesn't use quotas or absolute exclusions. 9

       If one's goal is selecting a diverse student body, a good employee, or a sympathetic juror, using a quota might not be narrowly tailored to that goal.  Considering race as one of many factors might be the better tailored alternative, 10 but is it still an acceptable alternative?  The example above suggests that it is not acceptable, even for traditionally holistic hiring decisions and peremptory challenge.  Why would the answer be any different for a university admissions decision?

       Unqualified vs. Qualified:  Nor can the distinction be that race-consciousness is impermissible only when it favors applicants who are unqualified. 11  Again, imagine the government employer who says:  "I would never hire an unqualified white (or a non-Jew) over a qualified black (Jew).  Absolutely not! I only give qualified whites (non-Jews) a slight preference."  Presumably the argument would be rejected out of hand.

       In fact, it's not clear that "unqualified" even has much meaning in most areas, including education.  "Qualified" is a matter of degree:  Some people seem more capable than others; some test scores suggest that one will have a 75% chance of graduating (a rough measure, of course), others suggest a 50% chance, others a 25% chance.  Calling a student "unqualified" means only that he falls below some arbitrary cutoff dictated by many contingencies, including the applicant pool quality, the number of slots, the funds available for remedial classes, and so on.  Distinctions between preferences that favor the unqualified over the qualified and those that favor the less qualified over the more qualified seem illusory. 12

       Benefiting Everyone:  It also seems hard to distinguish educational diversity programs on the theory that they "help everyone," as opposed to just the immediate beneficiary. 13  Efficient government operation helps everyone too.  If using race as a proxy for other attributes gives us even a slight improvement in the quality of police brutality investigators, or for that matter of policemen, society as a whole is being benefited. 14

       Need for Proxy:  Perhaps using race as a proxy is permissible in some cases but not others because it's more necessary in some cases than in others.  When it is hard to measure some criterion directly, otherwise impermissible proxies might become legitimate.

       But how does this distinguish university admissions from, for instance, various employment decisions?  Predicting future employee performance is notoriously hard.  Interviews are of little help:  The applicant is on his best behavior, and may hide some of his attitudes.  Whether someone comes from a particular ethnic group may not be a very reliable measure of, say, his attitude about women.  But his answer to the question "What's your attitude about women?," or even to a subtler battery of questions, may be no more reliable.  The employer has as hard a choice in front of it -- and has as much need for the use of proxies to supplement its meager information about the applicant -- as does the admissions office.

       Moreover, even if a job interview does provide a decent sense of an applicant's likely experiences, outlooks, and ideas, the need to use race as a proxy only retreats, not evaporates.  If two candidates' resumes and interviews show that they seem equally qualified for a job, race may be the most reliable remaining tiebreaker.  Say one is looking for aggressive litigators, and one discovers that 50% of the blacks who seem like they would make aggressive litigators actually do make aggressive litigators, but only 20% of the East Asians who seem like they would make aggressive litigators actually do.  If one must choose between an aggressive-seeming black and an equally aggressive-seeming East Asian, then the need to use race as a proxy resurfaces:  By definition, one has no even mildly reliable way of deciding between the two other than using race.

       Accuracy of Proxy:  Another possible distinction is the accuracy of the proxy -- perhaps it is permissible to consider race when it's an accurate proxy for certain experiences, outlooks, or ideas but impermissible when it's an inaccurate proxy for others.

       But this doesn't practically distinguish diversity admissions decisions from proxy-based hiring decisions, either.  Under current law, employment decisions may not be based on race as a proxy no matter how well race correlates with the relevant criterion; the same is true for the decisions involved in peremptory challenges.  At the same time, it is far from clear that race is a remarkably strong proxy for those attributes that contribute to intellectual diversity.

       Specific v. General:  Some have suggested to me that the diversity rationale is permissible because it rests on general assumptions rather than specific stereotypes.  Maybe it's fine for the government to act on the assumption that blacks generally tend to reflect the experiences and attitudes of American blacks as a class, but wrong to act on the assumption that blacks tend to have certain specific traits or beliefs.

       It is not clear, though, how using race as a proxy for general attributes is severable from using it as a proxy for specific ones.  After all, how do we test the rationality of our race-as-proxy hypothesis?  We don't think blue-eyed people are likely to have any distinctive set of experiences and attitudes different from the brown-eyed; nor do we think so for Americans of Swedish extraction and Americans of Norwegian extraction.

       We believe race is a good proxy for some general set of race-correlated experiences and attitudes precisely because we have a pretty good idea of the specific experiences and attitudes.  We think blacks are more likely to have faced job discrimination, racial slurs, and police harassment.  We think East Asians are more likely to have faced somewhat different forms of discrimination and at the same time more likely to have certain cultural attributes that differ from those of non-East Asians.  Only by thinking about the specific experiences and attitudes of any group can we persuade ourselves that membership in this group should be considered a diversity factor.

       In fact, many of the impermissible uses of race as proxy can equally easily be characterized as general assumptions or as specific ones.  Quite a few people, when asked why they might discriminate against, for instance, blacks or Jews would say:  "Well, it's not that I think they have any specific bad traits.  It's just that generally -- doubtless based on the experiences that people in their positions have had to endure -- they just don't click with me.  On the other hand, I've found that Anglo-Saxons (or Irish or Italians), who tend to share my experiences and attitudes, are just a better fit for me."

       The same is true of a lawyer's peremptory challenges.  A lawyer who is prosecuting a black defendant accused of robbing a Korean store owner might want to strike blacks from the jury because he believes that blacks tend to share certain particular, identifiable beliefs; but he might also just generally prefer to have jurors who reflect the experiences and attitudes of the victim's racial group, rather than of the racial group to which the accused belongs. 15  Yet surely the law cannot distinguish the challenges based on specific assumptions from those based on general ones.

       Education:  Perhaps the difference lies not in the means but in the ends.  Maybe considering race as a proxy for attitudes in educational diversity programs is permissible because increasing intellectual diversity in educational institutions is particularly important. 16  Maybe achieving more educational diversity is a "compelling" interest in the sense that winning a verdict for your client or employing the best worker is not.  Likewise, maybe the answer to the business person who says "Why is it wrong for me to use race as proxy in my commercial dealings, but fine for you to use race as proxy in your college admissions decisions?" is just that "Your own decisions may be important enough to you, but they are not as important to society as is educational diversity."

       I'm not sure, though, that this response is correct.  To begin with, despite all the benefits that diversity of experiences, outlooks, and ideas brings to the educational process, education can go on tolerably well at varying levels of diversity.  Nonracial factors already make most educational institutions intellectually diverse in many respects:  People have diverse political opinions, diverse cultural backgrounds, diverse interests.

       True, college students are considerably more homogeneous in some ways than is society at large.  Most UCLA law students, for example, probably come from relatively middle-class families.  It also seems likely that their present opinions and outlooks on life are more homogeneous with respect to, for instance, the importance of education than are the opinions of non-college students their age.  But despite this homogeneity, the school is still highly intellectually diverse.  Some extra diversity may be better, but is the need for it "compelling" in the sense in which the term is usually used in constitutional law?

       On the other side, the interests in acquitting the innocent and convicting the guilty are quite important.  The interest in the efficient operation of government offices varies from office to office, but in at least some situations -- recall the police brutality investigator -- it is rather significant.

       We might doubt that the use of race as a proxy in jury selection and employment will, in the aggregate, serve government interests.  Race-conscious peremptories and race-conscious hiring might generally lead to more injustice (false convictions and false acquittals) and inefficiency (hiring based on irrational prejudices rather than rational generalizations).

       But why shouldn't we be willing to carve out an exception for the seemingly "good" uses of race as proxy, those that do lead to more accurate verdicts or greater efficiency?  After all, those who support using race as proxy for diversity purposes would probably agree that, as a general matter, considering race in admissions decisions -- for instance, as proxy for likely academic success -- is also unsound.  Still, they are ready to carve out an exception from the general prohibition for diversity decisions, presumably because diversity is so important.

       So why not generally allow use of race as a proxy in those situations where it seems to serve important interests at least as well as it does in the educational diversity context?  For instance, say a lawyer can introduce persuasive statistical evidence that a given group harbors certain prejudices.  Assume a certain racial or ethnic group is disproportionately likely to vote to acquit in rape cases even if the elements are proven, or disproportionally likely to convict even if the elements are not proven; reliable surveys are done, and members of that group disproportionately admit that they wouldn't follow the law.  Certainly if they admitted this at voir dire, they could be stricken for cause.  But not everyone does admit these things at voir dire.  Without the use of race as a proxy, it will be more likely that there will be people on the jury who will act unjustly.  Why not carve out an exemption here from the general requirement of race-neutrality?

       Likewise, there are certain jobs -- the police misconduct investigation I mentioned might be an example -- where getting the job done the best way possible is very important, and where race may be at least as good a proxy as in the educational diversity context.  Why not allow race-conscious decisionmaking in those situations?

       Perhaps the right answer here is a rule-utilitarian one:  Though we can imagine situations where race can be a valuable proxy in hiring or in jury selection, we think that considering race usually does more harm than good, and that the political and judicial systems aren't likely to do a good job of sorting the proper uses of race as proxy from the improper ones.

       But it's not clear why this rationale stops at educational diversity programs.  Why not say, as many do, that the correct rule-utilitarian answer is to ban all use of race as proxy, including in the educational context?  If we are unwilling to consider exceptions in employment or peremptory challenges, why consider an exception for educational diversity?  After all, even in the educational diversity context, some uses of race as proxy can do more harm than good. 17

       Finally, consider again the normative role of antidiscrimination law.  If someone asks, "Why is it wrong for me to use race as proxy in my business life and in my private life, but fine for you to use race as proxy in your college admissions decisions?," a rule-utilitarian defender of diversity programs can respond: In the aggregate, many people who use race as proxy in their business or private lives will do it in an irrational or bigoted way, so it's better for us as a society to conclude that this conduct is generally wrong.  On the other hand, in the aggregate, university officials who use race as proxy in their diversity programs will tend to do it in a good, socially redeeming way, so we as a society should conclude that their conduct is generally right. Is this, though, a particularly persuasive statement for the law to make?

       Antisubordination et al.:  There is, of course, an easy way to finesse the above arguments.  Educational diversity programs, one might argue, help minorities overcome the effects of racial prejudice, which should be the litmus test for whether a race-conscious program is constitutional.  Using race as proxy for goals other than diversity doesn't help overcome the effects of prejudice, and will probably only exacerbate these effects.  Maybe a few race-based employment decisions will help minorities, but perhaps those should, in fact, be allowed.

       This makes the whole argument internally consistent, but it makes it into an argument that is no longer based just on diversity, even into one that is not based on diversity at all.  To take this approach, one has to return to the contested theories about compensation, about the propriety of constitutional standards that differ depending on which group is being disfavored, or about the correctness of the antisubordination model as opposed to the equal opportunity model.  And if one accepts these theories -- if, for instance, the litmus test is whether a program is aimed at overcoming the effects of racial prejudice -- then there's no need to refer to the diversity justification.

Race and Religion

       Perhaps, though, diversity really is different.  Maybe there is something so uniting, so educationally enriching about diversity that the quest for diversity makes it proper to use race as a proxy for attitude, even though it would be improper in any other context.  Imagine then the following statement from a public law school's admissions office:

       Much to our regret, we have found that many large religious groups -- for instance, fundamentalist Christians and devout Catholics -- are grossly underrepresented among our students compared to the public at large.  We feel this deprives our students of the experiences, outlooks, and ideas that a diversity of represented religions can provide; it impoverishes discussions of such important topics as the law of abortion, the Establishment Clause, homosexuality, and others.

       We will therefore consider religion as one of the factors that can contribute to diversity.  As it happens, we find we already have many Jews, many atheists and agnostics, and many relatively unobservant Christians and Buddhists.  Because of this, we will give a preference to devout Catholics, devout Protestants, devout Buddhists, Muslims, and members of less common Christian sects.

       If your reaction is "Wow! Sounds great," this argument probably won't convince you of much (except perhaps what I say below in the "narrow tailoring" discussion). 18  But if it troubles you -- as it troubles me -- why do you find it troubling?

       Rationality:  Certainly the notion that religion is a proxy for experiences, outlooks, and ideas is quite reasonable.  One can make a strong case that religion, especially when devoutly observed, correlates with them at least as strongly as race or sex does.  Devout religious upbringings carry with them experiences that secular upbringings do not; and even more importantl , religion is inherently tied to a particular outlook on the world and a particular set of ideas.

       The theory that race inherently influences one's outlook is immensely controversial; the same is even true as to sex; but few doubt it as to religion.  Not all co-religionists agree on all issues, but if religion means anything, it must mean that the devout of a particular faith tend to see the world through a lens quite different from that used by the secular, the less devout, or the devout of other faiths.

       Constitutional Test:  Discrimination based on religion raises serious constitutional problems, but the test for discrimination among sects is strict scrutiny, the very same test used for race discrimination.  The test has been most prominently articulated under the Establishment Clause rather than the Equal Protection Clause, 19 but it's still the same:  "[W]hen we are presented with a state law granting a denominational preference, our precedents demand that we treat the law as suspect and that we apply strict scrutiny in adjudging its constitutionality." 20  The religious discrimination "must be invalidated unless it is justified by a compelling governmental interest and unless it is closely fitted to further that interest." 21  If race-based diversity programs pass strict scrutiny, why wouldn't religion-based programs?

       Degree of Underrepresentation:  Any diversity program may initially require evidence of lack of diversity.  If various racial groups are already well-represented at an institution, consideration of race might not be necessary to assure the diversity of experiences, outlooks, and ideas that racial diversity may bring.  The same is true if there are already plenty of people of all sorts of religions at the institution.

       My sense, though, is that the lack of religious diversity at many schools is at least as severe as the lack of racial diversity.  Certainly that is my impression of the UCLA law school student body.  The same seems especially true of the UCLA law faculty; there are a few relatively observant Jews and a few churchgoing Christians, but I would guess that fewer than 10% of the faculty went to services last week, compared to 37% of the public at large. 22  And focusing on the ultimate goal -- input into the educational process from particular experiences, outlooks, and ideas -- I can say I very rarely hear any distinctively Christian inputs from either my colleagues or my students. 23

       Cause of Underrepresentation:  There are many reasons, of course, why various religions might be underrepresented.  People of different religious groups may, for cultural or ideological reasons, tend to gravitate to different professions and different levels of education.  The disproportionate attraction of Jews to various professions is an obvious example.  Some religious groups tend to be poorer than others, which can also lead to underrepresentation.  Some may also argue that certain religious beliefs may simply be wrong; if so, the argument would go, education may wean people away from those beliefs, so those who are likely to have gotten a good enough education to be admitted to UCLA might have abandoned some of their foolish dogmas.

       But the goal of diversity programs is generally said to be enhancing diversity regardless of the causes of the underrepresentation.  In the race-based diversity program context, the goal is to get more experiences, outlooks, and ideas without regard to whether the existing underrepresentation flows from cultural preferences for other fields, a history of discrimination, or other socioeconomic factors (some of which, such as recent immigration, may be relatively independent of discrimination).  Who is at fault for the underrepresentation is relevant to compensatory schemes, not to diversity plans.  The same should be true in the religion context.  Of course, it might be hard to get a considerable number of people from a particular religious perspective if there are few people of that group in the normal pool of most qualified applicants, but that's true in the race context, too.

       Risk of Fraud:  The one important difference between race and religion is that religion is easier to fake.  If it is easier to get into a school as a devout Catholic than as a lapsed Unitarian, some applicants might claim a convenient conversion.  And even people who aren't actively trying to defraud might subconsciously exaggerate the extent of their beliefs.  The problem is much less common in the race context, though it arises in situations involving either mixed parentage or hard-to-define classifications such as Hispanic.

       It is hard to see, though, why the risk of fraud would make religious diversity programs unconstitutional.  If fraud were rampant enough that applicants' claimed religions proved to be highly inaccurate proxies for their experiences, outlooks, and ideas, the program might no longer be well-tailored to the intellectual diversity goal.  But any diversity program that considers anything except race and sex would have to deal with this problem already.

       If you give preferences to musicians, chess players, and French speakers, 24 you have to deal with the possibility that people will claim skills they don't have.  Of course, there will be deterrents to such claims, from natural honesty to fear of discovery, but the same applies for religion.  And if religion is indeed a better proxy for the ultimate criterion (one's experiences, outlooks, and ideas) than race is, then it might remain a better proxy (or at least an equally good proxy) even after one factors in the fraud rate. 25

       Narrow Tailoring:  If the above analysis is right, it suggests that considering religion in diversity programs is not just permissible -- it might be required, so long as race is considered.

       Recall one of the main thrusts of Powell's Bakke opinion:  A program that considers only race isn't narrowly tailored to the goal of intellectual diversity because it elevates race above other diversity factors that could be equally important. 26  This is an underinclusiveness argument -- if the policy focuses only on race and ignores other key diversity factors, it is not narrowly tailored to the diversity goal.

       The same is true of diversity systems that give strong consideration to race but none to religion.  Religion says much more about the experiences, outlooks, and ideas that a person brings than does being a musician, a chess player, or a French speaker. 27  It might even say more than, or at least as much as, race or geographical background or even "exceptional personal talents, unique work or service experience, leadership potential, maturity, demonstrated compassion, a history of overcoming disadvantage, [or] ability to communicate with the poor." 28

       What justification is there for including race and these other factors in the consideration, but excluding religion?  It couldn't be that we're reluctant to consider religion because it is a suspect classification -- the same is true of race.  Perhaps it's that claims about religion are subject to falsification or exaggeration, but so are claims of compassion, overcoming disadvantage, or ability to communicate with the poor.  Maybe it's that there is no perceived underrepresentation of various religious groups -- that the school will get a good deal of religious diversity in any event -- but on some campuses and in some departments (and certainly on some faculties) this underrepresentation does exist.

       So excluding religion as a factor but including race might suggest that the program is not narrowly tailored.  It might also suggest, as underinclusiveness often does, that the actual purpose of the program isn't really the stated purpose; here, that the real purpose isn't actually diversity of experiences, outlooks, and ideas as such, but rather something else -- perhaps just racial or ethnic diversity, a justification that Powell's Bakke opinion specifically condemned. 29

The Trouble with Diversity

       The religious diversity hypothetical suggests a simple point:  There is something troublesome about sorting people by certain attributes, be they religion or race, even when this sorting is a rational and well-intentioned attempt to serve a legitimate, even important, social goal.  I would prefer that my law school be more religiously diverse, because I do think this would make it more intellectually diverse, and therefore a better educational environment.  But even the goal of intellectual diversity ought not be served by the means of sorting by religion.

       The same follows from some of the other examples where race (or sex) might be used as a proxy.  We condemn the use of race as a proxy not just when it is irrational or bigoted, but even when it is rational and unbigoted.  We believe people ought to be judged by their personal traits, not by the sex or the racial group to which they happen to belong, even when the group membership does provide reliable clues to the members' attitudes and experiences.

       We read the Equal Protection Clause this way with regard to government decisions, and the law often asks people to behave this way in their privat lives too.  Much of the grand civil rights project of the late twentieth century has been aimed at persuading people to shift to this approach -- no easy task, given that the use of race and ethnicity as proxy has been the dominant approach throughout human history, and continues to be dominant throughout much of the world today.

       Race-based diversity programs, it seems to me, are deeply corrosive of this project.  When the government condemns a practice as pernicious and immoral, and then engages in a similar practice itself, people are likely to ask:  How exactly does what you praise differ from what you damn?  The intellectual diversity justification does not, in my view, provide a satisfactory answer.  


    *    Acting Professor, UCLA Law School (volokh@law.ucla.edu).  I'd like to thank Akhil Amar, Stephen Bainbridge, Alan Brownstein, Evan Caminker, Julian Eule, Robert Goldstein, Jerry Kang, Kenneth Karst, and Jonathan Varat for their help.

    1.    Regents of the Univ. of Cal. v. Bakke, 438 U.S. 265, 314 (1978) (opinion of Powell, J.).

    2.    See id. at 307, 311-14 (distinguishing the interest in intellectual diversity from an interest in racial or ethnic diversity as such).

    3.    Cf. Powers v. Ohio, 499 U.S. 400, 410 (1991) (holding unconstitutional the use of race as proxy for juror bias in peremptories).

       I recognize that the peremptory challenge cases are complicated by the fact that they also involve the defendant's rights, not just the juror's; but the Court has concluded that jurors themselves have a right not to have their race be used as a proxy for their attitudes, even without regard to the defendant.  Id.; cf. Georgia v. McCollum, 505 U.S. 42, 59 (1992) (prohibiting the use of race as proxy in peremptory challenges even by the defendant's own lawyer).  In any event, there seems to be no reason why the presence of the defendant's interests should affect the propriety of using race as a proxy.

    4.    See J.E.B. v. Alabama ex rel. T.B., 114 S. Ct. 1419 (1994) (holding unconstitutional the use of sex as proxy for juror bias in peremptories).

    5.    Some might argue that diversity is always practically better than uniformity, but this strikes me as a rather controversial claim.  I doubt a court would be willing to embrace such a broad assertion in the absence of persuasive empirical evidence.  Cf. Stephen M. Bainbridge, A Conservative Contractarian Critique of "Progressive" Corporate Law, 82 Cornell L. Rev. (forthcoming 1997); Robert Drago, Share Schemes, Participatory Management and Work Norms, 23 Rev. of Radical Pol. Econ. 55, 58-59 (1991) (same as to gender homogeneity); Donald McCloskey, Bourgeois Virtue, Am. Scholar, Spring 1994, at 177, 183 (same as to ethnic and religious homogeneity); Marcel Fafchamps, Market Emergence, Trust, and Reputation, at 36 (unpublished manuscript, on file with author) (describing possible practical advantages of ethnic homogeneity in certain contexts).

    6.    I should say I'm making no great claim for the originality of the argument in Parts II and III, see, e.g., Richard A. Posner, The DeFunis Case and the Constitutionality of Preferential Treatment of Racial Minorities, 1974 Sup. Ct. Rev. 1, 7-15, though I hope some of the particular points I make below are unusual enough to be useful.

    7.    But see Regents of the Univ. of Cal. v. Bakke, 438 U.S. 265, 318 (1978) (opinion of Powell, J.) (suggesting, in my view incorrectly and not consistently with other parts of the opinion, that a program is not even intentionally discriminatory "where race or ethnic background is simply one element -- to be weighed fairly against other elements -- in the selection process").

    8.    Here, as in other places throughout this section, I speak of "Jewish" as an ethnic group rather than as a religion.  Of course, it's both; one can, for instance, be both a Jew and an atheist or an agnostic.  Cf. Shaare Tefila Congregation v. Cobb, 481 U.S. 615, 617-18 (1987) (holding that being Jewish is a "race" for purposes of 42 U.S.C. § 1982); Salehi v. INS, 796 F.2d 1286, 1288 (10th Cir. 1986) (claim of fear of persecution based on "Jewish ethnic and religious background"); Hampel v. City of Denver, 886 F. Supp. 756, 758 (D. Colo. 1994) (Title VII claim of discrimination based on "Jewish ethnicity").

    9.    See Village of Arlington Heights v. Metropolitan Housing Dev. Corp., 429 U.S. 252, 265 (1977) (The Equal Protection Clause "does not require a plaintiff to prove that the challenged action rested solely on racially discriminatory purposes. . . . When there is a proof that a discriminatory purpose has been a motivating factor in the decision, [rational basis scrutiny] is no longer justified.").

    10.    See Bakke, 438 U.S. at 314-18 (opinion of Powell, J.).

    11.    But see Akhil Reed Amar & Neal Kumar Katyal, Bakke's Fate, 43 UCLA L. Rev. 1745, 1772-78 (1996).

    12.    I recognize that many of the predictors used to determine who is "more qualified" are quite unreliable, and I certainly don't mean to endorse here any particular set of qualifications, such as grades, SAT scores, and so on.  Limited as the predictors are, however, decisionmakers have to look at something.  My only contention here is that, regardless of the qualifications one looks to, distinguishing preferences for the unqualified from preferences for the less qualified is unsound.

       I have heard some people suggest that because all predictors are imperfect, we should have fewer misgivings about using race to supplement those predictors, but I don't see how that follows.  The difficulty of deciding who will be a good student or a good employee is never seen as a justification for preferring, say, white candidates over minority ones; I see no reason why it should be a justification for preferring racial minorities over whites.

       Some have argued that belonging to a racial minority is itself part of the qualifications, precisely because minority students bring different perspectives and experiences into the classroom, but this assertion returns us to the main question:  Why is it impermissible to consider being white a qualification -- even when being white is a statistically reliable proxy for valuable experiences or attitudes -- but permissible to consider being nonwhite a qualification?

       Others argue that belonging to a racial minority group should be a qualification because minority graduates will provide role models for future minority students.  This argument, though, is not an intellectual diversity argument at all; to accept it, one has to accept a justification for race-consciousness beyond simple intellectual diversity.  Cf. Wygant v. Jackson Bd. of Educ., 476 U.S. 267, 288 n.* (1986) (O'Connor, J., concurring in part and concurring in the judgment) (distinguishing intellectual diversity justification from role model justification).

    13.    See Amar & Katyal, supra note 11, at 1773-75.

    14.    One colleague of mine has suggested that diversity programs in college admissions are permissible in part because they widely spread the burden, as well as the benefit:  Unlike proxy-based hiring, which may burden a few people, diversity admissions generally spreads that burden over thousands.  Cf. Wygant, 476 U.S. at 282-83 (plurality opinion) (endorsing the view that race classifications may become more permissible as the burden they impose becomes more diffuse).

       It's not clear to me that the diffuseness of the burden should generally be particularly important, but in any event it can't be the distinction here.  We don't tolerate the use of race as proxy in admissions generally -- for instance, the use of race as proxy for likely academic success -- though of course the burden is every bit as diffuse as with diversity admissions.  Likewise, the use of race as proxy in a hiring context where the burden of lost opportunity is spread widely -- say, thousands of people being considered for several hundred jobs, numbers that aren't far from the normal university admissions numbers -- seems no more permissible than in other hiring situations.

    15.    Some might believe that such lawyers do in fact have in mind some special stereotypes, and not just general assumptions.  But why shouldn't we be equally confident that admissions and hiring offices have some equally specific stereotypes in mind?  Conversely, if we're willing to give the benefit of the doubt to the university officials, why aren't we willing to give it to the lawyer in the example?

    16.    One could also argue that diversity programs are different because they ensure that people will be exposed to people of other races and will therefore learn to respect and appreciate them.  This, though, is no longer an interest in intellectual diversity -- diversity of "experiences, outlooks, and ideas" that would otherwise be left out -- but specifically in racial or ethnic diversity as such.  See Bakke, 438 U.S. at 307, 315 (opinion of Powell, J.) (drawing such a distinction, and concluding that only an interest in intellectual diversity was compelling).

       Moreover, regardless of what Justice Powell thought, I'm not sure that the interest in teaching respect and appreciation, important as it is, is important enough to justify race classifications where other interests do not; and it's not at all clear that race-conscious affirmative action programs do, indeed, increase respect and appreciation for those of other races.  Besides the risk of resentment (justified or not) from people who belong to the burdened groups, race-conscious programs artificially create a situation where race becomes a good predictor of success.  If a school admits students of group X with worse average predictors than those of group Y, the Xs will on average do worse; when fellow students see this, it may, instead of teaching respect and appreciation, reinforce the very prejudices that the system was meant to undo.

       A student from another law school (who was of Chinese extraction, for whatever that's worth) recently remarked to me that he and some of his acquaintances had a rule of thumb when selecting classes:  If there were many blacks and Hispanics in the class, he'd know his grade would likely be higher.  The grading was done on a forced curve, and it was a good bet that the black and Hispanic students -- who were generally admitted with considerably lower predictors than the white and East Asian students -- were likely to get the lower grades.

       This was hardly a praiseworthy attitude, but not an irrational or even a bigoted one.  When the school creates an environment where race is a good predictor of success, people learn to take advantage of this signal.  Use of race as proxy begets more use of race as proxy.  Maybe this is an unusual incident; maybe race-conscious admissions do, in the aggregate, lead to greater appreciation and respect for people of other races.  But this doesn't seem like an easy case to make on purely intuitive grounds, and I know of no evidence that supports it empirically.

    17.    See, e.g., Jerry Kang, Negative Action Against Asian Americans: The Internal Instability of Dworkin's Defense of Affirmative Action, 31 Harv. C.R-C.L. L. Rev. 1, 14-15 (1996).

    18.    Akhil Amar and Neal Katyal might approve of considering religion in diversity admissions; they suggest that growing up Amish would be a significant diversity factor, see supra note 11, and I take it they'd say the same about growing up as a fundamentalist Southern Baptist or a devout Catholic or a Scientologist -- not groups that are as culturally insular as the Amish, but certainly ones that have relatively distinctive outlooks, ideas, and (often) experiences.

       The President of Princeton University, quoted by Justice Powell in Bakke, might agree.  See Bakke, 438 U.S. at 312-13 n.48 (opinion of Powell, J.) (quoting the President as describing "some of the benefits derived from a diverse student body" as including "-interactions among students . . . of different races, religions, and backgrounds´").

    19.    The Court has often said that the Equal Protection Clause applies to religious classifications to the same extent as it does to racial ones, but it has said this only in dicta.  See, e.g., Miller v. Johnson, 115 S. Ct. 2475, 2485 (1995); Burlington N. R.R. Co. v. Ford, 504 U.S. 648, 651 (1992); American Sugar Refining Co. v. Louisiana, 179 U.S. 89, 92 (1900).

    20.    Larson v. Valente, 456 U.S. 228, 246 (1982).

    21.    Id. at 247 (citations omitted).

       The Constitution permits some preferences for religion generally, especially when the preferences aim at accommodating sincerely held religious beliefs.  But in such cases, the preference must turn on a person's specific beliefs, not on his sectarian affiliation.  See, e.g., Board of Educ. of Kiryas Joel Village Sch. Dist. v. Grumet, 114 S. Ct. 2481, 2497 (1994) (O'Connor, J., concurring in part and concurring in the judgment): What makes accommodation permissible, even praiseworthy, is not that the government is making life easier for some particular religious group as such.  Rather, it is that the government is accommodating a deeply held belief.  Accommodations may thus justify treating those who share this belief differently from those who do not; but they do not justify discriminations based on sect.  A state law prohibiting the consumption of alcohol may exempt sacramental wines, but it may not exempt sacramental wine use by Catholics but not by Jews. See also id. at 2493 ("[W]hatever the limits of permissible legislative accommodations may be, it is clear that neutrality as among religions must be honored." (citations omitted)); Employment Div. v. Smith, 494 U.S. 872, 886 n.3, 890 (1990) (requiring strict scrutiny for sectarian classifications but expressing approval for "nondiscriminatory religious-practice exemption[s]").

    22.    Larry B. Stammer, Church Attendance Falls to 11-Year Low, L.A. Times, Mar. 2, 1996, at B4.  Broken down by age range, the fractions of the population that have attended religious services within the last week are:
18 to 3034%
31 to 4931%
50 to 6846%
69 and over48%

    23.    Note the results of a recent study comparing the religious makeup of law faculties with the religious makeup of the full-time working population ages 30-75:
Religious Group Law Professors Full-Time Working Population
Catholic/Orthodox Christian13.7%26.0%
Protestant/other Christian32.4%60.5%
Total Christian46.1%86.6%
Jewish26.4% 2.0%
No Religion26.4% 8.3%
Other Religion 1.1% 3.1%
James Lindgren, Measuring Diversity tbl. 2 (May 26, 1996) (unpublished manuscript, on file with author).  The study did not purport to measure devoutness; nonobservant Protestants, for instance, would still be counted as Protestants so long as they identified themselves as such.

       Though this is beyond the scope of the Article, it's curious that the two groups that are most underrepresented among law faculties relative to the working population are Hispanics and Republicans (including people who lean Republican); the numbers for them are:
Group Law Professors Full-Time Working Population
Hispanics 2.3% 7.6%
Republicans(including leaning Republicans)12.9%41.0%

       I do not mean to endorse the proposition that faculties should aim for proportional representation, either by race, religion, or political beliefs; nor do I want to suggest that all supporters of the intellectual diversity rationale would favor such a system.  But the percentages do suggest that certain religious groups -- especially if one focuses on the devout of those groups, who are probably a fairly small fraction of the groups at large -- are so unrepresented that favoring them would increase intellectual diversity.

    24.    Amar & Katyal, supra note 11, at 1772.

    25.    Some readers suggested a related concern: that people's choice of religion might be influenced by the availability of government benefits.  I agree this is something we would rather avoid; perhaps that's one reason the Court has required strict scrutiny of sectarian preferences.  But we'd also prefer to avoid race classifications.  The pure diversity argument (as opposed to, say, the argument that diversity programs are permissible because they benefit nonwhites, and that the standard for laws that benefit nonwhites should differ from that for other race classifications) concedes this, and accepts that race classifications are allowed only if they pass strict scrutiny.  Why doesn't the interest in intellectual diversity trump our concern about influencing the choice of a religion, just as it trumps our concern about sorting people based on race?

       Perhaps one might argue that the concern about influencing choice of religion is so strong that it should lead not just to strict scrutiny of sectarian discrimination, but to a per se prohibition.  But, first, that's not what the Court has held and, second, I'm not sure why this would be so.  Influencing someone's choice of faith might be bad, but why is it any worse than race discrimination?

    26.    Bakke, 438 U.S. at 315-16 (opinion of Powell, J.).

    27.    Amar & Katyal, supra note 11, at 1772-73.

    28.    Bakke, 438 U.S. at 317 (opinion of Powell, J.).

    29.    Id. at 315; cf. City of Richmond v. J.A. Croson Co., 488 U.S. 469, 493 (1989) (O'Connor, J., plurality opinion) ("[T]he purpose of strict scrutiny is to `smoke out´ illegitimate uses of race . . . .").