Chapman Law Review
The Federalist Society for Law and Public Policy
CIVIL RIGHTS: RACIAL PREFERENCES IN HIGHER EDUCATION
2005 National Lawyer’s Convention
November 12, 2005
Copyright (c) 2008 Chapman Law Review
Professor Richard O. Lempert, Eric Stein Distinguished University Professor of Law and Sociology, University of Michigan
Dr. John Lott, Resident Scholar, American Enterprise Institute
Professor Stanley Rothman, Director, Center for the Study of Social and Political Change, Smith College
Professor Richard H. Sander, University of California, Los Angeles School of Law
Professor Douglas W. Kmiec, Professor of Constitutional Law & Caruso Family Chair in Constitutional Law, Pepperdine University School of Law (moderator)
PROFESSOR KMIEC: Our first speaker today is Richard Sander. He received his bachelor’s degree from Harvard, and a law degree and a doctorate in economics from Northwestern. He has taught at UCLA since 1989. He has done considerable empirical research on social policy and legal education. But beyond the classroom, he has worked tirelessly for many years to improve the enforcement of fair housing laws in Southern California and helped start a program that has substantially increased the participation of low income Los Angeles workers in the Earned Income Tax Credit program.
Please welcome Professor Richard Sander.
PROFESSOR SANDER: Thanks very much, Doug. Those were kind words.
I am very happy to be here at this meeting of the Federalists. I have given about 30 presentations on this work over the last year, and I think this is the third audience that might be sympathetic to some of my arguments. But beyond that, speaking as a–I guess I consider myself a progressive Democrat–I think that along with the majority of my colleagues, all professors across the board realize that the Federalist Society plays an enormously important role in law schools as a center of intellectual inquiry by trying to get debates going and getting discussions going across ideological divides. And that is an irreplaceable role that I appreciate and I hope you continue to foster.
I am going to be very brief, because we have a lot of material to try to cover in this panel, and I think the most interesting part of these discussions is usually the question-and-answer session. I have brought along a lot of material on PowerPoint to get into specifics. I left some copies of my work at the back, but I think those have disappeared. I encourage you to follow up on any of the things that I raise and try to engage all of us on this debate on the most specific terms possible.
Let me just say that globally, I think that in addition to trying to make some arguments about affirmative action, trying to get some data into the public realm, maybe the fundamental goal of this research was to provide an opening for discussion. Professor Hoffman and I are unusual because race continues to be an enormously sensitive topic, and I think it is nowhere as sensitive than at the American academia, and it is very hard to have open and informed discussions about the pros and cons about things like affirmative action.
Our debate over the last 25 years has mostly focused on whether affirmative action violates anti-discrimination laws, equal protection laws, and so on. That is a fairly straightforward, neutral, abstract ground on which we can engage the issue, and that is an important ground. But I think that ultimately, it is tremendously important to engage the empirical realities of what the effects of preference policies are. And I say that not just because I am very concerned about the effects of preference policies on minorities, but because I think that even if Grutter had been decided the other way, even if Congress passed something like Prop 209 nationwide, practices at schools would not change dramatically.
Doug mentioned the various idiosyncrasies of California. One of them is that we have Prop 209, and I have been able to look close-up from the inside into its effects. And although there are widely varying levels of adherence to the letter of 209, there is very little adherence to the spirit of it. Schools feel themselves to be under such tremendous imperative to achieve racial diversity that they have found lots of other ways to do it. So I do not think that the mismatch effect eight years after or nine years after Prop 209 has really been significantly difficult in California. We may be able to talk about that a little more in the discussion period, too.
But ultimately, change in practices will only occur if we change the hearts and minds of American academia. I know that is a frightening thought, but I really think that is where the battle needs to be engaged. My fundamental message is that there needs to be more discussion; there needs to be more research and more data on these issues. We have got to pursue it. And the law school world is a very good way to pursue it, because we have unusual conditions that make it possible to analyze what the effects of preferences are.
At the undergraduate level, most of the debate on preferences concerns graduation rates, which is ultimately somewhat fruitless. I mean, it does not really tell us anything about how successful the educational process is. If a school like Yale has a 98- or 99-percent graduation rate, you could say that almost all African-Americans at Yale ultimately get a degree.
The real question is what happens to long-term outcomes. What happens to the amount that you have learned in school? And the fact that law schools have fairly standardized curricula and grading systems and a bar that people take across the country makes it a rich venue to try to understand these questions.
So, I am going to very briefly summarize–what do I have? Like, five minutes?
PROFESSOR KMIEC: You have eight minutes.
PROFESSOR SANDER: I have eight minutes.
I am going to briefly summarize the key arguments in systemic analysis and save most of the specific data discussion for later. There are really five different arguments I am trying to make in the article. The first one concerns the admissions process. As Doug said, essentially I am trying to put forth two propositions. One is that what law schools do in their admissions process is essentially indistinguishable from what the Supreme Court ruled to be illegal in the Gratz case. In other words, schools do one of two things: they either use mechanical race norming procedures to equalize the disparities between blacks and whites on background credentials–same thing for Hispanics and whites, or Native Americans and whites–or they simply segregate admissions and try to achieve the same proportional result in admission outcomes as is represented in the applicant pool itself. And there are a variety of ways of showing that.
We have got quite strong data on it, and I think that argument is overwhelming. There has been relatively little critical commentary on that particular point; maybe Rick or John will make some points about that today, but I have seen very little engagement on that. And we are now starting to see other scholars produce data and empirical analyses reaching the exact same conclusion. So I think that point is pretty established, and that is a pretty devastating point, from a legal point of view.
But what is less appreciated is that if elite schools do this process of race norming, or segregating admissions, then the schools that are further down in the hierarchy of law schools really have no choice but to follow suit, or to be all white, or all white and Asian. So, for example, if Harvard, Yale, and Stanford, exercise aggressive racial preferences, and through that process absorb all the students who would be admitted to UCLA, on race-blind criteria, then UCLA must either choose to not have any minority students or must imitate those preferences and reach further down in the applicant pool. As a result of this, what I call the cascade effect, you see a striking similarity in the credentials gap, all up and down the legal academy.
In this talk today, I am going to use something called the Academic Index, which is simply sort of a normalization of LSAT scores and undergraduate grades to a scale that goes from zero to 1000. The law schools are generally comprised of students who have academic indices between 400 and 900. And the typical black-white gap in American law schools is 170 points, a very large difference that roughly corresponds to the test score gap that we see in studies from elementary school, on. That 170 points is essentially identical, whether you look at the most elite schools or schools that are ranked 150. The only place where you do not find it are the historically minority schools because they have very large minority populations, they do not exercise the same type of preferences, and they have uniquely equal student bodies in terms of the level of credentials of students of different races. One result of that, I think, is that Howard University, for example, is, I think, eighth on the list of law schools recruited by large law firms. It is a unique place where employers can go and find students who have very high academic achievement, blacks with very high academic achievement, and Hispanics with high academic achievement in very large numbers.
So, point number two is that as a result of these really large credentials, and not as a result of race, minority performance, particularly black performance, in law school suffers. The argument is not that LSAT and undergraduate grades are perfect predictors of performance; they are not, although they are better than is usually reported because most analyses look within individual schools and have restriction of range problems that limit their statistical power. But for groups, the LSAT and UGPA are quite predictive of overall results, so that we see the credentials gap very closely mirrored in the performance gap in first-year grades, second-year grades, and third-year grades.
Now, I think there is some debate about whether that credentials gap explains 100 percent of the performance gap. I think that it is pretty close to 100 percent; maybe 95 percent. But I do not think anyone has put forward an argument that it is less than 85 or 90 percent. In other words, under-performance by race or things that are uniquely the case about blacks or Hispanics are not driving the fact that there are worse performances in grades, and certainly not worse performances on graduation or bar exam scores. It is being driven by the system of preferences.
Now, again I get one pick up on the point Doug made, that it is important to distinguish here that if we got rid of–if we completely eliminated what I call the mismatch effect, you would not have identical graduation and bar results across racial lines because there is this difference in incoming credentials. My argument is that the mismatch effect, the effect of blacks having much lower grades in school, roughly doubles these disparate outcomes, so that we roughly double the number of blacks who do not become lawyers as a result of these policies.
The third point is that grades are tremendously important in terms of a whole range of outcomes. They are the most important factor in determining whether you graduate from law school. I think that is fairly intuitively obvious. And they also turn out to be extremely important in terms of your bar performance, and in terms of how you are evaluated by employers. And I think actually, they are quite significant in terms of later career performance.
Prestige clearly matters on some of these factors. It clearly helps to go to a more elite school in terms of graduating. And the evidence is somewhat mixed, but I would not be surprised if there was a net positive effect of going to a more elite school and passing the bar. It certainly is helpful in the employment market. But in every analysis that I have done and I have seen other people perform, grades do dwarf the effect of prestige. That means that if you have to choose between having lousy grades and having a lower-tier school, you are better off going for the lower-tier school and getting a good GPA. GPA is going to dwarf the effects of prestige and a range of practical outcomes.
And that, in turn, implies that there is something going on at law school that means that if you’re at the very bottom of your class, the bottom 10 or 15 percent, you are actually learning less than you would if you were at another law school. Now that strikes people as counter-intuitive, and it is the weakest part of my argument in the sense that I do not have any direct data sort of observing the learning process. I cannot say exactly where in the first semester or the second semester this is kicking in; from, like, weekly quiz results of students or something like that. But it is the only conclusion that follows from the fact that grades drive outcomes so much more heavily than school eliteness does. And it is supported by a lot of anecdotal evidence that professors, when they teach classes, tend to observe that final exams and other evaluations from the bottom of the class show a level of confusion that is really qualitatively different from that experienced by a lot of other students. So there is something going on when you are at the bottom of the class, regardless of your race, that really affects the quality and quantity of education that you receive.
As a result of that, you see these striking patterns that–well, why do I not go forward and show you a little bit of the math.
PROFESSOR KMIEC: Time is up.
PROFESSOR SANDER: Oh, time is up? Then let me just say to summarize quickly. One is that in the job market, we again see this very powerful effect of grades versus prestige, that across most of the range of law schools, grades dominate prestige as a hiring factor. I estimate that because of the lower grades black students receive as a result of preferences, their average earnings are about $10,000 lower than they would be otherwise. So it has an important and powerful effect after graduation, after bar passage.
And finally, the issue of what would happen to overall production of lawyers, essentially what I find is that because of the mismatch effect, there is a very large attrition of blacks. There is really no question about that. And the issue is, what would happen to the number of blacks admitted into law school if one reduced or eliminated preferences? First of all, that is only one policy option. I hope we will get to talk about all of the policy options during the Q&A.
But if we took the extreme step of eliminating preferences, I estimate that the number of blacks admitted to law school would only drop modestly, 10 to 20 percent, and that that drop would be offset by the dramatically lower attrition rates. But the real point here is not whether there would be an eight percent increase or an eight percent decrease, but that the traditional claim that preferences are the only way to have a significant minority presence in law school is clearly wrong, and I think that point has been largely abandoned by the critics.
PROFESSOR KMIEC: Our next presenter is Stanley Rothman, the Mary Huggins Gamble Professor of Government Emeritus at Smith College, Director of the Center for the Study of Social and Political Change. He received his doctorate in government from Harvard. He is the author of numerous books, including American Elites: Hollywood America, Social and Political Themes in Motion Pictures, Environmental Cancer–a Political Disease, The Least Dangerous Branch? — there is a question mark after that. And he is currently chairman of the National Association of Scholars.
Please welcome Dr. Rothman.
DR. ROTHMAN: I am getting too old for this. Thank you for the introduction. I am really pleased to be here, or at least ambivalent.
The current situation in law with respect to affirmative action or diversity is rather paradoxical in some ways because affirmative action–when students were admitted under affirmative action, the notion was that blacks had a special case, and so did American-Indians, because of the treatment which they had received, and they therefore were entitled to some special prerogatives. But if the decisions are made now on the basis of diversity, then the black exception disappears, and Hispanics have as much right as blacks, or African Americans, to be admitted on some sort of basis which–I will not call it quota, but which is quota-related.
This has two potential problems. One is that a conflict between blacks and Hispanics could escalate very sharply, under certain circumstances. The other is that, whereas the Affirmative-Action basis for special admissions had a moral case, the diversity is what makes it a pragmatic case, namely that diversity would affect all students in a very positive way, and therefore should be used. But if it turns out that the data shows otherwise, the Court would have to revise its decision. And that is something which I suppose the Court would not like to do, at least at the present time. We do not know what the Court will do ten years from now.
In any event, the argument that diversity enhances education and interracial understanding rests upon a fairly significant body of social science literature. Unfortunately, most of this literature relies upon surveys of students and sometimes faculty and administrators. Surveys are crude instruments, indeed, and the results often produce interminable debates which vary so very much. So much for social science.
Reporter data, truth in memories, raise a host of problems ranging from selective recall to various response sets, which may be inadvertent or unconscious. Many of the most prominent studies contain questions which are quite dubious, such as, “how much has a diverse student body helped you work more effectively and/or get along better with members of other races?” It is a pretty leading question. In short, many social science surveys that support the argument that we need the benefits of diversity in college curriculum and college enrollment cannot preclude the possibility that favorable responses represent a shared mentality, rather than represent a valid inference of real-world effects. In this case, the argument that diversity is beneficial becomes circular. Students are taught that diversity is opposed only by racists and the misinformed. They are then asked whether diversity is valuable, and their positive replies are often seen as proof that diversity is valuable.
In order to avoid such problems in testing the hypothesis that enrollment diversity programs benefit the academic community, we chose a more indirect approach. In addition to standard questions, we asked students, faculty and administrators–three major university constituencies–to evaluate the educational experience offered by the college with no reference to diversity. Then we correlated their attitudes with separate, though incomplete, measures of racial diversity, namely the proportion of African American students. The study is part of a larger one which compares American and Canadian universities, which we have not gotten to yet. We hope to get there.
Our U.S. sample totaled 1,643 students, 1632 faculty, and 808 administrators. Our Canadian sample, of course, is smaller. Our survey addressed several aspects of campus opinion with regard to diversity. There is strong support among faculty, students, and administrators for the need to create multicultural courses, but not to make them mandatory. None of the constituencies agreed with the statement that “this university pays too much attention to minority issues.” On the other hand, the majority of all three groups opposed hiring or admission using race or gender as counters.
Eighty-five percent of the students agreed with the statement that “no one should be given special preference . . . on the basis of their gender or race.” Only 56 percent of the faculty and 48 percent of the administrators agreed with them. Interestingly, 75 percent of the students, with only 57 percent of the faculty, supported relaxation of admission standards on minority groups in order to increase their chances for admission. However, all three groups–81 percent of faculty, 76 percent of the students, and 83 percent of administrators–massively oppose hiring faculty. I’ll let you speculate as to the reasons for the shifts in faculty views on the two questions.
We asked many other standard questions. I’ll report on just one more finding. Most of the members of all three groups believed that the effort to admit more minorities had no effect on university standards, and of the large minorities, 25 percent or more believe that academic standards have been lowered, and only very small numbers believe that standards have been raised.
Now, the major focus–this is a side effect–the major focus of the study is based on four general questions testing perceptions of educational environment and three questions on perceptions of discrimination and the treatment of minorities. Not all questions were asked of all three samples. The point is that these questions were asked at the very beginning of the questionnaire and had nothing to do with diversity or affirmative action. We did not ask them that. We asked them how satisfied are they with their university experience, which we asked only the students. How well does the school educate its students? How hard do students work at their studies; we asked that of all groups. And how well prepared academically students were upon entering; we asked only faculty and administrators that.
Respondents were also asked whether minority students were treated better, worse, or about the same as white students; the extent to which racial discrimination is a problem at their institutions; and whether they personally had been treated unfairly because of their race, ethnicity, gender, or sexual orientation, or religious beliefs, or political views. These questions were asked of all groups.
Now, according to the original diversity model, increases in black enrollment should produce positive assessments over time from students regarding enrollment and regarding their educational experience. However, the correlations were in the opposite direction. As the proportion of black students rose, student satisfaction with the university experience dropped as to their assessments of the quality of education and work ethics of the students in general. In addition, the higher the enrollment diversity, the more likely students were to say that they personally experienced discrimination. The same pattern of negative correlations between educational benefits of increased black enrollment appeared in the responses of both faculty and administrators. We did only a little bit with Asians and Hispanics because there were not enough of them in the sample, so this study will not be able to do very much with them.
Faculty members also rated students as less hard-working as diversity increased. Each valuation of college life that produced a significant correlation with enrollment diversity, defined as a proportion of black students, was tested by a regression that included a host of background variables, and which I shall not go into. But there are a lot of them; if you want to know, I will tell you later. It will take me the rest of the time just to reel that off.
PROFESSOR KMIEC: Two minutes.
DR. ROTHMAN: Two minutes. Okay.
Well, the higher the enrollment diversity, the more likely students were to say that they personally experienced discrimination. The same pattern of negative correlations between educational benefits and increased black enrollment appeared in the responses of faculty and administrators. Both groups perceived the decreases in educational quality and academic preparation as the number of black students increased. Faculty members also rated students as less hard-working as diversity increased. We did not ask them about diversity. We just asked them these general questions and did the correlations. And they are statistically significant, at the .01 level. They do not explain that much of a variance, but no social science study ever explains much of a variance. That is why people report the correlations.
Anyway, the argument has been made that actually admitting a larger number of black students is not the issue in terms of producing, more recently, better results. But you have to have enough black students so they can do other things, like have interactive seminars, have them meet in classrooms, get to know each other better, and see that they differ from each other. This argument goes one step back, namely that you can find the necessary, but not sufficient, conditions by having a larger number of black students in the class.
I am not persuaded by this. The studies I have seen are not persuasive. My experience does not persuade me that that is reasonable. Now, of course, we do not know for sure. I am trying to figure what to leave out. I–well, I am not going to say that. It will come out in the question period.
This does not mean that diversity is a goal that must necessarily be abandoned. The increased presence of black and Hispanic students produced in part by affirmative action or diversity does not seem to improve the educational interracial environment. However, the increased presence of Asian-Americans, who are often excluded from preferential programs such as Michigan’s, seem to have at least some positive impact. It may well turn out that the desired benefits of a more diverse educational environment will be more rapidly treated if we do not pursue the goal too impatiently. However, much work remains to be done before we can make any strong statement about the advantages and disadvantages of diversity policy. Thank you.
PROFESSOR KMIEC: Thank you, Dr. Rothman, for your tolerance of my time quotas.
The next two speakers are respondents, and they are both quite skilled in statistical analysis. For the benefit of the non-statisticians in the audience, I would ask each of them to translate their remarks.
Professor Richard Lempert is the Eric Stein Distinguished University Professor of Law and Sociology at the University of Michigan. He is currently serving as the Director for the Social and Economic Sciences at the National Science Foundation. Dr. Lempert received his BA from Oberlin, his M.A. and Ph.D. in sociology from Michigan, and his law degree from the Michigan Law School. He has authored and co-authored numerous books and articles, including An Invitation to Law and Social Science and A Modern Approach to Evidence. His areas of particular investigation include the jury system, dispute processing, and affirmative action, as well as capital punishment. Please welcome Dr. Lempert.
DR. LOTT: He wants to go last.
PROFESSOR KMIEC: Do not welcome Dr. Lempert at this time. We will quiz him later.
The intervening respondent is Dr. John Lott, who is well known to us because he is a resident scholar at the American Enterprise Institute. He received his doctoral training at UCLA. Dr. Lott has held a number of academic positions at Chicago, Yale, Stanford, UCLA, Wharton, and Rice; also the chief economist for the US Sentencing Commission; has done profound work on the bias against the guns. He is currently completing a book on the importance of reputation and deterring criminals.
Please welcome John Lott.
DR. LOTT: Well, thanks very much. I am honored to be here with such distinguished people on the panel. I suppose I first realized that there was a real problem in this area back when I was at Wharton. I kind of drew the short straw and had to be on the Admissions Committee for a couple years. Just to mention how fundamental this problem is, one of the things that I just could not get over was if you look at the number of African Americans who scored over 700 on the math portion of the SAT in the early 1990s, it was 26 in, I think it was 1992. And you had a school like Wharton where you have 4,000 undergraduates, and they average on the SAT combined totals of 1,450. Just try to figure out how you are going to–if even one school were to get ten percent of those students to go there, how many would be left for everybody else? So, the cascading problem that Richard brought up was something that I have thought about a little bit.
Now there are two sets of papers here. Both sets of these papers come to essentially the same conclusions, but I do not think the two sets of papers are equally strong. Let me start first with the research by Rothman, Lipset, and Nevitte. They provide two interesting papers. Clearly, of the two of these, the most important seeks to evaluate whether increased diversity improves people’s perceptions of quality of education. And they look at this diversity in proportion to the student body, and at surveys of broad areas from students, faculty, and administrators. You know, it is interesting, the diversity they have there is–I do not think it is really proved what they think it has. The regressions are cross-sectional, looking across schools at one point in time, and comparing a school’s characteristics with the survey data for these three groups.
Unfortunately, that is not quite the same thing as saying that you take a particular school, and you change its level of diversity in order to try to see how it affects people’s evaluation of the school. The bad effect that they find from increased diversity on polling data, for many reasons, does not imply the same result that would apply for any particular school.
Let me just give you a simple example from something like industrial organization. There are lots of regressions, kind of more mundane ones. We will try to look at firm size and, let us say, the cost of firm production. If you run a cross-sectional regression, some people say, well, larger firms need to have lower costs than small firms do. But it could be that each firm is kind of at its ideal level of production for its costs. They may have certain specialized assets that cause them to fit certain niches. And so, if I were to go and change the size of any of the large firms or change the size of any of the small firms, whether I make them smaller or larger, I may increase the costs that they have. It could be that at one point in time, they’re at their ideal point. And surely, that is a possibility that exists when you are looking at this type of data across schools with regard to diversity. You know, it could be that all these schools have the right level of diversity for producing people’s satisfaction with the schools and if you change it, you could be making things worse.
It is also very difficult to truly try to account for differences across places when you are using purely cross-sectional data. Ideally, you want to have what we call panel data, where you are going to be looking not only across places but by following each of those jurisdictions over time, so that you can go and see when a place changes in terms of its level of diversity; how things like the student satisfaction are changing relative to what they were before.
But you want more than that, in a sense, because there is one issue, and I suppose this kind of bothered me in the general discussion for a lot of these papers here; and that is you really want to have some type of exogenous change, something else that is kind of being imposed on these particular schools in terms of the level of diversity, rather than something that they’re changing, their selection over time, because what could be changing the level of diversity could also itself independently be causing changes in people’s perceptions about, you know, the quality of education that they are receiving.
And then, there is another general issue here, and that is just polling data per se. The way that this is set up is that different schools have high levels of diversity because, just generally, something else is causing both a high level of diversity as well as dissatisfaction. To put it another way, you want to ask why certain schools have higher levels of diversity than other ones. It is possible, and it could simply be by accident. It is also possible that they may have certain political views at the school that may make it less desirable for students to be there. For example, independent conflicts may arise as the school is pushing for more diversity. That is just one of many possibilities that could exist there.
Another problem is–and Rothman does recognize problems with survey data generally–if I go and give two people the same conditions, they may not come back to me in terms of the same answer, whether they are satisfied or not satisfied with the conditions. And I suppose one of the classic examples of that in labor economics is there are surveys for people on the job of whether or not they are satisfied, or what types of characteristics they are looking for in the jobs. And one of the things that people have shown for a long time is that if you take kind of the classic characteristics of male jobs–one of the things, for example, is riskiness of the job–it seems like when you ask men and women what characteristics they prefer for jobs, women seem to prefer the characteristics of men’s jobs much more than men do. If you have these 1-to-10 scales, women give much higher scores than men do.
But the problem is when people go back and say: how much more do you have to pay a woman to essentially take a risky job, than you have to pay a man to do a risky job? It turns out that you have to pay the woman much more. So, just because they go and they say, on a 1-to-10 scale, that they rank riskiness or find it more attractive relatively than men do, that does not mean that they’re answering the questions in exactly the same way that the men do with the same set of facts that they may be perceiving there. You could have the same type of phenomenon that exists across different schools.
I would like to turn to Dr. Sander’s research. The type of argument that he is making, or at least the results that he is finding, something that’s been around for a while–I mean, Milton Friedman and Tom Sole have argued these types of cascading effects and impacts that these can have for a while, but despite that, I think it is still extremely useful for someone to actually go out and try to measure these things, and try to do it creatively with the data that is available.
Again, this is difficult data to deal with, and it is difficult for many of the same reasons that I was bringing up with regard to the polling data because ideally, you want to have panel data. You want to have panel data where you have some type of exogenous change that is occurring, that is not a choice made by the individual school itself. And you want to have changes that occur over time.
There have been a lot of improvements in the data, and you can just see it across the different papers as they are being written. Rick was mentioning earlier, I guess, and maybe he will talk about this in the response time, I guess he has some new work I have not seen on Proposition 209. Maybe that can answer some of the questions because, ideally, 209 could provide an interesting test here. You have something else that is being imposed on the schools in terms of whether or not they are able to or would use affirmative action. Now, he can argue that they get around it. The question is how quickly they get around it. I would argue, and it could be that for a year or two, you do see a big change in how they admit the students, and then after awhile, they get around it and you go back to where we were before.
But in some sense, it is even more ideal that they get around it, at least from a science perspective, and that is because then you kind of have two tests. You have got the change. You can go and see when they change. Who gets in first; do I see a change in the grades; do I see a change in the rate at which these students graduate; do I see a change in their bar pass rates in one direction. And then, does it go back in the other direction once they start going around it? It is nice to have those types of additional tests.
The only other thing that I can think of that might be analogous is when schools have to deal with the bar association. At least, I have heard stories about the bar association. Without naming names of particular schools, the bar association will come down on a particular school and say you are not admitting enough minorities. And so, the school will go and change how it admits people. Now, if you have schools that are a particularly important share of the law students that are produced in a particular state, you could look at the overall bar passage rates to see the spillover effect on other schools, if students do not get into one particular school and now they are, to deal with the cascade and the overall impact issues that Rick–that is kind of motivating his discussion there in terms of what has happened to the total number of people who are going to become lawyers or whatever.
But you can still, even without it being a major share, look at some of the more direct questions about what happens to their grades in school and what happens to the rate at which they graduate. So those are the types of things I hope Rick is going to go into, hopefully with this 209 data and it just generally could be done.
Rick recognizes a lot of these weaknesses with the data and tries to deal with them. In general, I think the most important thing that he tries to do, and you can see it in many of the discussions here, try to deal with all broader selection problems. They are inherent in this data. Anyway, I read six papers over the last week, two of them by Sander and four of them by his critics. And just as an aside, the amount of name-calling was pretty amazing to me.
PROFESSOR SANDER: Not on my side.
DR. LOTT: No, do not worry. It was much more obviously, by a huge ratio, from the other side. It brings in considerations about, how do you respond to that? Do you just try to respond as an academic, just keeping it on the level? But I am not sure. You would have to ask some political type how best to respond to these things. But Rick, as he says, has pretty much kept in line.
Also, I just cannot help but mention this. It makes me wonder what the gain is from having law reviews. I am pretty cynical of the refereeing process in general, but I think it would provide at least some kind of controls on these things. I cannot even imagine what it would be like having second- and third-year economics students edit economics journals, let alone second- and third-year law students edit economics papers for law reviews. And also, I think the papers would be a lot shorter than what I have had to read over the period of time, so that would be a minor benefit.
Now the broad data, as Rick has pointed out and Doug pointed out, is pretty shocking. I mean, I will just give you one thing that I am going to concentrate on in terms of this discussion. If you look at the top-tier law schools, 52 percent of blacks in those schools have average GPAs that are in the lowest ten percent of their class. Only 5.6 percent of whites fall in the bottom ten percent of their grades. And even Rick’s critics acknowledge that these raw numbers are just amazing. The responses, though, avoid the most basic problems. Ayres and Brooks claim that Sander is wrong in the important sense that if you were to move students from the top-tier schools to the lower-tier schools, you would still have them falling in this very bottom percentage of GPAs. I will go more into that in a minute.
But the question of why they get low grades to begin with, let alone why they would still get these low grades even if we moved them to lower-level schools, is never explained, really. I mean, I could guess, knowing Ian Ayres pretty well, that he probably believes there is some type of systematic discrimination that is occurring in law schools which is preventing these people from going and getting higher grades. I am not really sure how that would actually work in practice with blind grading and law schools.
And there is one other general point I want to get into before I get into the numbers here, and that is rational expectations. It is kind of an economist term. But essentially, you believe the people should guess the problems in advance. One thing that makes me somewhat skeptical, at least coming into the results that Rick gets here, is why would people go to schools that they are going to end up doing so poorly at, not graduating, and then not being able to become lawyers to begin with? You know, it could be that people are just fooled into these types of things, but you would expect there would be something learned over time, and maybe after the amount of publicity that Rick’s paper has gotten, maybe we should see some changes in terms of people going through that. I doubt it, but it would be interesting to see if that was the case. Okay, so you still need to explain–let me back up for a second. One of the most interesting sets of data that exists here is the first- and second-choice decisions that students face–there are multiple tables now that look at this type of issue.
But the bottom-line question I think needs to be addressed in these studies is why students pick their second choice, versus their first choice. It could simply be that if I could go into a high school and a low school, the reason that I picked the low school, the reason was I decided at the last moment I don’t want to be challenged. Well, if that’s the case, that provides certain information on systematic biases that might exist there in terms of their grades.
I mean, it could be that the students that go to the second choice are students that would get as high a grade as if I were to forcibly move somebody who’s going to their high first-rate school, and move them to a lower-ranked school. So, that might under-predict the benefit from getting rid of affirmative action. It could be income that decides to do that. If you have the income to do it, you also have the issue of someone with a higher income might be able to concentrate more on–do we have 15 minutes?
PROFESSOR KMIEC: Yes.
DR. LOTT: So, I did want to get to a point here. You know, it is possible that the second-choice students just kind of–they read Rick’s studies, and they know that they would run into problems going to the other schools, and those would be kind of the ideal people that we’d have moving into it. But it’s still a question of trying to control and account for it because there are certain systematic biases.
The Ayres and Brooks paper attempts to redo some of the basic regressions that Rick has, including a variable for whether or not there is a second choice here. They claim to find that the second choice is not, you know, a student going there does not improve their grades relative to what they would have otherwise. I think part of it could be explained by not accounting for these biases that we are talking about.
But another thing has to do with the sample size that they have here. They are not comparing all first choice schools versus all the second choice. They have a very small subset of first-choice schools they are looking at. They are looking at first-choice schools when the student also reveals information on the second-choice schools, and that essentially cuts the sample size by about two-thirds that they have to make the comparison. And it makes it a lot less likely that they’re going to get a statistically significant result.
Even more than that, while Rick does not run regressions on this, he does provide kind of a table that shows the means on the stuff there. One thing that’s interesting is that the ones that are being excluded tend to produce an even greater difference between first- and second-choice schools in terms of things like bar passage rates. And so including those, you would imagine, and a regression would, not only because you have a larger sample size, but because the means are different there, be even more likely to produce statistically significant results.
So, there are other things here. I could go through the different regressions. I am not going to go through those. But let me just say a couple things to think about. We have seen a lot of institutional changes in law schools over time. You do not fail students anymore, really, in law schools. Why do you not fail them? I think part of it is this fact we are talking about in terms of GPAs. If you fail students these days, you would overwhelmingly be failing minority students. You also have the content of classes. I worry, from reading this type of stuff, whether maybe the emphasis on policy discussions very heavily in the top-tier law schools might put minority students at a disadvantage, if they are having these other difficulties going through and learning the material. On the other hand, it might not be too bad for some kid who scores really highly on the LSAT, because he can go and pick this up on his own. I will not go through this, but it is just a bit of an irony that the professors whose classes I have sat in on, that primarily tend to go and do policy discussions are probably the ones that push the hardest for affirmative action, and there might be a little bit of a tension there just to think about.
PROFESSOR KMIEC: Well, now a man who needs no introduction, because he has had one, but we welcome Dr. Richard Lempert. And in case anything gets lost in the translation, I want to get right to the kind of hardball part of the analysis here. Dr. Lempert writes that Professor Sander’s research has serious statistical, analytical, and data flaws, and that he provides no reliable support for his hypothesis or conclusions that he draws from it.
With that as an introduction, Dr. Lempert.
DR. LEMPERT: As I set up my computer, I’d like to ask two questions. I know that this is a sophomoric test for a show of hands, but I am curious. The first is, in terms of this discussion and debate, how many of you would have different views on affirmative action depending on whether Rick Sander’s arguments are scientifically accurate or, in fact, the real world is almost the opposite of what he says? Whose opinions would be changed?
DR. LEMPERT: Okay, that is good and interesting to see. And of course, it is not necessary for anybody’s opinions to be changed. If you view the Constitution a certain way, it does not depend on empirical evidence. But in a certain sense, there is no point in these discussions unless some people are open to science, as I am.
The second question, and this will not apply to the older faces I see out there, but I am also curious as to how many of you people who were in schools at a time there were black professors in law schools, encountered one or more black professors who you thought was really a good teacher or you felt had a real impact on your education?
DR. LEMPERT: Okay, I am going to come back to that in a bit. Is this on now?
Have you ever had this problem before? Do not even–here is a point we can begin with, which I think is maybe one of the few points that Rick and I are going to agree on. Rick has had this experience many times. He has gone before audiences who are presumptively unsympathetic to him. So now, we find the situation reversed; we will agree this is fair. And here are the questions I asked, at the same time.
[Brief pause to adjust PowerPoint slide]
PROFESSOR KMIEC: This is pretty much what we do in law school when we are not doing policy.
DR. LEMPERT: Sorry about all that.
First, I just want to give you some numbers about how many people are affected. We tend to talk in terms of percentages. You hear 80 percent, 50 percent. What are the numbers? I should emphasize, this is 1991 data. Unfortunately the best we have is for the cohort entering in 1991. But the situation may have changed since then, and in various ways. Only with the 1991 data can we ask how many people were affected, and who?
If we consider affirmative action admits to be people who get into law schools when their index scores–that is, their undergraduate grades and LSAT scores–would have predicted that they would have not been admitted, then in 1991 there were 2,740 black affirmative action admits. Who was hurt by this, we can ask. Well, if we consider those hurt, those people who were predicted to have been admitted who had not been admitted because their places were taken by non-index admits, there were 4,392 white students who one would have expected to have been admitted on the hard credentials, who were not.
However, let us look at the number of whites whom we may think of as affirmative action admits: they are 2-1/2 times the number of blacks; thus there were more than 6,000 white students who were admitted to law schools though their credentials would have predicted that they would have been rejected. You could still have admitted about 2,000 of these affirmative action whites along with all the admitted affirmative action minorities and also have admitted all of the whites who were predicted to be admitted and were not. Yet when whites who feel they should have been admitted to law school and were not feel they have been victimized by affirmative action, they focus on minorities admitted to law school with weaker credentials than theirs and not on the larger number of whites who were admitted although they did not have the index credentials that the schools that admitted them ordinarily demanded. That is just to give you an idea of the numbers.
My critique of Rick is simple, and I cannot say it any more kindly; I hope this does not count as name-calling. I do not think his work is very good science. It was called in one article “cold fusion.” I will not go that far, but I have serious problems with it. I think there are statistical and methodological errors in it. I do not think there is any support in these data for the mismatch hypothesis, and I think there is less than no support for his claims that attorney production would increase if we had no affirmative action.
At this point, some of you may want to scream. You all as lawyers know about, “he-said she-said” type cases, and I think you also know that the only thing worse is, “this expert said, that expert said.” How do we make any progress in resolving different social science claims, particularly in a brief commentary? In a certain sense, we cannot make truly serious progress, but you have on the web and the law reviews, articles that Rick has written, and that the critics have written. Also, we will publish–my co-authors–a web response to his reply in the Stanford Law Review. If you really are interested and concerned, check out the detailed discussions and you can decide whose arguments hold up, or give the conflicting pieces to people who know statistics and they can decide.
However, the situation is not as bleak as it may be because Sander admits to at least a number of the flaws in his work. He does this in his reply to our critique in the Stanford Law Review. He admits that our criticism of his estimate that the number of black lawyers would rise without affirmative action is apt–he calls this our strongest point. We certainly can agree, and this is one other point that Rick and I can agree on, that there would be huge fluctuations in what happens depending upon the number of whites who are applying to law school in a given year. He admits at another point that certain results depend on a questionable coding decision, where he lumps people as white when he does not know their race. And he also admits that his models are improperly specified because they do not take account of selection bias. Now, only the last, he sees as mattering. There are also what I call omissions by silence in his reply to critics. There are very serious critiques that I and my co-authors make in our article, which would expect to be answered if they were wrong, but they are not answered.
If you don’t have a social science background, here comes my one lesson in statistics. If you take away nothing from this discussion except this particular lesson, and none of the details of our discussion, you still would have benefited. Rick writes about the T statistic, and I will just quote the part that is in italics, “T statistics above 2.0 are usually taken to signify that the independent variable is genuinely helpful in predicting the dependent variable.” That statement, as we say bluntly in our article, making a point he never responds to, is wrong. It is wrong in “Statistics 101” sense. As he himself recognizes, T statistics are very sensitive to numbers and to the statistical qualities of the distribution. One may have a highly significant relationship, particularly in a data set as large as the bar passage study, which has data on more than 20,000 students, and it may have no practical importance whatsoever. One of the problems with the analysis is that it treats everything that is statistically significant as if it was practically significant. That is simply not true.
Listening to his presentation, I want to give you one more lesson, one that John alluded to, and this is that correlation is not causality. The fact that people with low grades in law school do poorly on the bar exam does not mean that the lower grades have caused poor bar exam performance. One cannot ascertain that from the data.
There is no evidence of a mismatch. This slide provides a reanalysis of the data. Rick’s hypothesis is very simple. If you hold index scores constant, because of the mismatch effect, people will do better the less challenging the law school they go to. Here we look at the data, and we hold index scores constant. We put an X for every prediction that is opposite the predicted direction, and P when it supports Rick’s hypothesis. You will see in this table, which talks about black graduation rates, and in this table, which talks about black bar passage rates, that the Xs overwhelmingly predominate. Contrary to the predictions of the mismatch hypothesis, if you have a given index score, you tend to do better in both graduation and bar passage, the more selective–not the less selective–the more selective the law school you go to. This is completely consistent, I should note, with Bowen and Bok’s study at the undergraduate level.
What is most important here, I think in these tables, is that the effect is strongest for students attending elite schools. John, who I think is not aware of all the data, suggested there was a huge problem in failures at elite schools. This is the one level of school where blacks tend not to have problems. They attend and they graduate and they get good jobs.
In my own research with some colleagues on Michigan Law school alumni, we found that over a 27-year period, about 95 percent or more of our affirmative action graduates–most of whom were black–graduated and passed the bar. Their incomes and job satisfaction were no different than were white alumni over that 27-year period, and they tended to do a bit more service. So the mismatch hypothesis in the bar passage data, as Rick originally analyzed them, are not supported.
He does make a selection bias argument, and in a sense he has to because the data when simply analyzed do not support his analysis. So it becomes a more complicated analysis. It is basically one that controlling for index scores, blacks do not do worse than average as school selectivity increases, but that is because blacks in more selective schools have disguised strengths that are not measured, and that if these blacks, say at a school like Michigan or Harvard, went to a school like, let us say, Iowa or DePaul, they would do even better than they do because of selection bias.
I have a couple of things to say about this–the first is this is completely inconsistent with the first point that Rick made in his presentation to you. The first point he made his discussion was that law schools, particularly the elite ones, select only mechanically on the basis of LSAT scores and undergraduate grade point averages. You cannot have that both ways. Either law schools select on features beyond the index credentials, in which case there may be something to the mismatch hypothesis, or there is no mismatch hypothesis–no selection bias, I should say–because selection bias depends on selection occurring on unmeasured variables. There is, I should note parenthetically, a kind of self-selection which Rick’s response doesn’t take notice of, which does open the door slightly to selection bias and raises some complications. But the key focus is on law schools. So, you can’t have it both ways; you can’t both claim that law schools, whatever they say, admit applicants almost entirely on the basis of index credentials plus race and argue that selection bias on unmeasured variables explains why, despite your mismatch hypothesis, black students at more elite schools seem to do better than black students at less elite schools in graduating and passing the bar.
There are other analytic problems. As John correctly alerted us to, there may be other things that explain outcomes. In fact, if you look at the differences between second-choice and first-choice people, you find that they differ dramatically in the degree to which their motives are driven by financial aid. A much higher proportion of second-choice people than those who went with their first choice said financial aid was very important in choosing what law school to attend. We know that schools tend to give their financial aid more generously to those they most want, so there is still selection bias in Rick’s analysis, and he does not correct for that.
What does occur if you abolish affirmative action is a devastating effect on black enrollments. If you look at current data or relatively recent data, 2003 data, if you just select on the basis of LSAT and undergraduate grade points, blacks would constitute 75 percent of the top ten percent of the class in the top ten schools, and only one percent of the top 11 to 25 percent schools; you can see those numbers, 1.65 percent when we get to the 26 to 50-percent schools. So there would be a devastating effect. Even if we double these on the theory that law schools would look at other factors, you have a miniscule representation of blacks. This has a devastating effect on the black professorate. Sander’s own data show this, yet 25 percent of 604 black law professors teaching in ABA-approved schools went to Harvard and Yale, and the top ten law schools produced 40 percent of black law professors.
I cannot comment on Rothman at all. I do not think his science is nearly as flawed. I just think it does not say terribly much. We can ask questions about that.
I want to just end with one note. Why do I care? I have spent I cannot tell you how many hours. My wife can tell you how many late evenings we have not spent together because of my reworking of these data. I do care. I care passionately about this. Partly, I care about good science, but beyond that I do care about affirmative action.
One reason is I care deeply about equality. It would be lovely to think that racism was behind us, but recent research is even more striking about what is going on; for example, research reveals that there is a cost to a black name. One researcher sent out job application forms answering real ads. The vitas are identical; many more and many quicker callbacks went to those with a white name than to those with a black name. In one study, it turned out the whites with a criminal record were more likely to get job interviews than blacks with the same vita without criminal records. The implicit association test shows mental processes that disadvantage blacks without people even realizing this. There is also a stereotype threat which seems to lead to poorer test performance, regardless of what has been learned, by stigmatized minorities.
The second thing is I care deeply about integration in the schools that I teach in and the society I live in. This partly is a function of me being raised in the 1960s and looking at the pictures of the South, but there is also a note of pragmatism. There was just an interesting column in the paper today about the French example, what is going on and how the lack of integration, the lack of caring for others, has led to sometimes violent social conflict. I want us all to live together in equality and in peace.
PROFESSOR KMIEC: We have very little time left, but I want to give Dr. Sander in particular, since his presentation is at the core of this, the opportunity for rebuttal. And I would ask those who have questions to be at the microphone, as you already are. And I would ask Dr. Rothman to lead off the question period.
PROFESSOR SANDER: I think Dr. Rothman is going to make a few comments while I change the computer.
PROFESSOR KMIEC: Okay, we are changing computers.
DR. ROTHMAN: I have been beautifully ignored here, so I do not have many comments to make. I would only say that the worst criticism one can get is that you did not do the study I would have done. The fact is we did not do the study that someone else might have done. We did not have the money, and we did not have the time. We may do some more of it later. And we agree that that raises some serious problems.
The point, however, is that we had a modest goal: to raise some questions about articles on affirmative action and diversity, which are widely published in our field, which dominate the field. We had some data which raised some questions about those. And all we do is raise questions, and it may not be much, but it is a beginning.
My colleague who is a statistician happens to be sitting in the audience, and I am going to ask him if he has anything he wants to say at this point. Bob.
AUDIENCE PARTICIPANT: Bob Lichter. I have worked with Stan in the past. I vetted some of the statistics on this article. Of course, this is what Dr. Lott raises as issues of research and design. And of course, it is always better to do longitudinal studies. But the real world does not always permit ideal solutions.
I think the strength of the Rothman paper is that it takes a body of literature that is overwhelmingly based on survey data that argues because people say diversity is good, that means it actually is good. In a sense, if we get rid of this problem of what we call social desirability response set by saying, well, one would think that this is true that if the schools that have higher minority representation, people would say nicer things about the school, they would be happier, so on and so forth. You would think this would also be true, but it is not the case.
Now it could be, of course, that there is some additional variable out there that is influencing both diversity levels and satisfaction. Dr. Rothman did not have time to mix a lot of the control variables. It could be that all over America, schools are at their perfect level of satisfaction, that if Villanova has more minorities than Temple, and dissatisfaction is also greater, it just happens that way it is possible. And we have to admit these possibilities. But I think a lot of people are going to advance by introducing real-world variables.
And if you look at this not as a case of trying to prove negative effects but basically trying to support cases that are made purely on survey data and try to support it using more strenuous measures and finding that it is not supported, I think that is a genuine advance in the argument.
PROFESSOR KMIEC: Thank you, Dr. Lichter. And now we will hear from Dr. Sander.
PROFESSOR SANDER: Yes, I will just add, I agree completely with what Dr. Lichtman just said. I think the key contribution of the survey research that Dr. Rothman has done is to point out the methodological weaknesses of what has been so heavily relied upon and try to set us on a path towards asking the questions in a more meaningful way, that are actually capable of having multiple answers. We are running late, which is a shame because, as I said, I think the key usefulness of these sessions is real engagement. But I am going to try to engage as many of the points as I can in just a few minutes.
This is one of the key regression analyses in systemic analysis. And Professor Lempert argues that I am misstating what the statistical significance is, and that becomes a serious problem because it leads me to think that relationships are meaningful when they are not meaningful. And he has argued specifically that equations like this do not really tell us much about actual performance.
The point of this equation is to illustrate, do a regression analysis, to try to show that law school GPA is tremendously more important than school eliteness in how you do on the bar. And it is something that is hard for non-experts to interpret. So it is particularly vulnerable to an argument that sort of says, well, that does not really prove anything.
So, let us see if this proves something more. This is a chart showing the proportion of students who graduate and pass the bar. It is essentially a crosstab, trying to illustrate the data behind the regression I just showed you. Students in the left-hand column are students who have GPAs in the bottom 10th of their class, at schools ranging from the most elite, that’s tier one, to the least elite, that’s tier six. In the right-hand column, you have GPAs of students in the top half of their class. You can see that there’s an astonishing difference. GPA matters enormously.
This regression equation is not one that is showing a tiny marginal effect. It is showing a fundamental effect. It is showing most of what drives these outcomes. And you look at this data in 20 different ways and you will come to that same result. What is disappointing about Professor Lempert’s analysis, so many of his criticisms and some of the other criticisms that have been published, is sort of trying to talk past the numbers, not really grappling with what the data really shows.
Another example of this is the first-choice, second-choice analysis. We have talked about it extensively, but you really have not seen what it shows. Out of the 2,000 students in the LSAT study, about ten percent of them chose to go to less elite schools. And on average, what that did is it eliminated about a third of the mismatch between them and their classmates. And this chart shows outcomes. The first column shows the rates at which whites graduate and pass the bar. The second column shows the rates at which blacks who go to their first choice schools graduate and pass the bar. You can see these enormous disparities–92 percent bar passage for whites; 59 percent for blacks. The last column shows the success rates for blacks who went to their second-choice school; in other words, those who reduced somewhat the size of the mismatch that they were under. And the third column shows the predictions that fall from a simple linear application of the mismatch theory, my favorite.
I think you can see that these are fairly powerful results. There are dramatic differences between how blacks do when they reduce their mismatch sum and when they are maximizing in their league. This data is overwhelming. It speaks for itself. If you analyze statistically and try to do a regression analysis–I have not published the regressions, but I have run all the regressions that Ayres and Brooks do–you find virtually every relationship that is predicted to be significant theoretically is in fact significant.
Ayres and Brooks did an initial analysis, where they found the difference in first-year grades between first- and second-choice students was not significant, and then they wrote up an analysis saying, well, this data does not work. I went through their data over a period of weeks. We found the errors in their analysis. They redid it, and they came to the exact same numbers that I did. I do not have any disagreement with Richard Brooks on what the numbers are. But they rewrote the text in such an elliptical way, not really wanting to change their conclusions, such that John Lott could carefully read it and conclude that they are saying that there is not a significant difference in first-choice rates. They do say there is a significant difference, and in fact, the Stanford editors forced Richard Lempert to change his article to acknowledge that there was a significant difference.
DR. LEMPERT: That is just completely false.
PROFESSOR SANDER: Now, it is not. Your text was changed to remove your claim that there was no significant difference between first- and second-choice grades. Go back to read your article. That change was made about a week before publication. There is no debate between Ayres, Brooks, and I on the fact that most of these relationships are statistically significant.
They find, for example–one area where they do find a difference is they say third-year grades are not significantly different between the first- and second-choice blacks. And that finding–they find, by excluding from their analysis, all students who drop out before they get to the third year. In other words, they eliminate most of the students who have bad grades because they have not graduated. So it totally skews their analysis. They get meaningless results. When he put the students back in you would devastate a systemically significant result.
So, I think that the debate has been ill served by a failure to not engage the data, and what we need are more social scientists who do not have a predetermined agenda looking at the state and carefully evaluating results. And I very much appreciate Dr. Lott doing that.
PROFESSOR KMIEC: There are a number of people waiting with questions. I would ask you to make them very targeted and very brief. We are going to take a few minutes. Please direct your question to a member of the panel.
AUDIENCE PARTICIPANT: This question is for Professor Sander. There are a few factors that I have noticed have not been mentioned so far that may be relevant.
First of all, I am wondering what the impact on the results that you found on racial preferences at the undergraduate level in the employment field; and also, what the, I guess–and also the reverse cascade effect for white students who wind up at schools that they would otherwise have gotten into a better school.
PROFESSOR SANDER: Okay. I followed the second part of your question is about, how does this affect whites. But say again the first part.
AUDIENCE PARTICIPANT: The first part is about the impact —
PROFESSOR KMIEC:: I think we actually only have time for the second part, so go ahead and answer the second part.
PROFESSOR SANDER: Well, the effect–I mean, it is certainly a clear policy implication from this research that whites overall should benefit from preferences, as they are used currently, rather than being hurt. In other words, if they actually do achieve an admission, whites are insulated to some extent from being at the bottom of the class and having the negative effects that would result from that.
It is a little bit hard to tell because Rick Lempert again sort of mentioned that many whites fall outside the normal–the middle range of index scores of individual schools. But the degree to which they fall out of those ranges tends to be much narrower than the area where blacks are falling. So we would expect if the system was a white-only system, the mismatch effect would probably shrink because there would be a smaller gap between the credentials of the bottom of the class and the credentials of the top of the class. I think it is most clearly true in the job market scenario that whites probably benefit from having higher average GPAs, sort of as a result of involuntarily being matched with less elite schools.
PROFESSOR KMIEC: We have one final question.
PROFESSOR SANDER: I just want to point out that reading from the text, Rick, your analysis says, “Ayres and Brooks find that students who attend their second-choice school had either received better final law school grades”–you were forced to change that from saying initially saying first-year law school grades, if you go back and look at the text.
DR. LEMPERT: We were not forced to change anything, at least —
AUDIENCE PARTICIPANT: Much of what I have heard from most of the panelists today kind of seems to ignore the vested interest that the establishment, big firms, the government has in affirmative action. And so–just two targeted questions with respect to that–Professor Sander, with reference to Howard University, I know it has a great reputation as a law school, but was the same 170-point gap from among the law students at Howard, as you would find with those same top minorities at the elite law schools? Because, the fact that the large law firms go there to interview means nothing, because every firm has policies where they need to bring in, or they are looking to bring in, a certain percentage of minorities, so that would obviously be a great place to go.
PROFESSOR SANDER: Yes.
AUDIENCE PARTICIPANT: And second, Professor Lempert, how can we accurately measure things like, “they graduated” or “they got good jobs,” when we know once again that the government, big firms, are going to go out there and recruit minorities because of their own policies and what they consider valid social policy? Thank you.
PROFESSOR KMIEC: Thank you for that question, and we will ask a brief answer from each respondent.
PROFESSOR SANDER: Yes, very briefly, Howard has smaller preferences because they have sort of a large built-in pool. They also have a much larger percentage of blacks, so that also reduces the mismatch effect on its own. So the result is that you get a pretty substantial number of Howard blacks who were the top 20 percent of their class, and those students do extremely well. And you will find that generally the historically black schools, they have higher bar passage rates and better outcomes than one would expect from just looking at their incoming credentials. It suggests that there is something positive going on.
PROFESSOR KMIEC: And Dr. Lempert.
DR. LEMPERT: With respect to that question, do we just see so much affirmative action of in the legal practice world that all these data are meaningless? I think not. At least in the Michigan data, when we looked at it many of our graduates who were on second and third jobs, as white or black–there is huge amounts of changing, as you now know, unlike the earlier world. And I at least find it very hard to believe that when you’re talking about average salaries on the order of $200,000, that firms are intentionally hiring senior people at those salaries, and the Harvard average, like $300,000 a year–just to have a black face in there, I think these people have to be valuable.
When you look at accomplishments, when you look at leadership roles, you find that the minorities–we looked at blacks in regressions, but we were looking at Hispanics and blacks for much of this. You know, they just have very high achievement rates. If you are an alumnus and as a faculty member, you see people come back, you are aware of the high achievement of your black and white students in elite law schools. So I just do not–you know, we have a discussion of exactly this issue because your reaction is a common one. In our article, we try to deal with the reasons why we do not think that has a major impact.
PROFESSOR KMIEC: Dr. Rothman, Dr. Sander, Drs. Lott and Lempert, ladies and gentlemen, thank you for your patience. Have a good one.