CHAPTER SIX

THE BREADTH OF MISMATCH

SANDER’S NARRATIVE ON LAW SCHOOLS and the evidence for “science” and “aspirations” mismatch from Chapter Three offer some particularly powerful examples of how mismatch can derail lives. These examples collectively affect many thousands of minority students each year, and even if we stopped here, we think we would have made a case for reform. But one might argue that these are special cases—that in law school and science curricula, students must race along a narrow path, and if they fall off, they are in trouble. This is a fair point; we examined law school and science curricula closely largely because in those cases it is easier to observe the consequences of mismatch directly.

In this chapter we consider higher education more broadly. What is the evidence that mismatch affects the general learning environment in college for those who receive racial preferences? Do black and Hispanic students end up flourishing in college and graduating at high rates despite whatever mismatch problems may exist? Are the benefits of getting a preference into a more elite school in the end worth the costs?

These are big questions—and honestly contested ones. One way of answering them is to look at an entire system of higher education that used large preferences for many years and then suddenly reduced them sharply. The two hundred thousand–student University of California did just that, and in Part III we examine closely what happened when racial preferences became illegal. In this chapter we take a different approach, using a range of evidence to understand, as best we can, how mismatch plays out across the broader college experience—how (and whether) it affects learning, graduation rates, and even long-term earnings. And importantly, we will introduce some of the people who spoke with us and our research associates about their experiences with mismatch. Though many questions we ask still await better data and further research, we think the chapter makes the general nature of mismatch far more tangible.

LEARNING

Mismatch is fundamentally about learning. Students who have much lower academic preparation than their classmates will not only learn less than those around them, but less than they would have learned in an environment where the academic index gap was smaller or did not exist.

It is hard to test for mismatch directly because colleges and universities rarely try to measure the learning of their students in ways that can be compared across classrooms or institutions. What we would most like is a controlled experiment, in which students are randomly assigned to classes where they are more or less mismatched and their rate of learning is observed. This sort of “randomized trial” is rare in education, and we know of only one attempt to study mismatch this way. Several years ago the World Bank helped to fund an experiment in Kenya in which thousands of elementary school children were randomly assigned to two types of classes: one that grouped students with a broad range of academic skills together, and one that separated them into high-preparation and low-preparation halves. Three distinguished American economists participated in the experiment’s design, observed its outcomes, and concluded that the “tracked” students (in both the higher and lower classes) learned more—suggesting that teachers taught more effectively when they calibrated their teaching to a narrower range of student preparation. Here, at least, reducing mismatch seemed to benefit students across the board.

We can only look in envy at research of this type. In American higher education the closest thing we have to a broad measure of student learning is a relatively young initiative known as the Collegiate Learning Assessment (CLA). Started by the Council for Aid to Education in 2001 as an effort to develop simple, generalized tests of college-level skills that would not rely on memorized facts or multiple-choice questions, CLA has grown rapidly and has to date been used by some five hundred colleges. CLA tests ask students to write essays about complex, somewhat real-world problems based on packets of background material they are given at the test. They integrate this information into a solution and describe it in their own words; the results are evaluated for verbal ability as well as analytic content. Nearly a quarter million students have taken these tests to date, and for researchers, the most useful information comes from CLA’s “longitudinal” component, in which the same student at a college takes the test shortly after arriving as a freshman, again at the end of sophomore year, and a third time as a senior. The longitudinal testing, if done systematically to a large random sample of students, would make it possible to measure both their progress and the general rate at which cognitive skills are acquired at their college. But the CLA is controversial because it is funded by the colleges that participate in it and lacks many of the hallmarks of careful research: Colleges sometimes choose which students participate in the tests, and an unusually high number of students who take the initial tests never take the follow-ups.

In a 2011 book that attracted wide attention, education scholars Richard Arum and Josipa Roksa used the CLA to assess the general effectiveness of college education in America. As the title of their book—Academically Adrift—implies, they concluded that student learning on college campuses was disappointingly modest. They did not directly test mismatch ideas, but their findings about black learning were particularly bleak.

In their national sample Arum and Roksa found that entering black freshmen had average CLA scores that were at about the 16th percentile of white freshmen—not good of course, but in line with the results of other tests like the SAT (see Chapter Two). But when students were retested at the end of their sophomore year, whites had moderately improved their performance as a group; the white sophomores were at about the 58th percentile of the white freshmen. Blacks, meanwhile, had barely budged; black sophomores were at about the 17th or 18th percentile of white freshmen. Blacks were not only lagging further behind; they were barely registering any improvement in skills during the first two years of college.

Note that this is in sharp contrast to high school. Although black scores on national achievement tests are far below white scores at both eighth- and twelfth-grade levels, both groups make steady progress during high school, and the gap does not significantly widen. The CLA data on blacks also contrast with the data on Hispanics; Arum and Roksa find that Hispanics improve their learning in college at roughly the same rate as whites. Because Hispanics receive racial preferences less frequently than blacks do and are much less likely to have academic indices far below those of their college classmates, each of these pieces of evidence is consistent with the idea that large college preferences for blacks, and the accompanying cascade that spreads the effects of preferences, are undermining their learning. Given the weaknesses in the current CLA (which could be overcome by increased funding) and the lack of any direct test for mismatch, we should not over-read these findings. But the findings to date suggest that there is indeed a crisis of black learning in college.

GRADES

As we noted in Chapter Two, students who receive large preferences tend to get low grades. Every academic study on the subject confirms this, though much of the available data captures this by using “race” as a proxy for preferences rather than measuring preferences directly. So, for example, a number of studies have shown that blacks at elite colleges have GPAs that place them somewhere between the 15th and 20th percentile of white students; only 5 percent of blacks and less than one-tenth of Hispanics end up in the top fifth of the class, and blacks are four or five times as likely as whites to end up in the bottom tenth. But of course, such statistics somewhat confuse the issue, because some blacks at elite colleges have received small preferences or none at all, whereas some white and Asian legacies and athletes undoubtedly have received preferences. An unusually careful study at Duke University found that when one looked simply at admission preferences and not at race, the race effect disappeared; that is, as Sander found for law school, blacks get low grades in college not because they are black but because they disproportionately receive large preferences.

The Duke study also tracked students through college and found that when one controlled for student majors and the different levels of grade inflation that exist in different parts of colleges, the students who received preferences did not catch up with their classmates. Indeed, as we discussed in Chapter Two, as the CLA study implies and as many other studies have found, as college progresses, students receiving preferences often fall further behind academically. Princeton scholar Thomas Espenshade and his coauthor Alexandria Walton Radford find that “Black students, for example, lag behind comparable whites by about nine percentage points after one year, but the gap grows by an additional eight percentage points by graduation time, resulting in a total deficit of more than 17 points relative to comparable whites.” Something similar happens on a smaller scale to Hispanics. There are a number of possible reasons for this, but the existence of mismatch—and the social effects of mismatch we will examine shortly—are certainly consistent with it.

STEREOTYPES AND STIGMA

Over the past twenty years an outpouring of literature has come forth on the claimed benefits of preferentially engineered campus diversity, driven in large part by Supreme Court decisions (explored in detail in Chapter Thirteen) that permitted racial preferences as a means to foster the virtue of “educational diversity.” The general questions such research raises are whether racial diversity encourages more interracial contact, a greater ability to work with people of different backgrounds, more understanding of differing perspectives, and higher racial tolerance.

Much of this research has been heavily and justly criticized as little more than propaganda. To see some of its intrinsic difficulties (and sometimes enormous silliness), consider a well-known study by Gary Orfield and Dean Whitla that surveyed law students at Harvard and the University of Michigan. Students were asked to report how many close friends they had (implicitly at the law school) who were of another race. Nearly all (92 percent) of the white students reported they had “three or more” such friends; only 37 percent of the black students and 29 percent of the Hispanic students reported “three or more” interracial friendships. Because both law schools had predominantly white student bodies, these answers created a mathematical impossibility: Whites appeared to be claiming at least six “close interracial” friendships for every one such friendship claimed by a minority student. Yet the authors soberly concluded from this finding (and other similar ones) that “white students appear to have a particularly enriching experience” from a racially diverse law school. A similarly profound finding from the study was that over 70 percent of the law students thought that “having students of different races and ethnicities” at the law school was a “clearly positive” thing.

Surveys of this type mostly tell us that students—especially white students—are eager to display their racial cosmopolitanism and their understanding of the official diversity line that university leaders promote at nearly all school events. Studies of student opinion that fail to control for the automatic desire of students to provide the “correct” response are almost worthless. A good deal of diversity research has these weaknesses and amounts to a sort of happy talk about interracial utopias on college campuses.

Thus, when white students are asked how many of their five closest friends on campus are of another race, they give impressively high numbers in response. But when, without any racial prompts, white students are asked to write down the names of their five closest friends, and the race of these friends is later determined, the number of nonwhites on the list is much lower. Or, to put the issue another way, when we compare the interracial friendships of white college students with those of the average white American, the patterns are not notably different. Carefully considered methodologies, in other words, tend to cast great doubt on the findings of much of the pro-campus-diversity research.

We think it goes almost without saying that, especially in a diverse society like the United States, students can benefit from an environment that is diverse in many ways—racially, socially, economically, religiously, and politically. But “diversity” research almost always fails to consider a critical question: What happens when students who bring diversity to a college or graduate program are both easily identified (by their skin color) and struggling to survive academically at very disproportionate rates?

It is not hard to imagine or to find some evidence for a whole interrelated series of consequences (which our discussions with experienced, sympathetic administrators bear out). Minority students—blacks in particular—may struggle heroically during the first semester but will very often be dismayed by their grades. As they start to see the gulf between their own performance and that of most of their fellow students, dismay can become despair.

“Their egos can take a big hit,” explains a black administrator, whom we will call Douglas, at a university in the Northeast. “All of a sudden they aren’t the top student anymore. And they are confronted with the notion of, ‘Hey, I’m not as smart as I thought I was.’ Some of them get here and right away start to think that they just can’t cut it. They are afraid to talk in class. They say that their classmates just seem so much better prepared. And so they very well may end up leaving.”

“I’ve been watching this happen for twenty years,” says Gary Hull, director of the Program on Values and Ethics in the Marketplace at Duke’s undergraduate school. “The paralyzing premise is ‘I must, but I can’t.’ That is, the students experiencing these mismatch obstacles arrive at Duke thinking that ‘I must’ succeed academically because ‘they have been told by everyone they trust’ that they are well qualified to succeed. The ‘I can’t’ realization comes after they arrive on campus and learn that they simply don’t have the preparation or academic ability to be successful students at Duke.”

So, to paraphrase Hull, they will not be able to live up to the high expectations of their families and friends. No matter how hard they work, they will get low grades and often feel lost during classroom discussions if they take tough courses. The reason is not that they are bad students; it is that the vast majority of their classmates at the highly competitive school have a huge head start in terms of high school education, academic ability, or both. They would be fine at a lower-tier school. But at Duke, “I’ve seen many of them withdraw into shells,” says Hull. “This causes some of those students to take on an unearned guilt, which then triggers a host of psychological maladies.” Researchers have come to similar conclusions about racial preference recipients developing negative perceptions of their own academic competence, which in turn harms performance.

Meanwhile, university administrators, who have typically misled (at least by implication) many black and Hispanic students during the recruitment process about their prospects for academic success, often take the position that the best way to build their academic self-confidence after they arrive on campus is to continue misleading them. The university thus never tells individual students who run into academic difficulty either that they were admitted via large racial preferences or that this put them at a great competitive disadvantage academically. So the students receiving large preferences may blame themselves and conclude that they are failures.

Beyond feeling self-doubt, sooner or later mismatched students are likely to notice that other students of color are having difficulty too. Their personal academic struggle takes on a collective, racial cast. They may not associate the problem with preferences, as they have been repeatedly assured that everyone admitted to the school is fully qualified and has only to seize the opportunities before them; they may turn to more sinister explanations instead, such as a hostile learning environment or even discrimination by their professors. Whatever explanation they settle on, the natural reaction is to withdraw into a racial enclave within the campus, seeking to foster a separate community in which the minority student can, in some sense, feel more confident and consider herself a better “fit.” Many universities encourage this by creating black dormitories and assigning entering students to them.

Indeed, it was no coincidence, we were told by Dr. William Hunter, whom we introduced in Chapter Three, that his disastrous early experience at Wesleyan turned around only after he moved out the “Black House,” formally known as Malcolm X House. Hunter had never wanted to be there in the first place. He had applied to room with a white friend from home, but Wesleyan informed him that he would be living in the Black House. There Hunter—a very strong student who had needed no racial preference to get into Wesleyan—lived amid many academically struggling inner-city kids. “I was astounded by how self-segregated they were,” Hunter recalls. “Even in the dining hall, there was a ‘black section,’ where all minorities were expected to congregate, at the risk of being ridiculed or ostracized. There was a great deal of peer pressure in this regard. Leaving the Black House was an essential part of my reasserting and rediscovering myself.”

Such harm from self-segregation is common, says Brian Corpening, assistant provost at the University of Oklahoma Health Sciences Center, who is black and is a veteran of four other colleges. “That just further isolates,” he says. “The whole point is for them to learn how to interact with people who are different from them and come from a different background. [Self-segregation] not only stratifies the students but, with the black kids, replicates what they came from at home.”

Corpening adds, “The existence of African American student advisers, Latino student advisers, Native American student advisers, and multicultural advisers tends to create a segregated environment within the overall college/university setting. It lets the overall infrastructure off the hook through the existence of separate ‘advisers,’ and it stunts the development of minority students—and, for that matter, white students—because the opportunities for dialogue and interaction become limited.” Schools should have a single “chief diversity officer,” not a diversity bureaucracy, he says, and the job should be “fostering interaction and sharing/learning among diverse groups on the campus rather than creating structures that maintain a sense of separation.”

For their part, white students may find the minority students’ behavior off-putting and clannish (even if the whites are clannish themselves). And plausibly, white and Asian students will notice that minority students often have weaker academic preparation, that they struggle academically, and that they often miss key points made in small classes. In other words, large preferences (and consequent, corresponding large gaps in college performance) could easily contribute to negative stereotypes about minority performance and perhaps minority ability in the minds of students of all races. None of these effects are quite what we have in mind when we talk about a healthy racial diversity on campus.

The danger that the widespread use of large racial preferences could reinforce—not dispel—negative stereotypes about blacks, American Indians, and, to a lesser extent, Hispanics would seem so obvious and so awful as to make this the foremost topic that scholars doing diversity research should address carefully. However, we are not aware of a single study that has addressed this topic. This seems incredible until one remembers the politics of diversity research and the broader politics of race on campus. The political cost of carrying out such research, as we illustrate in Chapter Eleven, would be high indeed.

Yet the interaction of preferences with racial attitudes is perhaps the most pervasive theme that has emerged from our interviews with former students who received preferences.

Jareau Hall breezed through high school in Syracuse, New York. In the top 20 percent of his class, he had been class president, a successful athlete, and sang in gospel choir. He was easily admitted to Colgate University, a moderately elite liberal arts college in rural New York; no one pointed out to Hall that his SAT scores were far below the class median. He immediately found himself over his head academically, facing far more rigorous coursework than ever before. “Nobody told me what would be expected of me beforehand,” he recalls. “I really didn’t know what I was getting into. And it all made me feel as if I wasn’t smart enough.” But just as surprising and upsetting was the social environment in which he found himself. “I was immediately stereotyped and put into a box because I was African American,” said Hall. “And that made it harder to perform. It may have been somewhat internalized, but people often made little derogatory comments or they would do things like come up and touch your hair. There was a general feeling that all blacks on campus were there either because they were athletes or they came through a minority recruitment program and might not really belong there. That was just assumed right away.” Hall dropped out after his freshman year, though he eventually returned to Colgate and obtained his bachelor’s degree.

“People thought I was only there because of affirmative action, so I tried to make myself invisible,” recalls the young Dartmouth grad we mentioned in Chapter Three. “You just felt that people resented affirmative action because they thought it led to students being accepted to college unfairly. They may not always have said that directly to me, but their remarks certainly implied it. And you would hear them say that about other minority students behind their backs—the idea that they didn’t belong there. So you just knew.”

“At elite institutions, the walls seem to whisper, ‘White males are superior . . . . African-Americans and Hispanics are inferior.’” So wrote Carol Swain, an African-American professor at Vanderbilt, in a book that often touches on mismatch themes. Swain suggests that her own rise to academic success occurred in part because she was not mismatched as an undergraduate. She started at a community college and received her bachelor’s, magna cum laude, from Roanoke College, eventually getting a doctorate at the University of North Carolina and an L.L.M. at Yale. Many black undergraduates whom Swain has encountered at elite schools, however, “seemed immobilized by the belief that they were incapable of reaching the high academic standards of the institutions. Many seemed to have internalized notions of black inferiority as well-meaning college advisers and other minority students contributed to their insecurities. At one college I attended,” Swain recalls, “a well-meaning advisor warned me during my first semester not to expect to perform there as well academically as I had in other settings . . . . I cannot see . . . how affirmative action policies which routinely place black students in institutions where whites and Asians are more academically accomplished can fail to engender this type of condescending mentality.”

Jocelyn Ladner-Mathis, a veteran college administrator who is now associate dean of liberal arts at Cuyahoga Community College in Cleveland, brought her own long experience with mismatch to bear on helping her own son choose a college in 2007. “I would think, ‘Thomas can get into this school, but can he get out?’ That is the first thing you pose to students. ‘You can get in, but can you get out?’ I have just seen so many kids stumble and get derailed along the way.’”

Ladner-Mathis, who herself had almost no contact with white people before becoming one of the first blacks to enter Mundelein College in Chicago, struggled academically but survived, and she considers herself a beneficiary of affirmative action. But she recalls, “A lot of my fellow African American students didn’t make it and dropped out.” Now as then, says Ladner-Mathis, many of “our students are coming completely unprepared, and it is really, really scary . . . . Affirmative action places you in situations where you can ultimately be successful. But it does come with a price. You can lose so much. You can lose your mind basically.”

Esther Cepeda, now a successful journalist, did well in school and college. Her mismatch experience began when she was given a scholarship to enter a math-intensive MBA program at Northwestern. “I think affirmative action helped me get in,” she recalls in a home interview. “It doesn’t feel good to know that I was not as qualified as other people for that program and I got in anyway.” And, as she elaborated in a 2012 op-ed, “I was not academically equal to my peers and woefully unprepared for the math-heavy statistical analysis needed to complete the basic courses in data mining.” Her low first-quarter grades put her on academic probation; she ended up leaving without a graduate degree, and “I left with serious bruises on my psyche and ego.” She spoke of the “sting that never seems to go away” from being “seen as someone who succeeded only because of affirmative action” and as “the official Hispanic, routinely called upon to enlighten my white classmates about Latino consumers’ struggles in the barrio with English language acquisition, gangs, and discrimination—none of which I’d ever had any experience with.” When asked about her “tough upbringing,” she would respond, “I didn’t have a tough upbringing. I didn’t grow up in a dump.”

Professor Phillip Richards, an African American English professor at Colgate, has sought to conceptualize broadly the systematic ways that racial preferences can harm black students and lead to “liberal re-segregation” of “marginal, academically inferior blacks” and to “an exceptionally self-destructive form of alienation.” We quote Professor Richards’s manuscript, with gratitude for his permission:

Colgate . . . creates a façade of racial diversity, while socializing blacks in an “integrated” world where they are implicitly prepared for a subordinate role in a globalized multi-ethnic sphere. The most recent Campus Climate Survey [at Colgate showed that] a majority of black and Hispanic students report that they are not pursuing the professional goals for which they came to Colgate. By any measure, these students are not being served in an educational sense by the university. And by implication they are not part of the community of scholars, teachers, and students at Colgate. . . .

[D]eparting cohorts of black students . . . rarely returned with their classmates for Colgate’s festive reunions. They did not wish to celebrate the assumption of their racial inferiority which underlies the institution’s central social knowledge.

The dull, lifeless work I receive from these black students tells me that they have not undergone Colgate’s special variant of elite socialization comprised of (for the most part exclusively white) fraternity and club life, as well as an ambitious academic regime. Unlike the upper class white majority of students, my black seniors do not show the analytic and expressive skills gained not only in class but also in the school’s wide range of extra-curricular activity. . . . And nothing so much characterizes the black American students’ essays than an anomic tone, the shadow of an eradicated selfhood set down in prose. Their failures are not only the consequence of academic inadequacy, but also the particular academic ethos at the heart of the school’s economically, socially, and politically flourishing life. No one could expect high academic performance from individuals so self consciously inadequate and consequently miserable as they.

Not all of these voices are singing exactly the same note. Certainly among the memoirs we have read and the people we have interviewed there are differing—sometimes widely differing—interpretations of why campus climates on elite colleges leave so many black and Hispanic students feeling like outcasts. But a remarkably pervasive theme that we derive from those we quote—and many other former students and administrators we have interviewed—is that a healthy social diversity across racial lines is extraordinarily difficult to build on a foundation of large racial preferences.

As we pointed out, diversity research rarely takes preferences into account. But when it does, the findings fit the experiences of our interviewees. For example, several studies have found that large racial preferences are directly associated with more negative self-images among the recipients. And then there is a study by three economists that should be considered the gold standard in the field—one that had actual data on the friendship networks of tens of thousands of students attending a range of colleges from moderately selective to the most elite and that explicitly attempted to understand the role of academic indices and relative academic position on cross-racial interaction.

In “Representation versus Assimilation” economists Peter Arcidiacono, Shakeeb Khan, and Jacob Vigdor found that students are much more likely to form friendships at college with other students whose level of academic preparation is similar to their own and that this is true both for same-race friendships and cross-racial friendships. Where racial preferences are large, this directly dampens the number of cross-racial friendships. The authors point out that, given the patterns they observe and the consequences of the cascade effect, our current racial preference system sharply reduces the number of cross-racial friendships that would occur if schools used smaller preferences.

Another suggestive example comes from law schools, where “study groups” have been a traditional survival strategy for first-year students. Anecdotal evidence—and common sense—suggests that students will tend to form study groups with students of similar academic strength. Our interviews have found that whereas Asians and whites often form study groups together, blacks and Hispanics often end up in same-race groups, plausibly because whites and Asians do not consider students regarded as “affirmative action admits” desirable study group members. Analyses of data on the effects of study groups shows that whereas Asian and white law students raise their grades by participating in a study group, black and Hispanic law students who do the same achieve no comparable benefit, even though they spend just as many hours studying as their white and Asian classmates. This is a result we would expect if “minority” study groups are almost entirely composed of students who are struggling academically.

One of the most widely discussed but often misunderstood strains of research in the diversity field explores something called “stereotype threat.” The idea is that people perform at their best when they are confident; factors that erode self-confidence, which often operate on a subconscious level, also erode performance. Hearing or being reminded of a negative stereotype about a group one belongs to hurts short-term performance; hearing a credible positive stereotype helps. If women are reminded before taking a math test that women tend not to do as well in math as men (a common but increasingly inaccurate stereotype), the average performance of women on the test goes down. Blacks are particularly vulnerable to negative stereotypes about black intelligence. In a famous series of experiments at Stanford that first established the idea of stereotype threat, Claude Steele and Joshua Aronson found that when black undergraduates were given a very challenging set of verbal puzzles to solve, their performance suffered significantly if, before starting the puzzles, they were told that the test would evaluate cognitive ability.

These results are often taken as evidence that black academic difficulties are merely an artifact of a hostile environment—make some simple modifications to the environment, and the “test-score” gap between blacks and whites might largely disappear. But that’s not quite what the research shows. Steele and Aronson found in their lab that when the test administrators did nothing to artificially “activate” stereotype threat, blacks and whites did as well on the complex verbal puzzles as would be predicted by their verbal SAT scores. When stereotype threat was activated, blacks did worse than their SAT scores would predict. In other words, the preexisting test-score gap is real, but performance gaps can be worsened (creating the so-called black underperformance problem) by activating stereotype threat.

Although there is a vast literature on stereotype threat, very few scholars in the field have considered what is probably obvious to most readers of this book: Selective colleges’ use of large racial preferences will very plausibly cause or aggravate stereotype threat. As we have seen, at these schools blacks are immersed in an environment not only where they are at a substantial academic disadvantage but also where their race often marks them as someone at an academic disadvantage. If a malevolent force wanted to establish a breeding ground for negative stereotypes of minority academic ability, it would be hard to come up with a more fertile petri dish.

One of the best studies of social interaction in college, undertaken by Harvard psychologist James Sidanius and three colleagues, examined intergroup relations at UCLA over a period of several years. They did not find that black and Hispanic students automatically performed worse simply by believing they had benefited from affirmative action. But they did find that “when black and Latino students were also concerned that negative stereotypes about their group’s intellectual ability could be true of them or that their academic performance would shape others’ views of their ethnic group . . . then their suspicions about having been admitted through affirmative action hurt their subsequent academic performance.” Given the difficulty of measuring the effects of subtle psychological processes on actual real-world grades, this is a powerful and remarkable finding. Stereotype threat is probably real, and it plausibly will be most severe for students admitted with the largest racial preferences.

GRADUATION

Probably the most famous book in the affirmative action literature is The Shape of the River, a magisterial 1998 study by William Bowen and Derek Bok, respectively the former presidents of Princeton and Harvard. Bowen was also the president of the Mellon Foundation, and with Mellon resources they created an extraordinary database known as the College and Beyond study (C&B), which gathered data on tens of thousands of students in three college cohorts, including colleges ranging from the “super-elites” to good state schools and historically black colleges. Bowen and Bok were both strong supporters (and implementers) of affirmative action, and their book was generally triumphant about the virtues and achievements of racial preference programs in college. For years, many took their work as definitive proof that blacks and other minorities benefited greatly from large admissions preferences into elite schools.

In fact, The Shape of the River had some notable flaws as a work of scholarship, many of which were detailed in an eviscerating 1999 essay by historian Stephan Thernstrom and political scientist Abigail Thernstrom. As we discuss in Chapter Fifteen, Mellon refused as a matter of policy to make the College and Beyond data available to other scholars to replicate and check Bowen’s and Bok’s findings. The Mellon Foundation did make the data available to scholars who passed through an arduous screening process, and many of these scholars—though not engaging in the forbidden acts of replication—have arrived at findings that are significantly inconsistent with Bowen and Bok’s original claims. Bowen and Bok also tended to be quite coy about directly testing the effect of preferences (on which they had incomparably good data) as opposed to examining relative outcomes of groups that benefited from preferences to differing degrees (e.g., blacks and whites). This along with many other research choices tended to blur and obscure their results.

One of most salient findings from Bowen and Bok was their claim that racial preferences bolstered the college graduation rate of black students. They demonstrated this by dividing the colleges in their database into three tiers of selectivity and showing that, other things being equal, blacks were more likely to graduate if they attended the top tier of colleges. This, they suggested, cut heavily against the mismatch hypothesis.

However, as others have argued, the Bowen and Bok findings can be just as easily read as a confirmation of mismatch. As we explained in Chapter Two, the peculiar workings of the cascade effect mean that racial preferences are smaller at super-elite colleges than they are at very elite colleges, and they are smaller at very elite colleges than they are at plain elite colleges. The black students for whom Bowen and Bok found the highest graduation rates—those attending super-elites—were, very plausibly, students who were on average substantially less mismatched than were black students at less elite schools. The authors could have greatly increased the power of their analysis had they examined how graduation varied with the size of a racial preference rather than by how it varied with school eliteness.

There were other, more technical but equally important problems with the Bowen and Bok analysis of graduation (we examine this further in an appendix posted to the website, www.mismatchthebook.com). But in a broad sense, they simply missed the point. The super-elites have such strong students and such deep support that their graduation rates are not far below 100 percent; even struggling students are given many opportunities to come back and finish. But these schools represent a tiny, tiny fraction of blacks receiving large preferences in higher education. As Figure 2.1 in Chapter Two suggested, perhaps the greatest harm done by the racial preferences used at the very elite schools is their cascading effect on somewhat less elite schools, which are effectively forced to use even larger preferences, and so on down the line, greatly aggravating the overall scale of the mismatch problem.

Studies that examine broader swaths of American higher education often find strong evidence that racial preferences produce lower college graduation rates. Economists Linda Loury and David Garman found that students who were mismatched had significantly lower graduation rates than those who were not. Economists Audrey Light and Wayne Strayer got the same result from their own cross-sectional analysis; sociologist Marta Tienda, in contrast, found little evidence of mismatch. All of these studies are, in our view, biased by the same “selection effect” that we discussed in detail in Chapter Five. These studies rely on comparisons of students who are at schools of differing eliteness; the analyst can observe only a few characteristics of each student (e.g., their SAT score); the “unobserved” characteristics in any particular dataset tend to favor the students who have (perhaps because of them) been admitted to the more elite school. These students will thus have higher graduation rates because of their unobserved characteristics, and that will skew the analysis to favor students attending more elite schools. It is, therefore, reasonable to assume that each of the studies mentioned above understated whatever mismatch effect actually exists. Taking this bias into account, these studies as a group provide substantial—if not definitive—evidence that mismatch reduces minority graduation rates. In Part III we will see a test of graduation mismatch that neatly avoids this selection effect and thus yields more definitive results.

EARNINGS AND OUTCOMES

Ultimately, a central purpose of racial preferences is to increase the presence of successful minorities in the economy and especially among the ranks of leaders in the professions, the sciences, business, and politics. How does mismatch affect these long-term outcomes? This is notoriously difficult to measure, and many of the issues are fairly technical, so we have relegated a full discussion of this question to the appendix posted on the book’s website, www.mismatchthebook.com. Here we content ourselves with a few key points.

We concede at the outset the possibility that the super-elite colleges and professional schools are sui generis, in a class by themselves. Going to Harvard or Yale confers lifetime reputational advantages and opens doors both at school and afterward that might make important differences in careers. Moreover, as noted earlier, the cascade effect means that the typical black at Harvard or Yale has received a smaller admissions preference than the typical black at a second-tier school and, thus, will be less harmed by mismatch.

Conversely, the hard evidence that elite schools really do confer important career advantages, especially to those entering with preferences, is surprisingly weak. Bowen and Bok present analyses that purport to show such advantages, but as we discussed in the earlier section on graduation, those results are susceptible to exactly the opposite interpretation that they offer. And Bowen and Bok’s analyses also show that ending up in the bottom third of one’s class (as do most students who receive large preferences) has a large, negative effect on long-term earnings.

Stacy Dale and Alan Krueger (the latter currently chairs President Obama’s Council of Economic Advisers) obtained access to Bowen and Bok’s data and performed a clever analysis similar to the first-choice/second-choice discussed in Chapter Five. That is, they found pairs of students who had been accepted by similar or identical pairs of colleges, with one student ending up at a more elite college than the other student. They then examined the earnings of these pairs fifteen years out of college. Their models suggested that students attending less elite schools earned as much as and perhaps more than similar students attending more-elite schools. Dale and Krueger did not examine mismatch specifically, but their results powerfully suggest that school eliteness is far less important to career outcomes than is generally thought. A wide array of other research suggests the same thing. When we control for the inveterate “selection effect” problem and look at a broad range of schools, there is no decisive advantage to attending a more elite school.

This implies a second generalization: Attending nonelite schools is not harmful to one’s career. Students with modest credentials who are interested in science and avoid mismatch by attending a less elite school have excellent long-term outcomes, with earnings well in excess of otherwise similar students who did not obtain science degrees. Students who attend historically black colleges also have very strong outcomes on a whole array of measures. Strong law students who attend nonelite schools regularly get better grades in law school and go on to outearn similar students who attended more-elite schools.

Some people find such statements counterintuitive. After all, a huge proportion of American elites have degrees from elite institutions, including a large fraction of congressmen and senators, most presidential nominees of major parties, most chief executives of large corporations, and most partners at the most elite law firms. Does that not imply that these schools are seeding grounds for greatness? Actually, it does not. The fallacy lies in the fact that top schools invest a good deal of time and effort in selecting the most talented applicants in the nation. For example, careful analysis shows that if we examine the proportion of the most talented law school applicants who go to the super-elite schools, and then examine the proportion of elite law firm partners thirty years later who attended those schools, the second proportion is lower than the first. A remarkable proportion of American Nobel Prize winners attended nonelite colleges.

A key reason behind these relationships—and this takes our discussion back to mismatch—is the effect of grades and, thus, learning on long-term outcomes. Doing well in school helps long-term outcomes, and doing badly hurts them. Again, there is a vast literature on this subject; we will highlight here one particularly interesting finding.

Starting in the mid-1980s the University of Michigan Law School began sending surveys to all of its alumni who had graduated fifteen years before, asking them detailed questions about their careers since law school. About half of the school’s graduates responded, and the school then matched these surveys against its own records to determine each graduate’s grades.

In Figure 6.1 we show the fortunes of nearly twenty years of Michigan law students who entered relatively large law firms after they graduated. This figure splits these graduates into ten equal “deciles” according to their law school grades. We can see two clear patterns. First, students with higher grades were much more likely to enter the highly sought jobs at big firms because the big firms preferred to hire students with strong academic records.

 

FIGURE 6.1. People with Good Law School Grades Succeed as Lawyers

Source: UMLS fifteen-year surveys of alumni for classes of 1972–1985; see text for details.

But second—and this is the key point—the law firm associates who had high law school grades were far more likely to still be at the firm fifteen years later, which, in this world, means that they survived the associate competition and became partners at the firm. Some associates leave because they don’t like the work (or the intensity of the work), and many others are fired if they fail to make partner. But the association between high grades and successfully obtaining a highly coveted partnership at these firms is, well, staggering. The associates with high GPAs were seven or eight times as likely to stay and become partners as those with low GPAs. As with the bar passage data we discussed in Chapter Four, these are correlations one rarely observes in human affairs.

There are a couple things that make the link between partnership and grades particularly impressive. Although law firms almost always take law school grades into account when making hiring decisions, they almost never look at grades when they make promotion decisions. Why should they? After all, they have plenty of their own internal information about how these associates have performed. Thus, looking at partnership decisions here is like conducting a “blind taste test”: The firms overwhelmingly promote people who happen to have had high grades without knowing or caring that they did. Moreover, the people with low grades in this analysis are exceptional people; they were able to get big firm jobs against the odds, despite their low grades, suggesting that they were more likely to have other important qualities, such as a magnetic personality, leadership skills, or a blood relationship to a senior partner. Yet despite these qualities, they did miserably in the competition for partnership. The strong implication is that law firms, routinely criticized for giving too much importance to grades in hiring decisions, may in fact be putting less weight on grades than perhaps they should.

This figure includes only white graduates from Michigan’s law school. We excluded minority graduates because many of these have lower grades, and one might then argue that the association between high grades and promotion was simply capturing, indirectly, racial discrimination by law firm partners. But when we do a similar analysis for minority lawyers, the pattern looks just the same. Indeed, these data help to explain another type of mismatch phenomenon, this one in the workforce. From the 1970s onward, law firms faced both internal and external pressure to use increasingly aggressive racial preferences in hiring to diversify their firms. By the 1990s the vast majority of large law firms were using explicit double-standards in hiring, and most black law graduates they hired (because of law school mismatch) had low grades. Few of these hires ever became partners at the firms they joined; most left after a few years, often with considerable bitterness at the lack of opportunity they encountered within firms. But these data along with much related research imply that the problem was low grades, not outright discrimination.

This research as well as a good deal of related work compellingly suggests that learning in school and the grades that measure that learning matter greatly for one’s long-term career success. We have tried to illustrate in this chapter how large racial preferences create a campus environment that is not conducive to learning and that in many ways undercuts the intellectual self-confidence that is the vital handmaiden to learning. It is harder to show with data the effect of mismatch on college English majors compared to students in more rigid curricula like college chemistry majors or first-year law students. But from what we can tell, the effects of mismatch are pervasive indeed.