Mismatch

The University of California is a large, complex system. Its ten campuses enroll well over two hundred thousand students and an almost equal number of staff. In the mid- and late-1990s eight campuses enrolled undergraduates. They range widely in eliteness, as Figure 8.1 in the last chapter suggests, from Berkeley and UCLA at the top, which rank on most lists of top twenty American universities, to UC Santa Cruz and UC Riverside, which are important research universities that rank probably among the top two hundred American schools in student strength and faculty reputation. These schools together comprise the first of three tiers in California higher education. Under the “Master Plan” adopted by state authorities in 1960, the eight UC campuses were each envisioned as research universities that would have graduate programs and would collectively make room for any California high school student whose academic index placed her among the top eighth of all students (these top students would come to be called “UC eligible”). A broader pool of students—those in the top third of the academic index—would be eligible for one of the twenty-plus campuses of the four-hundred-thousand-student Cal State system. Those who qualified for neither the UCs or Cal State could attend one of the dozens of community colleges operated by the state (some of them quite good). And, importantly, students who did well in community college nor at Cal State could transfer after their freshman or sophomore year to one of the UC schools.

This background is important for a few reasons. We focused on freshman admissions in Chapter Eight; but in addition to the roughly twenty-five thousand freshmen enrolled at the eight UC campuses collectively each year in the late 1990s, there were also some ten thousand transfer students who arrived as sophomores or juniors. At Berkeley and UCLA, which were sought after by transfers in particularly large numbers, they often made up between 30 and 40 percent of the senior class. Before Prop 209 many UC campuses applied very different standards to transfer students depending on their race. But in an era without overt racial preferences, transfers were an important mechanism to identify strong students of all races who, in early adulthood, were on a trajectory of rapid academic improvement. Managed properly, a transfer system was a good way to elevate talented students when they were ready rather than gambling on their success as freshmen.

UC eligibility also played a crucial role in shaping the effects of Prop 209. Because of the racial test-score gap, far more than one-eighth of Asian high school students in California qualified for UC eligibility, whereas far fewer than one-eighth of black or Hispanic students qualified. For many years before Prop 209 university administrators had sidestepped the “top eighth” rule through “special admissions,” but still, even before Prop 209 effectively shut down race-based special admissions, the UC system as a whole did not have anything comparable to the racial proportionality that one could find at most elite colleges. In the years before Prop 209 blacks made up just over 4 percent of UC freshmen and Hispanics made up 13 to 14 percent of UC freshmen—both numbers far below their relative numbers in the pool of high school graduates.

This meant that before Prop 209 the various UC campuses were competing intensely among themselves for the limited supply of black and Hispanic freshmen. Not surprisingly, Berkeley and UCLA often won this competition: In 1997 nearly half of all UC blacks and nearly a third of all UC Hispanics were at the Berkeley and UCLA campuses. The constraint UC eligibility created also meant that a substantial proportion of UC freshmen had academic indices that already qualified them for admission to at least one of the lower-tier UC schools. This is why, before Prop 209, racial preferences at Berkeley and UCLA were very large (and close to national norms), whereas preferences at the less elite UC campuses were generally modest.

How did Prop 209 change all these dynamics? As we have already observed, UC officials were very anxious about the prospective drop in minority enrollment, which they expected to be very large. They could not openly defy the ban on racial preferences, so instead they instituted a variety of ostensibly race-neutral measures intended to stem the decline. Some created or expanded socioeconomic preferences. Some began to favor applicants from inner-city schools. We suspect from our statistical analysis of admissions decisions at the time that some campus admissions offices, aware of the intense concern about diversity numbers, looked for signs in applications for minorities they could favor. So, for example, in the three years before Prop 209 a black or Hispanic applicant with an academic index in the low 600s had a 63 percent chance of being admitted to Berkeley or UCLA, compared to an 8 percent chance for a similar white applicant and a 7 percent chance for a similar Asian applicant. In the first three years of putatively race-neutral admissions, the black or Hispanic applicant’s chances fell sharply, to only 27 percent. But that was still far higher than the chances faced by whites (7 percent) or Asians (9 percent) with the same academic index.

These measures significantly moderated the effect of Prop 209. Without them, the cascade of blacks and Hispanics to less elite campuses would have been more severe; more students probably would have been displaced from the UC system altogether and into the generally less elite Cal State system. The use of race substitutes gave race-neutrality a less severe feel.

But the same race substitutes also preserved substantial gaps at every campus between the average academic index of whites and Asians on the one hand and blacks and Hispanics on the other. True, after Prop 209 there was much more overlap at every campus in the academic qualifications of students of different races, and both ends of the qualification spectrum were more racially heterogeneous. But overall, the academic index gaps between whites and Asians on the one hand and blacks and Hispanics on the other shrank by only 20 to 30 percent.

The simple answer is an emphatic yes. Minority graduation rates rose rapidly in the years after Prop 209, and on-time (four-year) graduation rates rose even faster. For the six classes of black freshmen who entered UC schools in the years before race-neutrality (i.e., the freshman classes of 1992 through 1997), the overall four-year graduation rate was 22 percent. For the six years after Prop 209’s implementation the black four-year graduation rate was 38 percent. Thus, even though the number of black freshmen in the UC system fell almost 20 percent from 1997 to 1998, the number of black freshmen who obtained their degrees in four years barely dipped for this class, and the entering class of 2000 produced, four years later, a record number of blacks graduating on time. The increase in black six-year graduation was less dramatic (63 percent before and 71 percent after Prop 209) but still substantial.

The improvement in Hispanic graduation rates was striking too, though, as always, the smaller initial racial preferences Hispanics receive imply that we should see smaller improvements from the new admissions regime. The four-year graduation rate for entering Hispanic freshmen averaged 27 percent in the six years before Prop 209 and 40 percent in the six years after. Six-year graduation rates rose from 69 to 74 percent. Because the Hispanic drop in enrollment after Prop 209 was smaller and because the underlying demographics of California produced steadily larger numbers of strong Hispanic applicants, the absolute numbers of Hispanic graduates rose at a stunning rate. The entering Hispanic freshman classes of 1992, 1993, and 1994 produced a total of 2,005 on-time graduates. The three classes of freshmen entering from 1998 through 2000 produced 3,577 on-time Hispanic graduates from the university.

Not all of these gains, however, should be attributed to the narrowing of racial preparation gaps. The early 1990s were difficult financial years for the university; improving fortunes in the late 1990s allowed the university to provide more student-support services and expand course offerings, both of which improved on-time graduation across the board. And, of course, reducing the number of academically weak black and Hispanic students would tend to boost the average graduation rates of the remainder. Is it possible to isolate these various effects from one another? The UCOP data (a dataset described in Chapter Eight) contains extensive information on every UC freshman enrolled in the years before and after Prop 209—not only on their high school record, test scores, and applications to UC schools but also on their performance and outcomes after they enrolled at UC. A team of economists led by Joe Hotz (a labor economist at Duke and former chair of economics at UCLA) and Peter Arcidiacono (also of Duke) examined Prop 209’s effect on UC undergraduate graduation rates for blacks, Hispanics, and Native Americans, using a wide range of controls to compare similar students attending campuses where their academic indices were closer to or further away from the median student in the entering class at that school—in other words, they compared students with varying degrees of academic match with their classmates. They found that Prop 209 had the effect of raising five-year minority graduation rates from 3 to 7 percent points. They also concluded that the effect would have been significantly larger had the university more rigorously implemented race-neutrality and thereby further reduced disparities in the academic preparation of students on each campus.

These findings perhaps do not settle the issue of whether mismatch always hurts black and Hispanic graduation rates, but they should weigh heavily in the debate. There is simply no other study that has so effectively handled the difficult problem of “selection effects” that we discussed in detail in Chapter Five. In this case, we can for once be confident that we are comparing fundamentally similar students experiencing different levels of mismatch.

Moreover, there are three reasons to think that Hotz and Arcidiacono are, if anything, underestimating Prop 209’s true effects. First, the impact of mismatch on four-year graduation appears to be larger than the effect on the five-year graduation rates they studied. Four-year graduation is generally the ideal, and students who avoid academic difficulty or who stay with their original major are more likely to finish in four years. Second, it is plausible that the better graduation outcomes for better-matched students will contribute to a broader academic climate of success, helping to propel the general increase in graduation rates. But third and most importantly, Hotz and Arcidiacono did not have data on and therefore could not take into account the positive effect of reduced racial preferences on transfer students.

As we noted earlier, an open transfer policy—that is, one that makes it relatively easy for students who do well at nonelite public schools to transfer up to elite ones—offers an ideal way to avoid mismatch while capturing what we would view as the true spirit of affirmative action. If high school seniors with good but not excellent academic records are able to attend colleges where they will not be overwhelmed, if they are able to demonstrate at those colleges that they can do excellent work, and if excellent students are able to transfer easily to more elite schools, then we are providing effective access without inviting the painful consequences of mismatch. We think California’s university system came somewhat closer to this ideal as it eliminated formal racial preferences, and the data on the outcomes for transfers supports this view.

During the four years before Prop 209’s implementation, less than 19 percent of black students who transferred into a new UC school as juniors graduated on time. That rose to 34 percent in the first four years after Prop 209, and 38 percent in the four years after that. (These are impressive numbers, as transfers often face especially great challenges in meeting new requirements.) The corresponding rates of eventual graduation for these black transfers were 61 percent before the new admissions regime and 74 percent after. The on-time graduation rate for Hispanic transfer students rose from 26 percent before race-neutrality to 42 percent afterward; eventual graduation rates rose from 74 percent to 81 percent. The number of black and Hispanic transfer students dipped for a few years after the change in admissions practices but then began to rise steadily. The number of blacks receiving bachelor’s degrees from the University of California as transfer students was more than 40 percent higher in 2007–2009 than it had been before Prop 209, an average of three hundred students per year. Most of these students started their college careers at community colleges or at Cal State. For Hispanics, the increase was over 60 percent.

Of course, Prop 209 had many other effects on black and Hispanic undergraduates. One of the most persistent claims by critics, often echoed by journalists and even UC officials, was that race-neutrality was “segregating” the UC campuses. Their argument was that fewer underrepresented minorities were getting into the elite schools; those that got in at all were clustered at UC Riverside, the least elite of the eight UC campuses. UC Riverside (which, relevantly, has a less idyllic campus setting than most of the UC schools) was henceforth to be the “minority” campus, and UC would enter into a sort of Jim Crow era.

The claim was ridiculous. As we discussed earlier in the chapter, the constraining rule of UC eligibility had always imposed limits on the number of blacks and Hispanics in the UC system, and before race-neutrality began Berkeley and UCLA lured a disproportionate number of these students to their campuses. Before Prop 209 Berkeley and UCLA had substantial numbers of black and Hispanic students, but many campuses had few indeed. Race-neutrality cascaded many blacks and Hispanics to less elite campuses, producing a more even distribution of students of all races across the campuses. Although UC Riverside did end up with the largest share of Hispanics of any single campus (it is, after all, located close to the largest residential concentration of Hispanics in the United States), it was still less than a quarter black and Hispanic after Prop 209.

We can objectively measure the degree of “segregation” across campuses with something called the “index of dissimilarity,” a measure often used in studies of urban segregation. For blacks entering as UC freshmen in 1997, the last year before Prop 209’s implementation, the index was .20, a very low number, indicating that the distribution of blacks vis-à-vis other races was only 20 percent off perfect proportionality. In 1998 the index fell to .18, and it dropped further in subsequent years. For Hispanics, the index of dissimilarity fell from .20 in 1997 to .15 in 1998. Substantively, this meant that many campuses that had very few blacks and Hispanics before Prop 209 (e.g., UC Santa Barbara and UC Santa Cruz) were significantly more integrated afterward.

For example, black and Hispanic grades improved, both in absolute and relative terms. As with graduation, there was already a trend in the UC system (and in American higher education generally) toward higher grades unrelated to Prop 209. But gains for blacks and Hispanics outpaced those of other groups. In the three years before the end of formal racial preferences blacks and Hispanics system-wide had freshman GPAs that put them at the 20th percentile of whites and Asians. As the credential gap between the races narrowed, GPA gaps narrowed by amounts that were identical or slightly larger; in 2001–2003 freshman GPAs for blacks and Hispanics had risen to the 30th percentile of whites and Asians. Academic probation rates for blacks and Hispanics declined commensurately.

Science persistence improved. For many years blacks and Hispanics applying to UC schools had been as likely to indicate that they planned to major in the natural sciences or engineering as white applicants did. But once they enrolled, students admitted with large preferences tended to have the same sorts of very high attrition rates that we saw documented in other settings in Chapter Three.

UC-wide, the number of black and Hispanic students graduating with STEM degrees steadily increased after the admissions reforms of 1998, and the number of science-interested students never graduating steadily fell. The share of black and Hispanic students majoring in STEM fields rose as well. Black and Hispanic engineering graduates rose by nearly 50 percent from 1997 to 2003, whereas the number of blacks and Hispanics majoring in ethnic studies and communications fell 20 percent.

As with the issue of graduation mismatch, the natural experiment of suddenly reducing the preparation disparities among different groups of students provides an excellent test of the science mismatch hypothesis. Marc Luppino, Roger Bolus, and one of us (Sander) completed an analysis of the UCOP data to measure the effect of science mismatch upon student persistence in science over a period from 1995 to 2003. We measured substantial mismatch effects for a variety of science outcomes, with magnitudes similar to those found by Smyth and McArdle (see Chapter Three).

Note how each of the accomplishments we have discussed—higher grades, higher graduation rates, and higher science-completion rates—is particularly remarkable in light of the others. Science courses grade harder, so it’s harder for the grades of a group of students to go up when more of them are studying science. Lower drop-out rates mean that the average final GPA of a group includes more students who hung in there, even if they struggled. It is remarkable that at the University of California, blacks and Hispanics in the post–Prop 209 era managed to simultaneously lift their achievement levels in multiple ways.

Preference programs at the doctoral level are much more subjective and ad hoc. Scores on such tests as the Graduate Record Examination (GRE) and college grades often provide a rough screen for admissions, but a history or physics department is likely to be much more concerned about whether an applicant has interesting ideas and has shown an ability to engage in independent research. Doctoral programs were much slower in the 1960s and 1970s to adopt preference programs, and because they rely on soft criteria and report little systematic data on admissions and performance, it is hard to measure the extent of preferences. But when they did, preferential admissions sometimes lacked any coherent standards, perhaps because many graduate admissions committees felt that even their subjective criteria might be unfair when applied to blacks and Hispanics.

John Ellis, a retired professor of German literature at UC Santa Cruz, provides an interesting window into the origin, operation, and effects of doctoral-level preferences on a UC campus. From the 1970s until 1986 he was dean of the school’s graduate division, and it was during his tenure that UC Santa Cruz began significant efforts to use preferences in recruiting graduate students. Ellis recalls that new federal programs in the 1970s provided rich incentives for schools to use preferences; the initial cost to the school of providing fellowships to black and Hispanic students was minimal, and he encouraged various departments to make active efforts to find promising candidates.

As Ellis tells it, his recruitment and outreach programs morphed in the hands of others into racial and gender preferences. The preferences were particularly large for black recruits, with a variety of harmful results. Often, the recruits were woefully “under-qualified students who were then psychologically harmed by their lack of preparation” and whose academic difficulties aggravated stereotypes about blacks’ academic abilities. Although Ellis felt that far less damage was done to the white and Asian candidates who, because of the preference programs, were displaced to slightly less elite schools, the loss of those strong-but-not-stellar students widened the preparation gulf between the highly qualified students admitted without preferences, and the preferentially admitted blacks and Hispanics.

Ellis is unusual for his passion on the issue, but the problems he identifies come up again and again when other UC faculty share their stories about the efforts of graduate programs to recruit black and Hispanic graduate students with large preferences and fellowships. As we discussed in Chapter Three, mismatch at the undergraduate level creates an extremely slender pipeline of black talent into doctoral programs. When schools aggressively compete for the small number of interested students, serious mismatch problems often ensue.

Again, the post–Prop 209 era at UC provides a chance to check the mismatch hypothesis with some admittedly imperfect data. Most graduate programs in the UC system were instructed to adopt race-neutral admissions for classes entering in 1997 and beyond. Total black graduate enrollment (not first-year entries but students in all years of study) began to fall in 1997, and from 1999 to 2004 it averaged about 25 percent below pre-209 levels. Yet the number of doctoral degrees granted to blacks crept upward and, taking into account the lagged time to degree, averaged about 20 percent higher in the post-209 era than before. Black graduate degrees in STEM fields held steady. Similar patterns occurred for Hispanics too, except that the overall decline in Hispanic admissions was smaller, and the growth in STEM graduate degrees was larger.

Once again we see a startling pattern: A reduction in racial preferences led to modest declines in black and Hispanic enrollment but an increase in black and Hispanic degrees. In the graduate school case it is plausible to conclude that Prop 209 caused two things to happen: Particular departments phased out the most aggressive preferences, and the aura of race-neutrality may have attracted some very strong black and Hispanic students.

One final development in UC’s post–Prop 209 era deserves mention. The improvement in black and Hispanic performance was particularly noticeable among the best students. Across all the UCs during the 1995–1997 period, less than one-tenth of blacks and Hispanics had GPAs of 3.5 or higher (compared to over one-quarter of whites and 30 percent of Asians). Both the number and proportion of blacks and Hispanics achieving high GPAs jumped after Prop 209, rising for them at a much faster rate than for whites and Asians.

It is not surprising that many more blacks and Hispanics would have had high academic achievement after formal racial preferences ended. Because, starting in 1998, the entire distribution of black and Hispanic academic qualifications moved up relative to whites and Asians, a larger proportion of black and Hispanic students were able to compete at the highest levels.

Diversity advocates have widely argued that blacks and Hispanics will perform better academically if they constitute a “critical mass” on campus. A recent, careful attempt to test this idea empirically found no support for the critical-mass hypothesis and some evidence that just the opposite was the case. The reshuffling of racial composition across the various UC campuses after Prop 209 provides an interesting opportunity to study this question. When we compare similar students across the various UC campuses and adjust for the racial composition of each campus, we find that black and Hispanic students earned slightly higher grades when the proportion of blacks and Hispanics on their campus fell. Strikingly, the effect is strongest for those black and Hispanic students arriving on campus with the highest academic indices. They are primed to do well, but they are also most likely to exceed the predictions of their academic indices on UC campuses with smaller minority populations.

This is only one finding, and we do not claim it has the rigorous support that lies behind findings like the research on undergraduate graduation rates at UC. But the finding makes sense in the context of the other research we have discussed. Many black and Hispanic students on UC campuses continued to be at a serious preparation disadvantage relative to the typical student on their campuses even after Prop 209. They were subject to some of the same pressures to withdraw from the academic competition that appears to occur at so many universities that use large racial preferences. But a smaller presence of academically struggling blacks and Hispanics may have made it more likely that they would, as individuals, have cross-racial networks and feel part of the broader campus community. That, very plausibly, helped to maintain their academic engagement.

The post–Prop 209 aggregate picture is nothing short of remarkable. If we compare minorities entering the UC system in 1995–97 (the final years of explicit racial preferences) with those in the post-209 cohorts, everything seems to be moving in the same direction:

• The number of blacks entering UC as freshmen in 2000 through 2003 is, on average, only 2 percent below pre-209 levels, and black enrollment jumps when we take into account transfers and lower attrition.

• The number of Hispanic freshmen is up by 22 percent over the same period, and again more when we include transfers.

• The number of blacks receiving bachelor degrees from UC schools rose from an average of 812 in 1998–2001 (the final cohorts entirely comprised of pre-209 entrants) to an average of 904 in 2004–2007 (the first cohorts entirely comprised of post-209 entrants). For UC Hispanics, the numbers rose from 3,317 to 4,428.

• The number of UC black and Hispanic freshmen who went on to graduate in four years rose 55 percent from 1995–1997 to 2001–2003.

• The number of UC black and Hispanic freshmen who went on to graduate in four years with STEM degrees rose 51 percent from 1995–1997 to 2001–2003.

• The number of UC black and Hispanic freshmen who went on to graduate in four years with GPAs of 3.5 or higher rose by 63 percent from 1995–1997 to 2001–2003.

• Doctorates and STEM graduate degrees earned by blacks and Hispanics combined rose by one-quarter from cohorts starting in 1995–1997 to cohorts starting in 1998–2000.