A Thumb on the Scale

The case for socioeconomic affirmative action

In February 2004, when Harvard allotted an additional $2 million per year in scholarship funds for undergraduates from families with incomes of $60,000 or less, the College estimated that 73.9 percent of matriculants came from the highest socioeconomic quartile of society, and 10.1 percent from the next-highest quartile. Just 6.8 percent of entering students came from the lowest quartile, and 9.2 percent from the second quartile (see “Class-conscious Financial Aid,” May-June 2004, page 62). In announcing the aid program, President Lawrence H. Summers observed that in America’s most selective colleges and universities overall, “only 3 percent of students come from the bottom income quartile and only 10 percent come from the bottom half of the income scale.…Children whose families are in the lower half of the American income distribution are underrepresented by 80 percent. These differences cannot be fully accounted for by native ability or academic preparation.” After the enhanced aid was publicized and the College stepped up recruiting among lower-income high-school students, more of them sought admission for the class of 2009, entering this fall (see "Brevia").

But does this effort suffice, given the responsibility of elite institutions like Harvard to fulfill their largest social purposes by promoting excellence and providing opportunity for students who can take advantage of the education on offer? William G. Bowen, LL.D. ’73, Martin A. Kurzweil ’02, and Eugene M. Tobin argue that more needs to be done. In their new book, Equity and Excellence in American Higher Education (University of Virginia Press), the seventh chapter—“Broadening the Quest for Equity at the Institutional Level: Socioeconomic Status, Admissions Preferences, and Financial Aid” (in part excerpted here)—makes the case for economic or class-based admissions preferences, along the lines of those already extended to “legacies” (and other categories of applicants). Noting that “the odds of both taking the [SAT] tests and doing very well on them were roughly six times higher for students from the top income quartile than for students from the bottom income quartile,” they document the overwhelming disadvantage lower-income students have in even entering the “credible pool” of applicants.

The authors also address the problem of college preparation, but conclude that waiting for long-term improvements in K-12 schooling is insufficient. In analyzing data on the 1995 entering cohort of 19 selective universities and colleges, they find ample room for putting a “thumb on the scale” to admit more applicants who have demonstrated their academic prowess despite the real disadvantages of lower socioeconomic status across a number of dimensions. (The 19 institutions include five Ivy League universities—Columbia, Harvard, Princeton, the University of Pennsylvania, and Yale; 10 academically selective liberal arts colleges—Barnard, Bowdoin, Macalester, Middlebury, Oberlin, Pomona, Smith, Swarthmore, Wellesley, and Williams—three of which—Barnard, Smith, and Wellesley—are for women only; and four leading state universities—Penn State, UCLA, the University of Illinois at Urbana/Champaign, and the University of Virginia.)

Though not excerpted here, the balance of the chapter makes the argument for sustaining race-sensitive admissions preferences alongside the new class-based criterion the authors propose. Class, they maintain, is not a proxy for race. The two preferences, they say, are complementary policies consistent with elite institutions’ aims of furthering educational excellence and equity.

~The Editors

Text excerpted from chapter seven of William G. Bowen, Martin A. Kurzweil, and Eugene M. Tobin, Equity and Excellence in American Higher Education. Charlottesville: University of Virginia Press, 2005. Reprinted with permission of the University of Virginia Press 


The continued successful pursuit of excellence in American higher education depends on opening the gates of opportunity wider. Historically, colleges and universities have been far more welcoming to white male students of European descent who were of the “right” religious persuasion, and usually from families able to pay, than they have been to others. Fortunately, substantial progress has been made in eliminating the most blatant forms of discrimination based on religion, gender, and national origin. But there is clearly more work to be done.

Race in America, and especially the treatment of African Americans, has been, and continues to be, the hardest marker of “disadvantage” to counter—in large part because “color” is such a visible marker and is associated with deeply ingrained stigmas and stereotypes. However, race is by no means the only lens through which to look. The more “privileged” colleges and universities (as defined by both financial resources and selectivity in admissions) might do more to address the disadvantages associated with low socioeconomic status (SES).

We should remind ourselves why there is such a strong case to be made for being broadly inclusive in “crafting a class”:

1. As part of their quest for excellence, colleges and universities want to attract the most promising students, and there has never been reason to believe that all outstanding candidates will be able to pay whatever fees are charged without help. Society at large needs all the trained talent it can marshal.

2. The quality (“excellence”) of the campus learning environment is improved for everyone when students from a wide variety of socioeconomic backgrounds are present. Students need to learn how to put themselves in other people’s shoes.

3. Finally, a commitment to this form of inclusiveness is an essential part of a broader affirmation of opportunity and is at the center of our concern for equity. An important societal goal is to enable individuals to move up the ladder of accomplishment as far as their talents, character, and determination will take them. This proposition, central to a well-functioning democratic society, is especially important at a time when education is more critical than ever before in determining access not only to the best jobs (and the accompanying economic rewards), but also to a broad set of less tangible opportunities that help us “live a life.” Moreover, the sense of democratic legitimacy is undermined if people believe that the rich are admitted to selective colleges and universities regardless of merit while able and deserving candidates from more modest backgrounds are turned away.



One of the key determinants of the composition of the student body at academically selective colleges and universities is the decisions made by the admissions office in choosing among large numbers of well-qualified candidates. Two other key determinants are the characteristics of the applicant pools (who applies) and the yields on admission offers (who comes). We know for the schools in our study that these factors combine to produce entering classes in which 10 to 11 percent of the students typically come from families in the bottom income quartile, just over 6 percent are first-generation college-goers, and only about 3 percent meet both criteria—that is, are from low-income families with no previous experience of college.

These percentages are clearly low in relation to the percentages in the underlying population. But it would be churlish in the extreme not to recognize how much these prestigious institutions, with their high degree of selectivity and their established constituencies, have accomplished. Success in enrolling the present number of students from low-income families can be attributed to a combination of active recruitment, a “need-blind” admissions philosophy, and a financial aid policy that meets the full need of all admitted students who elect to matriculate. [This means that] their admissions offices pay no attention to the financial circumstances of the student or the student’s family in making admissions decisions. These colleges and universities also devote substantial resources to financial aid and do very well in meeting the full need of their admitted applicants.

Need-blind admissions policies at many of the private institutions in our study evolved in the late 1950s and 1960s (starting with the Ivy League) in conjunction with the establishment of a larger “Overlap Group,” representing 23 East Coast colleges, that agreed on expected family contributions for students who were admitted to more than one of the schools in the group.

The development of this elaborate machinery for defining what need-based aid meant took place at a time when many of these schools had no well-thought-out financial aid policy and, in many cases, lacked the financial resources needed to meet the commitments they would have liked to make. The primary purpose of need-blind admissions was to reassure students from modest circumstances that their lack of money would not be held against them in the admissions competition—a practice that surely existed historically when colleges were struggling to pay their bills and that exists today at many financially hard-pressed colleges and universities. These policies remain in place today even though the Overlap Group was disestablished in the early 1990s as a result of a highly controversial anti-trust case.

Establishing these policies was no easy task. The situation at Yale in the mid-1960s is described vividly by Geoffrey Kabaservice in his account of the changes that [President] Kingman Brewster and his colleagues (including the newly appointed director of admissions, “Inky” Clark) brought to Yale. Kabaservice describes “the wall of biases against an applicant from an excellent, competitive public high school, such as New York’s Bronx High School of Science in the 1950s”:

The student came from a public high school, which Yale’s admissions officers did not visit. He scored highly on aptitude tests, which Yale discounted. He had a specialized education, which many influential Yale faculty members thought unfitted him for the liberal arts. He focused on science or technology, which Yale had traditionally considered unsavory. He was almost certainly from a nonwealthy family, which handicapped him in the era before need-blind admissions. And he had no Yale alumni connection or feeder school tradition to boost his candidacy. Furthermore, even if the candidate from Bronx Science was not Jewish himself, he came from a school that was predominantly Jewish…, at a time when anti-Semitism in Yale’s admissions was covert but active…. It is not surprising to find that during [Yale president A. Whitney] Griswold’s first five years in office, Bronx Science sent only 7 graduates to Yale, while Andover (not nearly as selective) sent 275. Over the same five-year span, almost all alumni sons who applied were accepted.

It is one thing to overturn atavistic policies and practices of the kind that Kabaservice describes and to agree that a very different admissions philosophy is called for; it is quite another to find the resources required to implement a full need-blind admissions policy and a meet-all-need financial aid policy. In considering the feasibility of enrolling more students from poor families, resource constraints have to be taken into account. Throughout the 1970s, even as wealthy a university as Princeton, with an exceedingly generous student aid endowment, struggled to find ways to meet the full need of all admitted students. Many liberal arts colleges outside the Northeast (Oberlin is a good example) have frequently been unable to meet full need. Brown University was not able to move to a full need-blind approach until 2002. Also, it was not until 2001 that Princeton met the full need of foreign students—and it is likely that many schools that say they meet full need today are really referring to support for American students.

Resources plainly matter, and later we provide dollar estimates of what it would actually cost to adopt one hypothetical approach to going beyond need-blind admissions in order to enroll more students from low-income families. But first we need to harken back to the factual picture of how applicants from low-SES backgrounds are faring in the current admissions competitions at the 19 schools in our case study—in relation to applicants from more advantaged families and also in relation to special groups who already receive preferential treatment.



The most striking single finding to emerge from [our] empirical work relates to SES: the probabilities of admission at the 19 schools in our special study are essentially identical, at any given SAT level, for three groups of non-minority students: those from the bottom income quartile, those who are potential first-generation college-goers, and all others. (We exclude minority students from these comparisons because we know that they benefit from race-sensitive admissions policies, and we do not want to confound the results of this analysis of how applicants from low-SES backgrounds fare by mixing together the treatment of race and the treatment of SES.) More refined statistical analysis confirms that applicants from low-SES backgrounds, whether defined by family income or parental education, get essentially no break in the admissions process; they fare neither better nor worse than other applicants.

The obverse side of this core finding is that, overall, students from affluent families also get no special break once they enter the admissions pool. These 19 schools say, overwhelmingly, that they pursue need-blind admissions policies. And our robust empirical findings suggest that in fact they are need-blind. For these schools, the real question is whether being need-blind is good enough.

The results for three special groups of applicants—underrepresented minority students, legacies, and recruited athletes—stand in sharp contrast. Members of each of these groups have a decidedly better chance of being admitted, at any specified SAT level, than do their fellow applicants, including those from low-SES categories. The average boost in the odds of admission is about 30 percentage points for a recruited athlete, 28 points for a member of an underrepresented minority group, and 20 points for a legacy. For example, an applicant with an admissions probability of 40 percent based on SAT scores and other variables would have an admissions probability of 70 percent if he or she were a recruited athlete, 68 percent if an underrepresented minority, and 60 percent if a legacy. Applicants who participate in early decision programs also enjoy a definite admissions advantage—about 20 percentage points at the 13 institutions for which we have data.

The question for faculty, administrative officers, and trustees responsible for setting policy is simply stated: looking ahead, is this set of preferences the best way to allocate scarce places at highly selective institutions?

•We believe that minority preferences serve several basic educational and societal goals. First, as was acknowledged by the Supreme Court in Grutter v. Bollinger, a diverse student body provides educational benefits to all students. Second, as Justice Sandra Day O’Connor’s majority opinion also emphasized, it is important that these highly regarded schools—which are in many ways gateways to wealth and power in this country—are seen to be accessible to students from all racial backgrounds. Finally, minority preferences correct at least in part for disadvantages that these students, as a group, have experienced—and will experience—in a country that is still affected profoundly by racial stigmas. Race matters in America, and that reality needs to be acknowledged.

•Legacy preferences serve to enhance the ties of alumni/ae to their alma mater. Schools have an understandable interest in keeping their alumni/ae—who can and often do promote their institutions with an evangelical fervor—feeling positively about their institutions. As Harvard’s dean of admissions, William Fitzsimmons, puts it: “[The school’s alumni/ae] volunteer an immense amount of their free time in recruiting students, raising money for their financial aid, taking part in Harvard Club activities at the local level, and in general promoting the college…. They often bring a special kind of loyalty and enthusiasm for life at the college that makes a real difference in the college climate…and makes Harvard a happier place.” In addition, all of these institutions rely heavily on alumni/ae donations, and rejecting a reasonably well-qualified legacy applicant can not only end support from the parent(s) but also damage relations with other alumni/ae.

To the extent that earlier cohorts were almost entirely wealthy and white and current cohorts are mostly wealthy and white, legacy preferences serve to reproduce the high-income/high-education/white profile that is characteristic of these schools. At the 18 schools for which we have legacy data, 50 percent of legacy applicants come from the top quartile of the income distribution, compared with 39 percent of non-legacy applicants. Of course today’s legacy pools include children of minority graduates, but this is a relatively recent phenomenon. Today, just 6.7 percent of legacy applicants at these 18 institutions are underrepresented minorities, compared to 12.5 percent of all other students.

The issue of fairness (equity) involved in legacy preferences was thrust into prominence in the context of the Supreme Court’s consideration of minority preferences at the University of Michigan. Even though a legacy preference does not raise the same kinds of constitutional questions as a racial preference, the two kinds of preference have become intertwined politically. Indeed, Justice Stephen Breyer asked explicitly about the relationship between the two types of preference, and the university systems in both Georgia and California prohibited legacy preferences after legal prohibitions ended racial preferences. Theodore M. Shaw, a lawyer for the National Association for the Advancement of Colored People’s Legal Defense Fund, posed his question this way: “What does legacy preference do to advance fairness and merit?”

In assessing the actual extent of legacy preference today, it is important to use the right numbers. Legacy applicants have stronger academic credentials than other applicants and would be accepted at higher rates if there were no legacy preference at all. Comparing legacies in competitive admissions situations with other candidates who present similar credentials gives a better sense of the extent of the true “break” legacies actually are given.

Dean Fitzsimmons notes that at Harvard “legacy status is basically used as a tie-breaker between comparable candidates,” and our data confirm that legacy preferences are generally reserved for candidates with strong credentials. When we assign the applicants to the 13 colleges and universities for which we have sufficient data on legacies to three SAT categories, we find that the legacy advantage is much more pronounced in the highest SAT range than elsewhere. The adjusted admissions advantage for legacies with combined SAT scores below 1100 is just over 6 percentile points (and not statistically significant), the advantage for those with SAT scores between 1100 and 1300 is 18 percentile points, and the advantage for those with SAT scores of 1300 and above is 25 percentile points.

It is important to recognize how much change there has been since the mid-1960s. At the time Brewster and Clark undertook to reform the Yale admissions process, more than two-thirds of alumni sons who applied to Yale were accepted, even though, we are told, “a disproportionate number flunked out or were placed on probation.” In those days, preferences for alumni children were deeply connected to privileges of many other kinds having to do with family background, social class, prep school education, and the like.

Defenders of legacy preferences as they are conferred at present argue that the admissions bar is still a high one, that relatively small numbers of applicants benefit from legacy preferences, and that the financial gains to colleges and universities from giving some weight to alumni/ae status are relevant. In addition, many colleges and universities also pay special attention in the admissions process to other candidates whose parents or other relatives are likely to make large donations, sometimes known as “development cases” or candidates on “the president’s list.” As John McCardell, the former president of Middlebury College, puts it: “If a handful of slots go to deserving applicants whose families can at least have the potential to improve in dramatic ways the quality of the education at Middlebury College, we would not be fair to our successors or predecessors if we were to overlook that reality.”

There is a clear trade-off between the goals of equity (which are not advanced by preferences of this kind) and excellence (which does depend on generating resources from alumni/ae and other donors). But to make the excellence argument, it is necessary that legacy/development preferences be conferred with something approaching surgical precision, that great care be taken in deciding how much of a “break” to give which candidates, and that the numbers not be too large.

We favor placing tight limits on preferences related to legacy status—and even tighter limits on those related to family wealth—because of their adverse impact on equity objectives, including the need for more, not less, socioeconomic diversity. But we do not think that ideological arguments against all such preferences should be allowed to overwhelm the legitimate interests of colleges and universities in maintaining a real sense of historical continuity on their campuses and in attracting the resources that they need to pursue their excellence objectives. (Another highly relevant finding is that legacy matriculants exhibit very little underperformance—1.4 percentile points in class rank.) More generally, we believe that all forms of preference, and all policies concerning tuition and financial aid, need to be examined together to see if an appropriate balance is being struck in pursuing the complementary goals of equity and excellence.

•Whereas minority admissions preferences serve educational and societal purposes, and legacy/development preferences serve institutional purposes tied to the pursuit of educational excellence, the preferences given to recruited athletes serve primarily the narrower purposes of the institution’s athletic establishment and the interests of trustees and alumni/ae with strong feelings about athletics. Athletes are recruited so that college teams can compete with teams from other colleges that are also recruiting athletes. There is no evidence that athletic accomplishments have any significant effect on giving or on how alumni/ae in general view their institution (actually, a majority favor placing less emphasis on intercollegiate athletics).

Thought of in the right way—as one component of an overall educational experience—athletics can be a valuable learning experience and a fine addition to other aspects of college life. Many of us know from personal experience that intercollegiate competition can be great fun, a healthy counterbalance to long hours in the library or laboratory, and, in all respects, a valued complement to academic pursuits. Regrettably, however, college sports in their current form represent a distinct threat to academic values and educational excellence. Recent research reveals all too clearly the ways in which intercollegiate athletics conflicts with core academic values at many of the country’s most selective colleges and universities that do not offer athletic scholarships.

Today recruited athletes, men and women alike, at the Ivy League universities and at leading liberal arts colleges, enjoy a huge admissions advantage. Recruited athletes are less and less representative of their classes (clustering in certain fields of study and bunching heavily in the bottom third of the class), they underperform academically (earning much lower grades than they would be expected to earn on the basis of their high school credentials), and they are far more isolated socially from their classmates than were their predecessors.

It is also clear that recruited athletes as a group do not contribute to racial or socioeconomic diversity. As [William] Bowen and Sarah Levin point out in Reclaiming the Game, “Recruited athletes in the schools in our study are in general appreciably less likely than students at large to be from underrepresented minority groups.” Furthermore, just 6 percent of recruited athletes enrolled at the institutions in our study are from the bottom income quartile, compared to 12 percent of other students. In short, giving admissions preferences to recruited athletes does little if anything to promote the equity goal of higher education (and may, in fact, detract from it) at the same time that it raises troubling issues for the pursuit of educational excellence.

In sum, the drift into recruitment-driven, high-intensity athletic programs at many of the country’s leading educational institutions, while not perhaps a “fatal flaw,” does indeed compromise the pursuit of excellence. It does so directly, by distorting priorities and misallocating resources (including highly coveted places in the entering classes of the most selective institutions). And it does so indirectly, by raising questions about our core commitments—about what matters most in crafting a class and in defining a campus ethos.

•Although they are not defined by demographic characteristics or by their interest in a particular activity such as varsity sports, early decision applicants can also be considered a “special group” from the standpoint of admissions preferences. Our data confirm the findings of Christopher Avery, Andrew Fairbanks, and Richard Zeckhauser [in The Early Admissions Game: Joining the Elite]: students who apply through early decision programs enjoy a decided admissions advantage. We find that applying early to one of the 13 institutions for which we have data, provided (in 1995) an adjusted admissions advantage of 19.6 percentage points. This is a striking statistic, and we must remember that the early application “craze” did not really start until 1999 or 2000.

What is most troubling about this form of preference is that it rewards applicants who are fortunate enough to know that applying early can benefit them substantially and who attend secondary schools, or are from families, with sufficient sophistication to know how to meet the early application deadlines. This group includes students at top-tier secondary schools, those with close connections to the college or university to which they are applying, and recruited athletes. Also, the very nature of early decision programs—which require students offered admission to matriculate at the school in question—makes it impossible for students who have financial need to compare financial aid offers from various schools in which they are interested. Minority students are of course disproportionately represented in the “need-sensitive” category.

These programs do not support the equity objectives that concern us in this study. [Nor can they] be said to promote the excellence objective of higher education. They are attractive to particular schools because they can help them recruit candidates who might otherwise have gone to a competitor. In addition, it is said that early decision programs help an institution improve its ranking in the U.S. News & World Report. But these institutional benefits seem far from compelling, especially when one counts the equity costs and takes a system-wide perspective.



Against this backdrop of evidence regarding the reasons for and effects of admissions preferences bestowed on various groups of applicants, for reasons that seem to us compelling in some instances and unpersuasive in others, we return to the central question: how should students from poor families and from families with no previous history of college attendance be treated in the admissions process? If we were to poll presidents of leading colleges and universities, we are confident that a large majority would want their schools to be more than “neutral” in considering such candidates. The brief submitted to the Supreme Court in conjunction with the University of Michigan cases by Harvard, Brown, Chicago, Duke, Dartmouth, the University of Pennsylvania, Princeton, and Yale is unequivocal on this point. The relevant section reads as follows:

Admission factors begin, of course, with the core academic criteria, including not just grades and test scores but teacher recommendations and state, regional, national, and international awards. In some cases, those criteria will be all but decisive, either positively (very rarely) or negatively (more often). In the vast majority of cases, however, they are not themselves decisive, and the process continues. Admissions officials give special attention to, among others, applicants from economically and/or culturally disadvantaged backgrounds, those with unusual athletic ability, those with special artistic talents, those who would be the first in their families to attend any college, those whose parents are alumni or alumnae, and those who have overcome various identifiable hardships. (Our emphasis.)

What is striking is the juxtaposition of this clear statement of intent with the equally clear empirical finding that, at the schools in our study, there is absolutely no admissions advantage associated with coming from a poor family and only a very small advantage (about 4 percentile points) associated with being a first-generation college-goer; at least that was the case for the ’95 entering cohort. We believe that this disjunction is due to a combination of a lack of data at the institutional level on the background characteristics of all applicants (data that are hard to assemble), financial aid constraints, a commitment to being need-blind (and therefore purposefully ignoring SES), and the lack of vocal champions for students from low-income families when tough choices are being made among a large group of very well-qualified candidates.

This last point deserves emphasis. The president of one of the universities in our study told us that he was not at all surprised by the finding that admissions at these highly selective institutions is truly need-blind, but not more than that. He said that when the admissions staff was considering an outstanding soccer player, it was as if “lights went on” in the room; everyone paid close attention, and everyone knew that the coach and athletic director were, in effect, watching closely and would have to be dealt with. Similarly, when the staff considered a legacy candidate with strong ties to the university, it was clear that representatives of the alumni office, the development office, and perhaps even the president’s office, were “present” in spirit if not in person. Minority candidates also were considered carefully and sympathetically, in part because of the active involvement of recruiters and admissions staff members with a special commitment to racial diversity. But, the president continued, when an otherwise “normal” applicant from a family of modest circumstances was considered, the process just moved right ahead without anyone making a special plea—perhaps, at least in part, because the immediate institutional interests to be served are less evident, or more diffuse, in the case of socioeconomic status.

[We do not] regard the status quo as acceptable. There are four reasons that we favor giving some preference to well-qualified applicants from modest backgrounds.

•First, this group of students is poorly represented in the most selective institutions [see page 50; for example, only] 3 percent of the students are both first-generation college-goers and from low-income families, whereas the national share of the same-age population in this category is nearly 20 percent.

•Second, enrolling more students from economically as well as racially disadvantaged backgrounds would unquestionably promote social mobility and be seen as representing a school’s serious commitment to enhancing opportunity in America.

•Third, it is essential to remember the odds that students from modest backgrounds had to overcome in order to get into the credible applicant pool in the first place. Students from families in the top income quartile were more than twice as likely as students from the bottom income quartile even to take the SAT. Among the test-takers, the odds of scoring over 1200 were three times higher for those in the top quartile as compared with those in the bottom quartile. Combining these probabilities, the key fact is that the odds of both taking the tests and doing very well on them were roughly six times higher for students from the top income quartile than for students from the bottom income quartile. Family circumstances have an enormous impact on the chances of applicants’ even being considered by the admissions staff at an academically selective college or university. Those who have “bucked the odds” to get into the credible pool would surely seem to deserve special recognition.

•Fourth, enrolled students who come from modest backgrounds do very well academically; in contrast to both recruited athletes and minority students, this group of students does not underperform.

For all of these reasons, relying simply on a traditional need-blind approach to admissions does not seem to us to be sufficient. Are the claims of equity really being met [when] prospective students are so stratified by SES in their pre-college years? [Although] multiple strategies can be used to improve the pre-collegiate preparedness of these students, we see no reason to wait (perhaps a very long time) for such strategies to make a real difference. Meanwhile, we believe that colleges and universities could take additional steps now to increase socioeconomic diversity on their campuses and to enhance social mobility in America.

To be sure, it is necessary to maintain perspective and not exaggerate the nationwide consequences of enrolling more students from low-income families at schools of the kinds included in our study. As one commentator put it, “Even if Harvard were tomorrow to get its share of low SES students up to the share of the overall population, the national SES gaps in enrollment and educational attainment would remain very much what they are today.” Other policies, at all levels of government and at all levels of the educational system, are needed if major progress is to be made. But we should not allow the enormity of the overall problem to justify inaction at the “local” level.



Those of us who favor encouraging academically selective colleges and universities to do more to enroll students from modest family backgrounds need to examine the implications of adopting specific, realistic alternatives to the traditional approach that combines need-blind admissions and need-based aid. The most direct alternative is simply to “put a thumb on the scale” when weighing the qualifications of applicants from lower-SES categories, much as colleges do now when they consider minorities, legacies, and recruited athletes—an approach sometimes called “class-based affirmative action.” But what kind of thumb—and how heavy a thumb? What would be the effects of such an approach on the composition of the entering class (including the number of minority students), the academic qualifications of enrolled students, and financial aid requirements?

To provide tentative answers, we have simulated the effects of one particular thumb type and size for the schools in our study. We decided to see what would happen if students with family incomes in the bottom quartile were given the same admissions advantage, within each SAT range, now enjoyed by legacies. Any number of other “hypotheticals” could be tried out, and any school interested in doing more for applicants from low-income or first-generation families would have to devise its own approach.

The idea of a “legacy thumb” appeals to us in part because there is a nice kind of symbolic symmetry associated with it. Why not give the least privileged group of applicants the same advantage conferred on those whose parentage has given them a special place in the competition for admission? [And as] we pointed out earlier, the general practice is to give legacies a significant boost only after they have done very well on their own in building a strong academic record, and to give those with stronger records bigger boosts. There is much to be said for giving special consideration to applicants who have demonstrated considerable achievement on their own. Using the actual admissions advantages currently enjoyed by legacies as a rough metric has the further advantage of emphasizing that, within each SAT interval, holistic assessments of all applicants would continue to be made—as opposed to more or less automatically taking all applicants from lower-SES backgrounds whose credentials placed them over some threshold.

What do we find when applicants from low-income families receive the same admissions advantage as legacies (and underrepresented minority groups also retain their current degree of admissions advantage)? The admissions probability for such candidates at the schools in our study could be expected to increase substantially—from 32 percent at present to 47 percent—which is, coincidentally, essentially the same as the admissions probability for minority applicants. The admissions probability for non-minority students from the bottom income quartile would increase even more dramatically—by almost 20 percentage points, from 30 percent to nearly 50 percent. Especially interesting is the effect on the admissions probability for all other applicants: it falls, as it would have to, but only by about 1 percentage point—from 39 percent to 38 percent. The explanation is, of course, the relative sizes of the applicant pools.

Turning to enrollment, our simulations show that the share of the class comprised of students from low-income families could be expected to increase from 11 percent to about 17 percent. The minority share would (by assumption) stay constant at just over 13 percent, and the share of all other students would decline from 79 to 74 percent. (These percentages add to more than 100 percent because there is an overlap between students from low-income families and minority students.) A side-effect of this kind of shift in admissions policies might be to reduce somewhat the number of students from wealthy families at the most selective institutions and, in effect, redistribute some of them to other liberal arts colleges and universities—which might be healthy all around. There is another potential effect: giving a “boost” to applicants from low-income families might encourage more such applicants to apply to highly selective schools.

There are three kinds of potential “costs” associated with this hypothetical “legacy thumb” policy. First, there could be effects on the academic profiles of the institutions and on academic outcomes. It would not be unreasonable to worry about some reduction in average SAT scores if high-scoring students from high-SES backgrounds were replaced by less well-prepared students from low-SES categories. But in fact, in our simulation the average SAT score of the members of the entering class remains essentially unchanged. A large part of the explanation is that at present large numbers of high-testing students from low-income families are being turned down; they are suffering the same fate as many other disappointed candidates with high test scores who applied to these selective institutions. Another reason for the modest impact on SAT scores is that the legacy admissions advantage is greatest at the higher reaches of the SAT distribution (and is virtually non-existent at the lower end). Thus, relatively speaking, it is the students with high SAT scores from low-income families who would benefit disproportionately from receiving legacy-type admissions preferences.

Second, there would be significant dollar costs involved in providing financial aid to a larger number of needy students (and perhaps some additional institutional expense for support services). We estimate that for our group of private liberal arts colleges, with an average of 500 students per class, grant aid funds would have to increase about $460,000 per class, per year, or just under $2 million for all four classes per year (an approximately 12 percent increase), if current financial aid policies were maintained. For the private universities in our study, with 1,500 students per class, the necessary increase could be expected to be about $1.4 million per class, per year, or between $5 and $6 million for all four classes (which is also approximately a 12 percent increase).

The third potential cost is also related to institutional finances. These institutions rely heavily on alumni/ae donations, and it is reasonable to ask whether enrolling a substantially higher number of students from low-SES backgrounds would result in less wealthy or less generous alumni/ae. Analyzing the raw data, we found that alumni/ae who came from the bottom income quartile in 1976 were somewhat less likely than their classmates to have made a donation in the nearly 20 years that they had been out of school. Former students from the bottom income quartile gave at a 38 percent rate, compared to rates of 50 to 59 percent for those from the two middle quartiles and a 62 percent rate for those from the top quartile.

The decision to make a gift to one’s former school is influenced by many factors, such as one’s experience at school, one’s general attitude toward charitable giving, one’s responsibilities to family members—and of course one’s income and wealth. Controlling for gender, race, SAT scores, college grades, employment status, sector of employment, and income, we find that the giving gap between alumni/ae from the bottom income quartile and other former students disappears. Thus we find that enrolling more students from low-income families could be expected to diminish alumni/ae donations, but because of characteristics other than socioeconomic background per se, especially choices of sectors in which to work. In other words, there is no “future donations” trade-off between similarly situated alumni/ae from different socioeconomic categories. However, we know that alumni/ae from low-SES backgrounds will not be, in general, “similarly situated.”

We understand the need to live within what are often tight institutional budget constraints, and we certainly do not mean to suggest that putting a thumb on the scale would be a painless process for any institution.

Still, substantial as these financial costs may be, it is easier to imagine moving some distance in the direction of admitting students from lower-SES backgrounds by means of admissions policies—as far as resource constraints would allow—than it is to believe that the powerful pre-college “conditioning” of the pool of applicants can be fixed any time soon. We are strongly in favor of doing whatever can be done to enhance college preparedness for students from disadvantaged backgrounds, but that will be a long and difficult process. Meanwhile, there is a strong argument in favor of doing more at the undergraduate level to help some substantial number of deserving students almost immediately.

Institutions interested in doing more for potential students from low-income families may also elect to provide even more generous financial aid to these students than they do already. Harvard and the University of Virginia have both elected to replace loans with grants for students from lower-income families. We hope that these initiatives succeed. However, we suspect that giving a boost in admissions may be more effective in altering the socioeconomic composition of classes than improving financial aid offers to those who are admitted. The relatively high yields on current offers of admission to applicants from low-income families (higher than the average yields on all offers of admission) suggest that present-day financial aid policies are less of a problem than is reliance on need-blind admissions. It is fine to be more generous—and being more generous certainly could increase yields further and help with recruitment—but we think consideration should be given to the alternative of putting a thumb on the admissions scale (maybe even a thumb and a half). And if any institutions are going to take the lead in moving beyond need-blind admissions, it is almost certainly going to have to be the wealthy, academically selective institutions that we have studied.           


William G. Bowen is president of the Andrew W. Mellon Foundation and president emeritus of Princeton University. With Harvard president emeritus Derek C. Bok, he wrote The Shape of the River: Long-Term Consequences of Considering Race in College and University Admissions. Martin A. Kurzweil, research associate at the foundation, is a student at Harvard Law School. Eugene M. Tobin is program officer for the Liberal Arts Colleges Program at the foundation, and former president of Hamilton College.

You might also like

Using the Law for Good

2024 Radcliffe Medalist Sonia Sotomayor on civic engagement and optimism

Equality and Justice

A Radcliffe Day panel discusses pluralism and progress. 

Close Call

Ending a tumultuous year, Harvard tradition is served in the 373rd Commencement—with plenty of thunder from the stage.

Most popular

Harvard Discloses Administrator and Investment Manager Compensation

The annual release on leaders’ most recent pay

Harvard Corporation Rules Thirteen Students Cannot Graduate

Faculty of Arts and Sciences May 20 vote on protestors’ status does not confer “good standing.”

AWOL from Academics

Behind students' increasing pull toward extracurriculars

More to explore

Bernini’s Model Masterpieces at the Harvard Art Museums

Thirteen sculptures from Gian Lorenzo Bernini at Harvard Art Museums.

Private Equity in Medicine and the Quality of Care

Hundreds of U.S. hospitals are owned by private equity firms—does monetizing medicine affect the quality of care?

Sasha the Harvard Police Dog

Sasha, the police dog of Harvard University