“In the interests of putting the concept of ‘equality of educational opportunity’ into practice, [colleges] might want to consider abandoning altogether the use of grades and tests in admissions, and instituting instead a lottery system for choosing among their applicants.” So wrote the eminent education scholar Alexander Astin in a Science letter more than 50 years ago. Proposals for lottery admissions to college have surfaced countless times since then, appearing in the pages of The Atlantic, The New York Times, the Los Angeles Times, and The Chronicle, among other media outlets.
A key fact about nearly all these proposals is that they have been intended for America’s most selective institutions, not the nearly 600 four-year colleges and universities with open-door policies or the additional 700 with acceptance rates greater than 75 percent (as of 2018-19). A small number of institutions with acceptance rates below 25 percent — fewer than 100 schools as of 2018-19 — attract an enormous number of applicants, along with a disproportionate amount of media attention.
A recent proposal for an admissions lottery appeared in these pages, in Matt Feeney’s “The Abiding Scandal of College Admissions.” Feeney begins with a bracing critique of the status quo, hitting some deserving targets head on: continuously shifting admissions criteria, résumé padding, and the carefully crafted “authenticity” that is currently in vogue.
But Feeney, like many of his predecessors, makes a leap in logic: Because the current system is flawed and seems arbitrary in some respects, we should abandon any pretense of a rational process and simply institute a lottery. Other aficionados have argued that in addition to improving fairness by giving everyone an equal chance, lottery admissions would increase ethnic and socioeconomic diversity and reduce anxiety among candidates and their families.
This is a misguided view. I’m currently a researcher at ETS and first wrote about lotteries when I was a professor of education at the University of California at Santa Barbara. If you dig into the data as I have, you’d see that a lottery is not the refreshingly simple solution to college admissions that many of its proponents claim.
First, let’s consider what may seem like an obvious question: Would admissions lotteries give all applicants an equal chance? Not if they are created along the lines that most lottery backers have proposed. A fascinating aspect of the discourse on this topic is the degree to which proponents, having argued for the egalitarian nature of lotteries, then torture the lottery concept into a form they regard as palatable.
Thresholds specifying the minimum performance levels required for lottery eligibility are tacked on, along with unequal weighting for candidates and miscellaneous other variations and exclusions. Some advocates have suggested that highly desirable candidates be excluded from the lottery and simply be guaranteed admission. To make sure key student groups are represented, others have recommended that the lottery pool be stratified, with random selection taking place within each stratum.
Nearly every proposal has specified that in order to be eligible for the lottery, applicants would have to exceed a threshold. Feeney calls this a “sane approach [for selecting among the] glut of qualified applicants.” The key word here is “qualified.” Who are the “qualified applicants,” and how can we identify them?
Most lottery advocates suggest that the threshold for inclusion in the pool should be a minimum admissions-test score. Even the Harvard Law emerita professor Lani Guinier, a staunch opponent of admissions tests, has advocated the use of lotteries for applicants who exceed an admissions-test-score threshold.
In The Atlantic, Alia Wong proposed an even more elaborate threshold for a lottery that Harvard might consider. To qualify for the lottery pool, she suggests, Harvard could establish a set of criteria, “say, an SAT score of 1470 or above, a 3.5 or higher GPA, a demonstrable interest and aptitude in particular nonacademic activities, a record of overcoming obstacles, and so on.” Wong’s lottery includes not only a rigorous threshold but an option for weighting the candidates differentially. “To continue to promote diversity, the school could give extra weight to certain applicants depending on, say, their zip code, the kind of high school they attended, their income, and their race.”
Guinier too suggests that candidates could be weighted unequally in a lottery, to increase the likelihood of selecting “applicants with the skills, abilities, or backgrounds that are valued by the institution. For example, a weighted lottery could enhance the chances that certain students will be selected by putting their names in two or three times … the institution could weight the names of students who come from underrepresented communities or demographic groups, or students who make five-year public-service commitments to work in disadvantaged communities upon graduation.”
This begins to look familiar. Indeed, Robert K. Fullinwider and Judith Lichtenberg, authors of Leveling the Playing Field, joke that by weighting applications by factors like high-school grade-point average, class rank, and admissions-test scores, an institution could devise a lottery that would produce much the same entering class as its current admissions procedures.
The impulse to impose additional rules and constraints on the lottery reflects the discomfort that even its proponents feel about an entirely random process. Except in cases where we truly believe that no legitimate distinctions can be made among the candidates, we tend to shy away from the random allocation of resources. The admissions lottery once used at the University of Illinois at Urbana-Champaign provides a useful illustration. In 1970, the university’s College of Liberal Arts & Sciences admitted a portion of its freshman class using a lottery. More than 800 applicants, some of whom were top students in high school, were rejected, leading to widespread anger among candidates and their families. One mother of a rejected applicant lamented, “We didn’t know that there would be a lottery or we may have started looking for another school in the first place.” The public outrage was so extreme that the lottery results were ultimately rescinded and all the rejected candidates were admitted.
As the political scientist Peter Stone argues,
The random allocation of goods must always be justified in terms of a specific conception of justice. This conception would spell out precisely what gives a person a claim to a good, and what makes one claim stronger or better than another. A lottery would then make sense whenever the conception does not provide a basis for distinguishing between the members of a group of claimants.
A lottery is not impartial, Stone notes, if “there are relevant differences between claimants to a good.” Clearly, the University of Illinois applicants and their families strongly believed that such differences existed. A similar result could be expected if a random draw were used to select college classes today.
Now let’s consider what happens when a lottery with a threshold is used as a means of considering only “qualified” candidates. Are these lotteries an equitable alternative to conventional selection procedures? Are they likely to produce diverse and well-performing classes, while reducing competitive frenzies and debilitating anxieties? Three relevant studies suggest the answer to these questions is, at best, unclear.
The studies, all assuming moderate thresholds, come from a range of disciplines: one by the political scientists Bernard Grofman and Samuel Merrill, another by the economists Anthony Carnevale and Stephen Rose, and the third by myself. The Grofman-Merrill study was a theoretical analysis; the other two used a simulation approach based on national samples of applicant data. In my own study, the lottery was modestly effective at increasing the percentage of students from underrepresented ethnic groups, relative to other selection methods. The other two studies, however, found the lottery to be entirely ineffective for this purpose. Grofman and Merrill found that “a lottery-based system with a realistic minimum threshold will result in only a minuscule rate of minority acceptance compared to that of whites.” In addition, all three studies raised concerns about the academic quality of the selected classes. According to Grofman and Merrill, the likely consequences of a lottery with a threshold are “not nearly as attractive as they might first appear to reformers.”
In short, lotteries without thresholds — the University of Illinois scenario — carry the risk that unqualified candidates will be selected while impressively talented and accomplished ones are rejected. Lotteries with moderate thresholds do little for diversity and may produce classes that are not prepared for college work. Imposing extremely high thresholds may come close to producing a lottery pool in which candidates are equivalent in their qualifications, thus satisfying the conditions that Stone would consider appropriate for a lottery. But using a lottery with very high threshold will do nothing to broaden educational opportunity. In my study of a national sample of applicants to selective schools in 2004, a lottery that included candidates with test scores in the top 10 percent produced a class in which only 2.5 percent were members of underrepresented ethnic groups, and only 6.6 percent were in the lower half of the socioeconomic distribution. This, of course, is not the outcome desired by lottery proponents.
What about the familiar claims that these lotteries would reduce craziness, competitiveness, and candidate anxiety?
If the lottery’s thresholds or weights include the factors their proponents call for — high test scores and GPAs, commitments to public service, aptitude for nonacademic activities — it’s hard to see how candidates’ anxiety would be reduced. Applicants would be faced with deciding whether to invest energy in exceeding the threshold or improving their weights, knowing they might then be eliminated through the toss of a metaphorical dart. And would candidates be more content to be rejected by a lottery than by an approach that took account of their individual characteristics and contributions, however imperfectly?
Lotteries are seductive because they ignore “bad reasons” for admission — inflated résumés, wealth, celebrity, political connections, and legacy status, for example. But they are also blind to talent and genuine accomplishment and eliminate any kind of conscientious individualized evaluation of applications. As a former admissions dean noted in The Chronicle, “while college admissions may be perceived by many to be a crapshoot, no one gains if we actually turn it into one.”