From Cambridge to Berkeley, from Ann Arbor to Austin, college administrators and faculty are scrambling—and it’s not just because the Trump administration is slashing billions of dollars from their budgets and calling them out on their violations of the rights of Jewish students. They are also still reeling from the Supreme Court’s decision two years ago to ban racial preferences in college admissions.
To say that racial diversity on campus has been the central preoccupation of elite institutions of higher education—their collective mission, in other words—would be an understatement. It begins with the rainbow aesthetic of the dozens of catalogues that all high-school students receive in the mail once they make it clear they are college-bound. Having taken literally dozens of tours of elite colleges in the past two years with my children, I can report that grand claims about the glories of campus diversity make up about a third of every school’s presentation.
College administrators and faculty have not only convinced themselves that a sufficient quantity of black and Hispanic students was necessary to the mission, they won over parents and students as well. So convinced have they all been over the past two decades that they engaged in a kind of mutual quiet conspiracy to gloss over the fact that the makeup of a campus’s population is only an aspect, and really a minor aspect, of the fact that families are forking over tens to hundreds of thousands of dollars for the purpose of getting an education.
In the pursuit of the larger mission, the admission in large numbers of those kids whose backgrounds deemed them contributors to campus diversity necessitated elevating their acceptances over those of other kids. In 2023, the Supreme Court heard a case in which Asian American plaintiffs sought to prove they had been treated unconstitutionally by America’s most elite institution because they had been disfavored as a group—diverse, to be sure, but the wrong kind of diverse. That case was Students for Fair Admissions v. Harvard. The high court ruled that the admissions practices at the school were a patent violation of the equal-rights clause of the 14th Amendment. As Chief Justice Roberts wrote in the majority opinion, “The guarantee of equal protection cannot mean one thing when applied to one individual and something else when applied to a person of another color.” He said flatly, “Eliminating racial discrimination means eliminating all of it,” and that ended affirmative action as we knew it.
Or did it?
The demographic makeup of the class of 2028—the first admitted after the Court’s decision in 2024—suggests that at least some colleges were playing games rather than obeying the Court’s edict. First, consider a school that did seem to change its approach. At MIT, the share of black students dropped from 15 percent to 5 percent, while Hispanic students fell from 16 percent to 11 percent. This happened because MIT stopped weighting minority applications more favorably when it came to evaluating them against those of students who were neither black nor Hispanic. Now check out Princeton: There was no change in the racial composition of its 2028 class. At Harvard, which was the litigant in the case, black students dropped from 18 percent to 14 percent of those admitted, but the share of Asian American students—whose unequal treatment by Harvard had been the basis for the lawsuit in the first place—remained the same. At Yale, the percentage of black students remained the same, but there were more Hispanic students. And at Duke, the percentage of both black and Hispanic students rose.
During the litigation that led to the Supreme Court decision, university leaders claimed that overturning racial-preference policies would make it all but impossible for schools to provide a diverse environment. Then it happened, and at the schools mentioned above, very little changed. How could that be, given the fact the evidence collected in the case showed a gap of more than 250 points in average SAT scores between blacks and Asians at Harvard? Was that gap closed in a single year? Would anyone believe such a thing?
So how are these schools now succeeding at providing the kind of racial diversity to which they are committed as a form of civil religion but without taking race into account—which, under the Supreme Court’s ruling, they are no longer allowed to do? Some schools are claiming they can do it through better recruitment. But if that were the case, they probably would have done it already. The bald fact is that there are simply not enough black and Hispanic students with the academic records necessary to populate the most elite schools without someone putting a thumb on the scale. And the schools that are still playing fast and loose with race in admissions will be found out soon enough when the courts make them open up their records again to see if they are complying with the law.
The second option floated by many university leaders is to sort by class rather than race, meaning they can offer admission to more socioeconomically disadvantaged students. “What Levels of Racial Diversity Can Be Achieved with Socioeconomic-Based Affirmative Action?” This was the question asked in a 2018 paper by Sean Reardon, a professor in the Graduate School of Education at Stanford University, and two of his colleagues from UC Irvine and the Brookings Institution. They set out to understand whether it would be possible to find workable, racially neutral admissions policies that would still achieve a form of educationally beneficial racial diversity.
The authors constructed a model of such a policy in selective colleges and universities using two different strategies—giving students a leg up if they come from a lower-income background, or engaging in targeted recruitment of applicants by race. They found that even tipping the balance in an extraordinary fashion using socioeconomic status alone—600 points on the SAT, more than double the benefit Harvard afforded black students versus Asian American students—achieved almost nothing when it came to achieving greater racial diversity.
This is not surprising. Similar real-world experiments at UCLA Law School and another study in Texas found that taking a student’s economic background into account was not enough to achieve the same kind of racial diversity that explicit racial preferences created. As the authors note—perhaps stating the obvious— “race-based affirmative action leads to racial diversity because it can select directly the students who will contribute most to racial diversity.” In order for class preferences to achieve the same kind of results, they write, there would have to be a “strong relationship between [class] and race.”
It turns out there is not.
For while the black poverty rate is twice as high as the white poverty rate, there are still many more poor white people in America than poor black people—about two and a half times as many. In other words, an admissions officer who wanted to admit more black students using only the socioeconomic status of the applicant would not have much success. Thus, in 2018, the authors concluded that although race-conscious affirmative action is “imperfect,” it is the only workable strategy if your goal is to have more black students.
_____________
So what is a college administrator to do, since class-based preferences aren’t going to get you there and the Supreme Court has said you cannot use race-based preferences?
Well…what if there were a way to find out a student’s race without asking for it? What if you could kind of quietly figure out whether a kid was black and use stealth to achieve your aims? You’d need some pretty sophisticated help to get there.
Enter the College Board, the nonprofit behemoth that administers most of the nationwide educational testing in the United States. The PSAT, SAT, and Advanced Placement tests given to high-school students are all devised, produced, and scored by the College Board, which generates a staggering $11 billion a year in revenues from them.
In 2019, the College Board was all set to launch what it called the Environmental Context Dashboard. The idea was to score two categories into which all students fell, on a scale from 1 to 100, and then average them. One score would be for a student’s school environment. The other would be for the student’s neighborhood. These were supposed to give colleges a score to tell them what level of obstacle a student had overcome, and it came to be known as the “adversity score.”
This deterministic conceit rubbed many people the wrong way. Even people who wanted schools to engage in some kind of preferential treatment for students from disadvantaged backgrounds were upset. Jessica Pliska of The Opportunity Network, “which works with students from underrepresented communities on their paths to college and career,” wrote in the New York Times: “Let me get this straight: We’ll rely on stereotypes that stigmatize students by assigning them a score based not on their own experiences, but on contextless, aggregated data about crime and poverty based solely on location? We’ll make gross assumptions across race, class and community in evaluating a student, an actual human, regardless of personal experience to determine a score they will never even see?” And in the face of the backlash, the College Board backed off and said it would just go by the SAT. The organization’s CEO, David Coleman, told the New York Times that they had been wrong to “retreat from the notion that a single score is better. So in that sense, we’ve adopted a humbler position…. The College Board should keep its focus on scoring achievement. We have acknowledged that we have perhaps overstepped.”
Except it didn’t retreat. Not at all.
In place of the “adversity score,” the College Board launched Landscape, which it says “provides consistent high school and neighborhood information for all applicants to help admissions officers fully consider every student, no matter where they live.”
This “tool” for admissions officers does not create any kind of numerical score. Rather, it gives college admissions officers access to the kind of information that would have been used to create the adversity score. In deliberately obfuscatory language, the College Board says Landscape helps “provide de-identified applicant information to generate contextual information in their Landscape instances” to colleges. Students and parents do not have access to the information about the kid, as they do with a SAT score. It is kept privileged. Landscape has been offered free of charge to any university that wants to use it.
The fact that the public has deliberately been denied access to information collected and disseminated about a student applying to a college—even if that student is somehow “de-identified”—raises profound concerns about exactly what kind of information is being collected and how it is being used. It might seem as if the College Board’s decision to provide information on a variety of different factors offers schools more to go on than the SAT, but it can also provide colleges with a way to place more weight on certain kinds of “adversity” to give schools the admissions outcomes they want.
Some of the information provided by Landscape is easily accessible and publicly available. For instance, the tool will tell admissions officers useful things about a student’s high school, like the percentage of students receiving free or reduced lunch, information about average SAT scores, or the percentage of students who take AP exams. Parents and students are able to see this information for themselves from sources such as Great Schools or programs like Naviance.
What else can the Landscape tool tell admissions officers? It has information about a student’s neighborhood, based on his individual census tract. It’s important to know that census tracts are very small—much smaller than zip codes or even what most people would think of as a neighborhood. As demographer Nicholas Eberstadt tells me, “It’s possible to have flawed assumptions about the individuals within the group,” but with something as small as a census tract, “you are less likely to have big errors.” In other words, you can pretty accurately draw a sociological portrait of a kid living in a specific census tract, even without a name.
And here is where we can start to reasonably wonder what the College Board is up to. After a family’s level of college attendance, the methodology page for Landscape on the College Board’s website lists “household structure” as the next factor in the equation. It provides “a measure based on neighborhood/high school information about the number of married or coupled families, single-parent families, and children living under the poverty line.”
Enter racial profiling. Now, while there are many more children living under the poverty line who are white than black, black poverty tends to be more concentrated. If you find rates of poverty in a census tract that exceed 30 percent, chances are that you are dealing with an urban area with a higher concentration of blacks. “Blacks are the most segregated racial group in America,” Scott Winship of the American Enterprise Institute tells me. “Poor whites are just less likely to live around other poor whites.” While measuring poverty at the level of census tract may seem like a clear proxy for a student’s economic situation, it can clearly also help narrow down the likelihood that a student is from one racial background or another.
For anyone who is even slightly familiar with demographic differences by race, using single parenthood as a factor is a way of homing in on something. In 2022, about 24 percent of white children were living in a home with only a mother, compared with 63 percent of all black children. Thus, the higher the percentage of households in a census tract that are single-female-led, the greater the likelihood that its residents are African American. Hispanic families have lower rates of marriage than do white families, but still 42 percent live with two parents.
Another factor listed in the Landscape tool is “housing stability.” This is described as “a measure based on neighborhood/high school information about vacancy rates, rental vs. home ownership, and mobility/housing turnover.” While it may seem intuitively as if this is a race-neutral factor—we associate living in the same home for years or decades with financial stability—the truth is that people in urban areas are much more likely to change residences, and they are much more likely to rent as opposed to own their homes.
If you find a census tract where income is low and a high percentage of people are renting, you are also more likely to be looking at an application from a black student. To put it simply, the white student from West Virginia may be living in a trailer park, but his parents probably own the trailer, and they might have stayed in the same trailer for a long period of time.
Landscape also includes “median gross rent as a percentage of household income.” Urban households are more likely to spend more of their money on housing, and the places with the most expensive housing costs are large cities with significant populations of poor black and Hispanic families—New York, Los Angeles, Washington, D.C., etc. Again, this is not to say that there are not plenty of poor white people living in these areas, but poor whites are more likely to be concentrated in rural areas including Appalachia, the Upper Midwest, and the Western Mountain states. Like the others, this criterion seems implicitly designed to surface minority candidates.
A similar dynamic is true when it comes to crime. Landscape considers “the predicted probability of being a victim of a crime in the neighborhood or neighborhoods represented by the students attending the high school.” Urban crime is much more common than rural crime. USA Facts notes that “in 2021, in urban settings, 24.5 out of 1,000 people aged 12 or older reported being the victims of violent crimes, and 157.5 reported being the victims of property crimes. In rural settings, those figures were 11.1 and 57.7, respectively.” Incorporating crime rates by census tract into any admissions decisions, even controlling for income, will likely favor black students.
This will also occur when you include the “percentage of households receiving food stamps/SNAP.” According to an analysis published in the journal Public Health Nutrition, “Of the 89,048 household-by-quarter observations, 15,613 or about 17.5 percent were current SNAP participants. Participation was highest among black households (26.6 percent), followed by Hispanic households (21.0 percent) and white households (16.4 percent).” SNAP is more likely to be used by single-parent households, and, as we’ve seen, black applicants are more likely to come from single-parent households, so using this factor may also increase the likelihood of selecting a black applicant.
Particularly when used with a large sample of applicants, the Landscape tool could clearly be a means of steering admissions in a particular direction without anyone ever having to acknowledge they have a hand on the tiller.
_____________
The factors listed as being incorporated into the Landscape tool provide plenty of fodder for admissions officers looking to increase black admissions rates, but there are also other seemingly “race-neutral” factors that could be added should the College Board decide to tinker more. These include infant-mortality rates, which are higher among black families. This may be harder to get for a census tract because there are so few of these incidents that such data might unintentionally reveal personal information.
Other possible factors that could be used include a simple look at population density. Though it would not be immediately obvious that living in a more densely populated area would create adversity for an applicant—some of the wealthiest applicants live in the densest urban environments—it would certainly achieve the goal of getting more urban applicants admitted than rural ones. Living in a one- or two-bedroom dwelling might be included as a sign of adversity. Not having a car could also be a factor. Few poor families in densely populated urban areas own cars, while most households in poor rural areas have one by necessity.
It’s also interesting to see which factors easily gleaned from Census data are not included in the algorithm—factors that might have favored rural applicants if they had been used. Not having access to running water, an indoor toilet, or kitchen appliances could have been used, but it does not seem to be a factor in the Landscape tool. Such amenities are always present in urban environments, even in the poorest neighborhoods, but not in some rural ones.
Similarly, the lack of a nearby medical facility is not included. Poor areas in large cities all have nearby hospitals and clinics; small towns do not. If the tool considered a lack of access to public transportation a factor in adversity, that would favor rural rather than urban applicants, but it does not come into play.
Finally, it is interesting that among the school-related adversity factors, per-pupil expenditure for the public schools gets no consideration. Wouldn’t a school that lacked funding be something that could hold a student back? Well, as it happens, per-pupil expenditures in large cities are much higher than in poor rural areas, thanks in part to the hold teachers’ unions have on city government.
There are other reasons to be suspicious of the true purpose of Landscape. “Can better information on applicants’ backgrounds increase socioeconomic diversity at selective colleges?” This was the question asked by a group of scholars in August 2022 regarding its effectiveness. The authors, a professor at Georgetown and three researchers at the College Board itself, concluded that it actually had limited effectiveness for that purpose.
Looking at admission decisions of 43 selective institutions for over 3.7 million applicants who applied to attend in the 2014–2015 to 2019–2020 school years, the authors found that “on average, applicants from the most challenging backgrounds experienced a five-percentage point increase in the probability of admission in the year of Landscape adoption.” Still, since the vast majority of students had “low-challenge” levels, the actual makeup of the pool of admitted students changed only modestly. Moreover, “the likelihood of applicants from higher-challenge backgrounds enrolling at pilot institutions did not change after Landscape was used in the admissions process.”
For years, advocates had been arguing that colleges needed to find ways to enroll more economically disadvantaged students. Evidence has been mounting that kids from wealthier backgrounds are getting an advantage above and beyond what they might deserve based on their academic record—and almost everyone thinks they’re gaming the system somehow. But are they? Admissions officers surely know which students come from privilege, and they seem fine with it. Harvard’s Raj Chetty and his team found that three factors are at work in getting wealthier students a leg up—legacy admissions, more extracurricular activities, and the recruitment of athletes. If schools wanted to give more weight to adversity, all they would have to do is give less weight to all of these other factors. They don’t need the College Board to help them.
And if colleges desired true socioeconomic diversity, they could make the application process simpler. Using the Common Application as it was intended (as a single universal form rather than having many selective schools demand additional essays) would be a start. They could eliminate the essay altogether, since wealthier kids often have theirs edited extensively or written outright by parents, coaches, or college counselors. They could also ask the College Board to restore the essay section to the SAT, so they can be sure the student has written it by himself. Even the financial aid forms could be eliminated. A tax return from parents would tell colleges everything they need to know about a student’s financial situation. Home ownership, crime rates, household structure, and all of these other factors may be interesting, but they are not necessary considerations for getting more lower-income kids into selective colleges.
So one gets the sense that Landscape is a way of surfacing race-based information surreptitiously. And there might be others in the offing. An organization such as the Harvard-based Opportunity Insights, for instance, could help admissions officers look at “social mobility” to understand their applicant’s neighborhood environment better. Incorporating such a factor might give admissions officers a clue into the history of the census tract a kid comes from, because knowing the years in which social mobility declined could offer clues to the racial makeup of a neighborhood.
Recent research conducted by Opportunity Insights revealed that “between children born in 1978 and 1992, employment rates declined for low-income white parents compared to both low-income Black and high-income white parents.” Understanding the direction in which things are moving in a particular census tract and how they have moved in the past might tell admissions officers something about the race of its residents—and applicants.
There are many new sources of information about students’ backgrounds in data that are not publicly available but that colleges might be able to tap into as the years pass. Would companies from Silicon Valley, for instance, be interested in creating their own adversity scores? The amount of data they collect is staggering, and their ability to interpret it is becoming nearly infinite due to the application of artificial intelligence. Though the politics of high tech seem to be changing these days, one could see how a certain progressive mindset in the Valley would be excited to help. As Nicholas Eberstadt notes, “If you have a Niagara Falls of data, you can use that to do all sorts of things.”
Facebook, for instance, might be able to provide data based on census tract or zip code about the number of social connections a student may have. Research has shown that information about real-world social mobility can be approximated by virtual social connections. And Facebook has plenty of other personal information, including data about political affiliations and consumer purchases. And if Amazon decided to get involved in such a project, college admissions officers would hit the jackpot.
There is no evidence so far that these companies are cooperating with colleges to accomplish any goals related to admissions. One might expect a large backlash about data privacy if they did. But companies regularly sell customer data to other organizations, and as long as they are not providing any personal records—but simply aggregated information—it’s not clear they would be doing anything illegal by selling to schools. In any case, the real issue is not how the data are collected. It’s the way the data are used that should be triggering suspicion—and the very real sense that elite universities are doing whatever they can to stand, like George Wallace long ago, in the schoolhouse door to keep out the students they have deemed undesirables.
_____________
In May, the College Board emailed families about three National Recognition Awards, for which students could be eligible beginning in 2025. As Sara Harberson, an eagle-eyed college counselor, noted on her blog: “Two of the awards have been around for several years: Rural/Small Town Recognition Award and First-Generation Recognition Award. …Noticeably absent are the three racial/ethnic awards that have been available for years: National Recognition Award for Black Students, Hispanic Students, and Native American/Indigenous Students.” The College Board notes the absence of these prizes on its website by explaining that the “complex and evolving legal and regulatory environment” has “reduced the value of these awards.”
Between the Supreme Court’s 2023 ruling and the Trump administration’s crackdown on DEI through executive orders, you can see why the College Board acted as it did. But the $11 billion organization has in no way walked back its support of such racial-classification schemes. It’s doubtful that David Coleman and his colleagues have adopted a “humbler position,” as he put it back in 2019. They’ve just decided to adopt a cleverer position—one in which the College Board is colluding in the creation of a complex new system for schools to identify the race of a student without explicitly asking for it. University admissions offices know exactly what the College Board is selling. Now it’s time for the public, and the administration, and the Congress, and the courts, to hold them to account.
We want to hear your thoughts about this article. Click here to send a letter to the editor.
