Is Test-Optional College Admissions a DEI Policy?
Credit: Cheryl Senter / Bloomberg / Getty Images- A February letter from the Education Department warns colleges to eliminate all DEI activities and policies, including test-optional admissions.
- The assumption is that a test-optional environment results in a more diverse applicant pool and entering class.
- A new study from Dartmouth College claims its experience with test-optional admissions suggests exactly the opposite.
- Tools colleges use to obtain demographic information on applicants may or may not include race.
A Feb. 14 “Dear Colleague” letter from the Education Department has left higher education scrambling for answers.
Issued by Craig Trainor, acting assistant secretary for civil rights, the letter warns educational institutions to eradicate any vestiges of diversity, equity, and inclusion (DEI) programming.
In doing so, the department claims the ruling in the 2023 Students for Fair Admissions (SFFA) v. Harvard case, in which the U.S. Supreme Court banned race-conscious admissions, extends beyond admissions to include “all … aspects of student, academic, and campus life.”
Colleges were given 14 days to change their activities and policies or risk losing federal funding, including student financial aid.
Specifically addressing admissions, the letter says colleges cannot search for “cues” in an applicant’s essay to determine race, even though the Harvard decision suggests students can use essays to address their racial identity.
Institutions also cannot use “non-racial information as a proxy for race,” the letter continues, pointing directly at the elimination of standardized testing “to achieve a desired racial balance or to increase racial diversity.”
The assumption here is that test-optional policies favor underrepresented students, including people of color, by removing a barrier long assumed to disadvantage students from these backgrounds.
But is this true?
Dartmouth Study Examines Test-Optional Outcomes
Not exactly, says a new working paper from the National Bureau of Economic Research.
Authored by three Dartmouth College professors, the paper — a January publication that might be described as a prescient coincidence in light of the Feb. 14 letter — provides a rationale for why the college decided to reinstate standardized testing in 2024.
Like most colleges, Dartmouth went test-optional during the pandemic. It subsequently convened a working group, including these three professors, to determine if resuming test requirements was warranted. This new paper presents a more thoroughly researched, academic treatment of the working group’s report.
Standardized tests such as the SAT and ACT are more accurate than high school grades and teacher recommendations as predictors of academic success at Dartmouth, the authors found. This conclusion aligns with the results of a 2024 study by Opportunity Insights, a research group based at Harvard University, which examined data among Ivy-Plus schools from 2017-2022.
And that correlation between test scores and academic success remains strong across income and demographic groups.
Yet during Dartmouth’s testing moratorium, 46% of applicants requested that its admissions office not consider their scores (some students send scores when taking the exam but later choose the test-optional route).
It turns out this decision actually hurt many students from less-advantaged backgrounds. Why? Because many such students fail to submit scores that, considered in context, would improve their chances of admission.
The authors note that “test scores can allow lower-income high achieving students to stand out regardless of alumni connections, interviewing skills, ability to travel to campus, or the use of admissions counselors.”
Conversely, the more advantaged students are “likely easier to evaluate with or without test scores since they frequently attend high schools already known to admissions offices, making the high school’s grading policies and level of academic rigor better understood.”
Fearing their scores aren’t up to par, some less-advantaged students decline to submit SAT or ACT results, but what they don’t know is how they stack up against similar applicants.
To quantify the extent of the advantage students are forgoing, Dartmouth researchers compared data from entering classes during test-optional and test-required admissions cycles. They discovered that with an SAT score in the 1400-1440 range, less-advantaged students had an admissions rate of 8% compared to 3.8% for more-advantaged students.
Moreover, first-generation applicants with SATs between 1450 and 1490 had an 8.1% chance of admission when submitting scores versus a 2.8% probability when opting not to.
Even a score in the 25th percentile for Dartmouth enrollees can still be “highly competitive and very meaningful for less-advantaged students,” the authors determined.
Overall, with an SAT score of 1400 or more, less-advantaged applicants can increase their chance of admission 3.6 times by submitting scores.
These findings, the authors claim, suggest that test-optional policies “confer a disadvantage on high-achieving, less-advantaged applicants.”
Equating Disadvantage With Race and Ethnicity?
Notably absent from this discussion of less-advantaged students is any mention of race.
The authors claim Dartmouth’s implementation of its test-optional policy didn’t increase the “demographic diversity” of the applicant pool but indicate nothing about racial diversity.
That’s partly because of the tool they used to determine what constitutes disadvantage: the College Board’s “Landscape”. This service, employed by Dartmouth’s admissions office, offers information on students’ neighborhoods and high schools, providing the context necessary to make nuanced decisions about achievement.
Using this tool, the researchers discovered that 39% of Dartmouth applicants come from the least advantaged to moderately advantaged high schools.
Landscape examines data such as the percentage of students receiving free lunch, median family income, and crime rates.
According to the website Insight Into Diversity, Landscape helps colleges pursue “a more diverse student body,” offering a “concrete method to identify and support students from historically underrepresented groups by identifying high schools and neighborhoods that have a diverse population.”But does that diversity include race?
Not specifically. Landscape doesn’t offer any data on the racial makeup of high schools and communities per se.
Inferences can be made, though, calling into question the legality of using Landscape given the Supreme Court’s ruling.
A Brookings study found that “a university policy providing a substantial ‘admissions bump’ to lower-income applicants might help maintain racial and ethnic diversity at selective colleges.”
The relationship between low-income, first-generation, and race is messy at best. Race and ethnicity are closely entwined with socioeconomic status, although about half of first-generation undergraduates are white.
While Landscape ignores race, a similar tool called “Opportunity Atlas,” produced by the Opportunity Insights group at Harvard, does not. Atlas examines economic and social mobility based on neighborhood data, family earnings, incarceration rates, gender, and race, allowing users to control for certain variables.
Yale University began using Atlas after the SCOTUS decision. A 2024 Yale Daily News investigation attempted to determine if its use might help explain how the university’s first incoming class following the affirmative action ban saw stable enrollments among Black and Latino/a students and diminished numbers of Asian Americans.
Peer institutions such as the Massachusetts Institute of Technology, Columbia University, Brown University, and Cornell University experienced declines among underrepresented groups during that first admissions cycle.
Noting this incongruity, SFFA threatened Yale with legal action, along with Princeton University and Duke University, which experienced similar stability among students of color.
Perhaps Opportunity Atlas might be called into question. Mark Abraham, executive director of DataHaven, a community service organization providing demographic information, told the Daily News some observers might suspect the use of Opportunity Atlas equates to considering race in admissions, though others may think otherwise.
Stanford Law School Professor Richard Ford told the Daily News that using Opportunity Atlas “certainly could” explain Yale’s success in maintaining the diversity of its incoming class, suggesting the tool provides “exactly the kind of analysis” that could mitigate the effects of the affirmative action ban.
Yale reporters couldn’t determine if other members of the Ivy-Plus group use Opportunity Atlas. Only the University of Pennsylvania responded to their inquiry, stating that it does not.
Like Dartmouth and other peer schools, Yale has resumed mandatory testing, reaching the same conclusions about the advantages the policy confers to less-advantaged applicants.
“When used thoughtfully as part of a whole-person review process,” Yale University noted in its public announcement on reinstatement, “tests can help increase rather than decrease diversity in our class,” while inviting students to apply without test scores “can, inadvertently, disadvantage students from low-income, first-generation, and rural backgrounds.”
In Dartmouth’s case, the college didn’t suffer declines in people of color in its class of 2028, the first cohort admitted following the SCOTUS decision. In fact, the percentage of students from underrepresented racial and ethnic backgrounds actually increased from 26.5% to 28.2%, yet somehow the college escaped SFFA’s watchful eye.
The college now assumes a return to standardized testing will aid that trajectory.
So while new research suggests test-optional policies in fact help diminish student diversity in the broadest sense, at least among elite colleges, the Trump administration’s warning posits exactly the opposite assumption. Given that most selective institutions are now returning to mandatory testing, however, the government’s concern in this regard may be largely moot.