Just before the October 2019 federal court decision upholding Harvard’s admissions policy that factors in a student’s race, a new analysis of the university’s applicant data revealed that it might not be black and Hispanic students who are getting the biggest preferences but rather privileged white students, especially white athletes.
Specifically, the analysis found that more than 43 percent of white students admitted to Harvard between 2009 and 2014 fell into four preferential categories: athletes, legacies (the children of alumni), the children of big donors or faculty and staff children. The authors — three economists — constructed a mathematical model that attempts to replicate Harvard’s admission decisions and concluded that three-quarters of these white students would probably have been rejected if they had not had these admissions preferences and were treated as ordinary white students.
What’s interesting about this study is that the lead author is Peter Arcidiacono, a Duke University economist who served as an expert witness for Students for Fair Admissions, the plaintiffs who sued Harvard, during the trial. But this study, “Legacy and Athlete Preferences at Harvard,” which was posted on Arcidiacono’s website on Sept. 18, 2019, was not funded by the anti-affirmative action group. After the court decision on Oct. 1, 2019, Arcidiacono agreed to answer my questions by email.
“I personally find these preferences that favor the elite distasteful,” Arcidiacono wrote in the email interview. “I also think that it points towards flaws in the holistic admissions system.”
Holistic admissions, in theory, is supposed to offset advantages that high-income applicants have, such as the ability to afford test prep tutors. But holistic admissions, Arcidiacono found, was also allowing Harvard to give preferences to many things that money can buy.
“I found it shocking that at Harvard recruited athletes are dominated by whites,” Arcidiacono wrote. “And among non-recruited athletes those who score best on the athletic rating are white legacies.”
For example, Arcidiacono found that recruited athletes, legacies, and dean’s interest list applicants (a euphemism for big donors) are all over 68 percent white. The admissions rate for recruited athletes is above 85 percent. Harvard recruits athletes who often benefited from extensive family resources to excel in their chosen sports, from sailing to ice hockey.
In the study, Arcidiacono and his research partners, Josh Kinsler at the University of Georgia and Tyler Ransom at the University of Oklahoma, used data that was released to the public during the trial, including tables that Arcidiacono produced as an expert witness. In that capacity, Arcidiacono had access to detailed raw data for each applicant, but that data, which was not made public, wasn’t directly used for this study.
The draft paper, dated Sept. 11, 2019, may be revised prior to publication. Arcidianco is planning on writing about additional aspects of the Harvard admissions case and hopes to publish all of the analysis more formally in the future.
From the publicly available data on more than 160,000 applicants, the authors estimated the probability of being admitted for each part of student’s application, from strong grades and extracurriculars to being an athlete or a legacy applicant. Of course, some legacies might have strong grades and test scores and might have gotten in without the preferential treatment. In the model Arcidiacono built, he was able, in effect, to “turn off” any particular part of a student’s application and predict what the admissions decision would have been. That is how he predicted that three-quarters of the white admitted students who were athletes, legacies, donors and faculty children probably would have been rejected.
Of course, there are many unobservable, unquantifiable aspects of a student that a paper (or online) application doesn’t reveal, such as a great sense of humor or artistic creativity. Arcidiacono’s model necessarily ignores these qualities. Arcidiacono also ignored the “personality” ratings that admissions officers awarded each candidate because he said he considers them to be racially biased. (Admissions officers picked a personality score for each candidate, from 1 to 6, based on their perception of the applicant’s personality traits, such as a sense of humor, kindness or courage.)
How much you accept Arcidiacono’s conclusions that white students are advantaged at the expense of Asian American students depends on how much you accept judging students on their observable academic qualifications and extracurricular activities and how much you want to add sometimes subjective qualities to the mix.
In this analysis, Arcidiacono didn’t focus on whether Asian American students were discriminated against, which was at the heart of the affirmative action case. But this study provides more evidence of indirect bias against Asian American applicants since Arcidiacono’s data showed that they are less likely to fit into a category of athlete, legacy, donor or faculty child.
“Harvard’s argument was that Asian Americans weren’t as multidimensional as whites,” he said in the email interview. “This was based on Harvard’s four profile ratings: academic, extracurricular, athletic, and personal. Asian Americans are stronger than whites on the first two. So basically Harvard argues that Asian Americans aren’t multidimensional because they’re not on the sailing team and score worse on the personal rating where bias is apparent.”
Not surprisingly, Arcidiacono said he was “very disappointed” by the federal court decision. The case is expected to be appealed and could end up in the Supreme Court.
This story about Harvard admissions was written by Jill Barshay and produced by The Hechinger Report, a nonprofit, independent news organization focused on inequality and innovation in education. Sign up for the Hechinger newsletter.