Harvard consistently rated Asian-American applicants lower than others on traits like "positive personality", likeability, courage, kindness and being "widely respected", according to an analysis of more than 160,000 student records filed on Friday by a group representing Asian-American students in a lawsuit against the university.
Asian-Americans scored higher than applicants of any other racial or ethnic group on admissions measures like test scores, grades and extracurricular activities, according to the analysis commissioned by a group that opposes all race-based admissions criteria. But the ratings of the Asian-American applicants' personal qualities significantly dragged down their chances of being admitted, the analysis found.
The court documents, filed in the federal court in Boston, also show Harvard conducted an internal investigation into its admissions policies in 2013 and found a bias against Asian-American applicants. But Harvard never made the findings public or acted on them.
Harvard admitted only 4.6 per cent of its applicants this year. That has led to intense interest in the university's closely guarded admissions process. Harvard had fought furiously over the last few months to keep secret the documents that were unsealed on Friday.
The documents came out as part of a lawsuit charging Harvard with systematically discriminating against Asian-Americans, in violation of civil rights law. The suit, brought by Students for Fair Admissions, says Harvard imposes what is in effect a soft quota of "racial balancing". This keeps the numbers of Asian-Americans artificially low, while advancing less qualified white, black and Hispanic applicants, the plaintiffs contend.
The findings come at a time when issues of race, ethnicity, admission, testing and equal access to education are confronting schools across the United States, from selective public high schools like Stuyvesant High School to elite private colleges. Many Ivy League schools, not just Harvard, have had similar ratios of Asian-American, black, white and Hispanic students for years, despite fluctuations in application rates and qualifications, raising questions about how those numbers are arrived at and whether they represent unspoken quotas.
Harvard and the group suing it have presented sharply divergent views of what constitutes a fair admissions process.
"It turns out that the suspicions of Asian-American alumni, students and applicants were right all along," Students for Fair Admissions says in a court document laying out the analysis. "Harvard today engages in the same kind of discrimination and stereotyping that it used to justify quotas on Jewish applicants in the 1920s and 1930s."
Harvard vigorously disagreed on Friday, saying its own expert analysis showed no discrimination and that seeking diversity is a valuable part of student selection. The university lashed out at the founder of Students for Fair Admissions, Mr Edward Blum, accusing him of using Harvard to replay a previous challenge to affirmative action in college admissions, Fisher versus the University of Texas at Austin. In its 2016 decision in that case, the Supreme Court ruled that race could be used as one of many factors in admissions.
"Thorough and comprehensive analysis of the data and evidence makes clear that Harvard does not discriminate against applicants from any group, including Asian-Americans, whose rate of admission has grown 29 per cent over the last decade," Harvard said in a statement. "Mr Blum and his organisation's incomplete and misleading data analysis paints a dangerously inaccurate picture of Harvard's whole-person admissions process by omitting critical data and information factors."
In court papers, Harvard says a statistical analysis could not capture the many intangible factors that go into Harvard admissions. Harvard says the plaintiffs' expert, Duke University economist Peter Arcidiacono, had mined the data to his advantage by taking out applicants who were favoured because they were legacies, athletes, the children of staff and the like, including Asian-Americans. In response, the plaintiffs say their expert had factored out these applicants because he wanted to look at the pure effect of race on admissions, unclouded by other factors.
Both sides filed papers on Friday asking for summary judgment, an immediate ruling in their favour. If the judge denies those requests, as is likely, a trial has been scheduled for October. If it goes on to the Supreme Court, it could upend decades of affirmative action policies at colleges and universities across the country.
Harvard is not the only Ivy League school facing pressure to admit more Asian-American students. Princeton and Cornell and others also have high numbers of Asian-American applicants. Yet their share of Asian-American students is comparable with Harvard's.
In Friday's court papers, the plaintiffs describe a shaping process that begins before students even apply, when Harvard buys data about PSAT scores and GPAs, according to the plaintiffs' motion. It is well documented that these scores vary by race.
The plaintiffs' analysis is based on data extracted from the records of more than 160,000 applicants who applied for admission over six cycles from 2000 to 2015.
They compare Harvard's treatment of Asian-Americans with its well-documented campaign to reduce the growing number of Jews being admitted to Harvard in the 1920s. Until then, applicants had been admitted on academic merit. To avoid adopting a blatant quota system, Harvard introduced subjective criteria like character, personality and promise. The plaintiffs call this the "original sin of holistic admissions". They argue that the same character-based system is being used now to hold the proportion of Asian-Americans at Harvard to roughly 20 per cent year after year, except for minor increases, they say, spurred by litigation.
White applicants would be most hurt if Asian-American admissions rose, the plaintiffs say.
In its admissions process, Harvard scores applicants in five categories - "academic", "extracurricular", "athletic", "personal" and "overall". They are ranked from one to six, with one being the best.
Whites get higher personal ratings than Asian-Americans, with 21.3 per cent of white applicants getting a one or two compared with 17.6 per cent of Asian-Americans, according to the plaintiffs' analysis.
Alumni interviewers give Asian-Americans personal ratings comparable to those of whites. But the admissions office gives them the worst scores of any racial group, often without even meeting them, according to Mr Arcidiacono.
Harvard says that while admissions officers may not meet the applicants, they can judge their personal qualities based on factors like personal essays and letters of recommendation.
The university says it was implausible that Harvard's 40-member admissions committee, some of whom were Asian-Americans, would conclude that Asian-American applicants were less personable than other races.
University officials did concede that its 2013 internal review found that if Harvard considered only academic achievement, the Asian-American share of the class would rise to 43 per cent from the actual 19 per cent. After accounting for Harvard's preference for recruited athletes and legacy applicants, the proportion of whites went up, while the share of Asian-Americans fell to 31 per cent. Accounting for extracurricular and personal ratings, the share of whites rose again, and Asian-Americans fell to 26 per cent.
What brought the Asian-American number down to roughly 18 per cent, or about the actual share, was accounting for a category called "demographic", the study found. This pushed up African-American and Hispanic numbers while reducing white and Asian-American. The plaintiffs say this meant there was a penalty for being Asian-American.
"Further details (especially around the personal rating) may provide further insight," Harvard's internal report said.
But, the plaintiffs say in their motion, there was no further insight because "Harvard killed the study and quietly buried the reports". Harvard says the review was discounted because it was preliminary and incomplete.
NYTIMES
