Experts Are Confused by Harvard’s Race Data. Here’s Why. | News | The Harvard Crimson

After the Supreme Court struck down affirmative action last summer, experts eagerly awaited Harvard’s demographic data for the Class of 2028 — hoping it would give a clear and unambiguous picture of the ruling’s impact on Harvard College’s admissions.

Except, it didn’t.

The school’s decision to change the scale for reporting racial breakdowns makes it more difficult to accurately compare this year’s numbers to prior years, according to three experts. Instead of computing percentages from the whole class, as in the past, Harvard calculated percentages from only the 92 percent of students who chose to report their race.

For an accurate comparison, the College stated that it recalculated the Class of 2027’s demographic data on the same scale as this year — out of the 96 percent of students in last year’s class who reported their race.

However, the data for the Class of 2027 used for comparison in Wednesday’s press release was inconsistent with numbers the school previously shared even when accounting for the College’s change in scale, per a Crimson analysis confirmed by independent experts.

“This year, we are reporting the share of students who responded to the question regarding their race/ethnicity so we are reporting data only for students who reported their race to us,” College spokesperson Jonathan Palumbo wrote in a statement. “This approach provides a clear view of the students who are studying at Harvard.”

Palumbo, however, declined to provide additional information about the methodology used to compute the racial breakdowns or about the source of the data when asked by The Crimson to explain potential discrepancies in the data. He also declined to answer questions about the decision to change the school’s data reporting practices.

Dean of Admissions and Financial Aid William R. Fitzsimmons ’67 declined an interview for this article through a Harvard spokesperson.

Richard D. Kahlenberg ’85, who served as an expert witness for Students for Fair Admissions in its lawsuit against the University, said the change in methodology made it difficult to interpret the figures released by Harvard and understand the demographic shift in the Class of 2028.

“When I saw the numbers, it was hard to know what to make of this change,” Kahlenberg said. “Harvard had been presenting data in a consistent fashion for a number of years, and then this year decided to change the presentation.”

“It introduced — I think for many people — an element of confusion,” he added.

The College reported in its press release that — out of those who chose to share their race in the Class of 2027 — 18 percent of students identified as Black, 37 percent as Asian American, 14 percent as Hispanic, 2 percent as Native American, and less than 1 percent as Native Hawaiian.

However, when members of the Class of 2027 enrolled in May 2023, the College had announced that 14.1 percent of students identified as Black, 29.8 percent identified as Asian, and 11.1 percent identified as Hispanic or Latino.

The Crimson independently converted Harvard’s previously-reported demographic data for the Class of 2027 to the College’s new scale — calculating racial demographics as a percentage of the 96 percent of students in the Class of 2027 who Harvard said chose to report their race, rather than as a percentage of the entire Class of 2027 student body.

With this methodology, the numbers shared by the College in May suggest that, out of those who chose to report their race, approximately 14.7 of the Class of 2027 identified as Black, 31.0 percent as Asian American, 11.6 percent as Hispanic, and 2.4 percent as Native American or Hawaiian — several percentage points different from what the College reported Wednesday.

There were also inconsistencies between the Class of 2027 race data reported by the College in October to the Common Data Set and the numbers shared on Wednesday. In the Common Data Set, the College reported that 33 of 1,644 students in the Class of 2027, about 2 percent, had not reported their race. However, Wednesday’s press release said this number was 4 percent.

The Crimson’s calculations were made using data available to the public. Palumbo declined to clarify the reasoning behind the difference in the data, if a different data source was used for the two reports, or provide information about when the data was collected.

The Crimson’s calculations do not account for any changes in the Class of 2027’s makeup since May 2023, for the data released at the time of enrollment, or October, for the Common Data Set. This could include admitted waitlist students and any students who later decided to delay or cancel their admission.

Each year, between 90 and 130 students defer their admission to Harvard College, according to the College’s admissions website. However, this decision is typically made before the College reports data to the Common Data Set.

Beyond the data inconsistencies, the Class of 2028 percentages — computed only from students who reported their race — are missing responses from a large chunk of freshmen. The 8 percent of students who chose not to self-identify their race this year doubled last year’s 4 percent.

And, it’s unclear if the remaining 8 percent is distributed identically across demographic groups to those who reported their race — or if that group is skewed. UCLA law professor Richard H. Sander ’78 said he believes Asian American students are most likely overrepresented across those who declined to report their race.

“We saw the same dramatic increase in ‘no race’ applicants at the University of California after Prop 209,” Sander wrote in an email, referring to a 1996 California ballot proposition that outlawed the use of affirmative action policies at the state’s public institutions.

He added that, after California’s affirmative action ban, the sum of UC applicants who declined to self-identify their race and those who self-identified as Asian American “corresponded with the share” of applicants who self-identified as Asian American before the rule change — implying that Asian-American applicants disproportionately declined to declare their race.

Sander projects that, as a result, the proportion of Asian American students in Harvard’s Class of 2028 is likely larger than the 37 percent the University reported.

—Staff writer Elyse C. Goncalves can be reached at elyse.goncalves@thecrimson.com. Follow her on X @e1ysegoncalves or on Threads @elyse.goncalves.

—Staff writer Matan H. Josephy can be reached matan.josephy@thecrimson.com. Follow him on X @matanjosephy.

—Staff writer Neil H. Shah can be reached at neil.shah@thecrimson.com. Follow him on X @neilhshah15.