That Recruitment Letter From Harvard Probably Doesn’t Mean Much

The message, emailed to more than 100,000 high school students, was seductive and flattering: “Your strong grades and standardized test scores indicate to us that Harvard and other selective institutions may be possibilities for you.”

Harvard encouraged them to apply. But many of the recipients had little chance of getting in, especially if they were black, according to a new analysis of the university’s admissions data by three economists.

As the college application season heats up, students across the country are getting letters and emails inviting them to apply to schools they may never have dreamed of attending — schools that are trying, for their own reasons, to attract as many applications as possible for the slots they have to fill. So-called recruit-to-deny strategies have been an open secret of elite colleges for years. Lower acceptance rates can help push up schools’ reputations and rankings by making them look more selective.

But the new analysis, a working paper for the National Bureau of Economic Research, argues that Harvard recruits students differently based on race, intentionally drawing applications from a large portion of African-Americans in particular who effectively have no chance of getting in.

While the paper acknowledges that attracting a diverse applicant pool could promote diverse incoming classes, it raises questions about whether such recruitment strategies amount to a cynical enterprise by college admissions offices, in which students are being sold false promises to serve the schools’ interests.

“It’s about appearances,” said Peter Arcidiacono, a Duke economist and co-author of the study. “You can say, ‘Look, we’re trying.’”

Colleges say they send out recruiting letters because they want to find the rare diamonds in the rough, the promising students who might not apply without a nudge. The recruitment letters are based mostly on scores for the PSAT, the practice version of the SAT that is taken in a student’s sophomore or junior year, well before the time for college applications. So even if the scores are not quite up to snuff, Harvard is gambling that some of these students — who seem promising at that stage — will improve enough by the time they apply to be worthy of admission.

But the authors of the working paper argue that Harvard made the test-score cutoff too low for recruiting African-American students, without changing its reliance on high scores for admission.

The analysis is based on data that Harvard was forced to release during a trial, a year ago, in which the college was accused of discriminating against Asian-American applicants. A federal judge ruled in Harvard’s favor this fall; the plaintiffs, Students for Fair Admissions, representing a group of Asian-Americans who were rejected by Harvard, are now appealing.

Dr. Arcidiacono was an expert witness for the plaintiffs at the trial. The study was written with economists at the University of Georgia and the University of Oklahoma.

A spokeswoman for Harvard, Rachael Dane, declined to comment on details of the study. But she said that the recruitment efforts it describes are a valuable source of students who are eventually admitted, accounting for 60 percent of freshmen in a typical year and more than 80 percent of minority freshmen.

For their recruitment drives, colleges buy names, ZIP codes, race and ethnicity and other information about students with test scores within a specified range from testing companies like the College Board and ACT. Students are often flooded with as many as 50 electronic solicitations via email, Snapchat and Instagram.

For the Class of 2018, Harvard sent out more than 114,000 letters and admitted 2,047 students. Almost half of those who qualified for a recruiting letter were members of underrepresented minorities.

Test-score cutoffs for the recruitment letters varied by race, gender and geography, and sometimes changed from year to year, according to testimony in the admissions trial. To get a letter in the fall of 2013, white and Asian-American men had to have scored at least 1380 on the SAT (converted from the equivalent on the PSAT), and black students and other underrepresented minorities had to have scored at least 1100.

While the numbers of applications Harvard gets from students of all races have increased over the years, the report said, African-American applications have soared, driven by students with lower SAT scores.

The increases were especially large for the classes applying between roughly 2003 and 2007, with the share of applicants who were black growing to 10.1 percent from 6.4 percent, the report said. But despite the growth in applications, the share of admitted students who were black stayed the same.

Harvard appeared to have deliberately encouraged black applicants whose SAT scores were lower than 550 on any of the SAT subsections, according to the study. In 2007, for example, there were more black applicants with scores below 550 than above 640 — a pattern “not mirrored by other racial groups,” the report said.

The African-American applicants who were admitted “overwhelmingly” had math SAT scores over 640, while “virtually no admits” had scores under 550.

As recruiting changed and the number of applications rose, Harvard’s admission rates for African-American students declined steeply, and fell more in line with the rates for other racial groups, because many of the students in the expanded recruiting pool were hopelessly unqualified, the report said. Before 2003, black students were at least 70 percent more likely to be admitted than Asian-Americans, but by 2007, the two groups had similar admission rates.

The report connected the rise in black applicants to two events: Harvard’s introduction of a more generous financial aid program in early 2004, and the Supreme Court’s rulings in landmark affirmative action cases involving the University of Michigan, Grutter v. Bollinger and Gratz v. Bollinger, which were decided on the same day in 2003.

The authors said they could not be sure why Harvard had changed its recruiting practices around that time. But they speculated that the university may have been trying to balance out any sharp disparities in admissions rates across racial groups, in order to “downplay the magnitude of race-based preferences” and avoid future litigation.

At the Harvard trial, William Fitzsimmons, dean of admissions, offered his own reason for why the university had lowered its test-score cutoffs for recruiting African-American and Hispanic applicants. “It really comes down to the economic disadvantage,” he testified. “These are students who have less of an opportunity, on average at least, to prepare well and to do well on standardized testing.”

He said that Harvard was careful not to send out too many search letters, because it “could have a negative effect” by discouraging students who saw other people be rejected.

Studies by scholars like Caroline Hoxby of Stanford have found that the vast majority of low-income high-achievers do not apply to any selective college, possibly because of a dearth of role models or of information about opportunities.

Akil Bello, a college admissions consultant who works with underprivileged students, said a weakness of the new study of Harvard’s recruitment was that it drew inferences from purely statistical data. “I think in the complex ecosystem of admissions, these one-note arguments are reductive, problematic and probably off-kilter,” he said. “But I think it’s on Harvard to be more transparent if they want these narratives to go away.”

Mr. Bello said that in his experience, students who applied to Harvard did not do so because they got recruitment letters. He said they usually were among the higher-achieving students in their schools, and were encouraged to apply by their parents and counselors.

“Generally, if somebody is looking at applying to a highly selective place like Harvard, there are informed adults around them, helping them navigate that,” Mr. Bello said.

It is optional for students to allow colleges to buy their names and addresses, and those who are contacted by colleges are 12 percent more likely than their peers to enroll in a 4-year college, according to the College Board.

Susan C. Beachy contributed research.