When a professor posted a Twitter thread about how Cornell University improved diversity in its computer-science PhD program, it quickly went viral.
“[W]e made a big step in improving diversity of the program. Let me tell you about it,” wrote David Bindel, the PhD admissions chair for Cornell University’s Computer Science department, earlier this month.
It’s clearly an issue that other higher ed leaders are interested in hearing about—the tweet has been shared more than 750 times and drew nearly 2,000 likes.
Computer science is “not just white dudes slinging code, and we all suffer for it when the world thinks that’s what it is,” Bindel wrote.
So how did Cornell do it? Bindel tells EdSurge that last year, Cornell had one Native American applicant, four African American and 17 hispanic applicants to its computer science PhD program, out of a pool of about 850 people. To encourage greater diversity in the next class. Bindel reached out to coordinators of the McNair Scholars Program, an organization that seeks to get undergraduate students—many of whom are underrepresented in graduate education—ready for graduate studies across the United States.
Bindel tapped into his personal network as well, which to some extent involved friends who now teach at different institutions. He also reached out to current Cornell computer science graduate students who had contacts.
Bindel says the rest of the application-and-review process was “pretty standard.” However, the committee added two new questions to the evaluations this year. The first was on how well the students had done with the opportunities they had.
“The idea was somebody who went through Stanford and sat in the back of a research group, then got their name on a couple of papers might not actually have done as impressive a job with the resources that they had as somebody who didn’t necessarily get a paper out, but did really impressive research as an undergraduate at a university where they had less mentoring,” Bindel says.
The committee also considered if the applicant brought a unique perspective to their research area. Bindel says both of those questions did affect decisions for some students.
“It wasn’t just that this got us more underrepresented minorities; it got us different students,” he says.
Last year, Cornell admitted 135 computer science PhD students out of a pool of about 850, targeting a class size of 40. This year, the applicant pool was about 1,300, and Cornell admitted 147 students, targeting a class size of 50. Out of the 147 admits, 59 have chosen to attend. Specifically, 52 have chosen to matriculate this upcoming year, and seven others have deferred admission.
As for the makeup of those 59 students, 30 are non-U.S. citizens from countries that include China and India. Of the latter, seventeen are men (two are Hispanic, four are African American, two are Asian American and nine are Caucasian), and 12 are women (one is African American, one is Pacific Islander, one is Asian American and 9 are Caucasian). In comparison, in Fall 2017, 42 students accepted offers, including four deferrals, Bindel says. Of those offers, 18 were United States citizens—and the only offers to underrepresented minorities were to two African American men, both of whom accepted.
“Overall, we actually got more selective this year than we did last year,” Bindel says.
Bindel adds that this year, there was a disproportionately high yield among female students and underrepresented minorities. A key piece of the department’s recruiting, he says, is the graduate student visit day, where prospective students meet with faculty and see what the campus is like. Bindel thinks it helped that students who attended saw that other admitted students were from a similar racial or gender background.
Hakim Weatherspoon, an associate professor in Cornell’s computer science department who works on increasing the number of underrepresented minorities who pursue a PhD, says the lack of diversity in computer science faculty and graduate students goes back to undergraduate education—and can go back to high school and earlier as well.
He thinks not everyone is aware of computer science. And once underrepresented minorities find out they’re interested in the subject, some may fear pursuing it because they have less experience than those who started learning the subject sooner.
He adds that another reason could be that underrepresented minorities don’t have examples in their families and universities of those who’ve been successful in the field. And finally, he thinks isolation can be a factor—meaning, if a student is the only one or one of the only ones of his or her race in computer science classes. He thinks that extends to the graduate level. Increasing diversity, he says, is not just about getting people to apply, but getting them into a cohort as well.
“If you’re only one student, every other year in a PhD program, you’re going to be isolated.”
