A new National Science Foundation grant will allow Valerie Barr ’77, Jean E. Sammet Professor of Computer Science to further her research into computer science education — and to continue to shake up the world of computer science.
The five-year grant funds the creation of the Alliance for Identity-Inclusive Computing Education, a collaboration with multiple organizations and universities, including the College, Duke University, the University of Oregon and Northeastern University.
The goal is to develop tools and strategies in computing education that increase the diversity in a field dominated by middle- to upper-class white and Asian able-bodied cisgender males and to improve retention and degree completion rates of historically underrepresented groups in high school and undergraduate students.
The grant will allow Barr to go beyond her longtime work of “getting students into the door and exploring computer science opportunities,” she said.
“This grant focuses on how to keep students in the room. Many are getting into their first class but don’t stay because of a variety of factors. They won’t stay if it’s an inhospitable environment.”
Nicki Washington, professor of the practice in the Department of Computer Science at Duke and a researcher for the project noted that diversity attrition in computer science is a complex problem.
“Prior efforts emphasize that it’s simply about access, courses and training. But that’s not the case when they’re in a class … dealing with problematic peers and faculty who then go on to lead these companies and shape the technology industry. We need to be creating better graduates,” she said.
One way to combat the attrition is by developing robust connections within the field to support diverse students.
“We want to make sure we grow our network as much as possible, whether that is through adding collaborators at the partner level or affiliates who are going to take the tools that we create into the classroom.”
One component of the project that Barr is overseeing is the critical role of teaching assistants in supporting students. “At research universities, faculty members can do all they want with pedagogy, but the students are usually working with teaching assistants. If they have a bad experience, they’ll leave. So my efforts are toward creating an infrastructure for teaching assistant-mentor training.”
Barr is building off the College’s Megas and Gigas Educate, or MaGE, peer mentoring program for computer science students, developed by Associate Professor of Computer Science Heather Pon-Barry, Professor and Chair of Computer Science Audrey St. John, Professor of Psychology and Education Becky Wai-Ling Packard and CS Program and Lab Coordinator Barbara Rotundo. As part of the program, upper-level computer science students serve as peer mentors to students in beginning computer science courses.
The second piece of Barr’s projects is an analysis of students getting computer science degrees versus other disciplines. The normal way of talking about women getting computer science degrees is as a percentage of total degrees awarded in the field. But those numbers are skewed because of demographic disparities in the undergraduate population.
“The question is, ‘Of women getting degrees, what percentage are getting computer science degrees?’” she said.
Barr will also be looking at race and ethnicity categories within the field and the difference between intent-to-major indicated by students when they enter college and actual degree acquisition four years later.
Except for a brief period in the mid-1980s, Barr said the portion of women earning computer science degrees never exceeded 2.5% — about a third of the percentage of men.
“As a comparison point, in biology the percentage of cohort degrees earned is roughly 7.5% for both men and women. [But] engineering has an even larger gender gap than does computer science, with 2.5% of women obtaining degrees versus 10% of men,” Barr said.
The Duke team and four other awardees from this funding cycle join eight existing NSF-funded alliances and the NSF INCLUDES National Network, which consists of more than 3,000 partners dedicated to broadening participation in STEM fields. With this grant, the alliance can further add partners and resources, including those from MaGE, to a nationwide database.
Barr said the database would allow schools to upload and compare their degree data to national numbers on gender, race and ethnicity. Until now, many efforts to diversify the field are focused on the “deficit model,” which focuses on access, curricula and training, she said.
“We’re flipping that theory on its head and looking at how to fix the environment of computer science education so it’s inclusive and students are making decisions equally regardless of gender, race or ethnicity.”