SIAM News Blog
SIAM News
Print

Promoting Diversity in Data Science Education

By Lina Sorg

The recruitment and retention of women and other minorities has been a significant challenge for science, technology, engineering, and mathematics (STEM) fields for the last several decades. As universities work to maintain a more diverse student body in STEM-related studies, they should pay particular attention to the burgeoning field of data science, which has experienced exponential growth in recent years. Simply put, diversity is necessary to meet the growing demands for data scientists in many industry settings. During a minisymposium presentation at the 2020 SIAM Annual Meeting, which took place virtually this week, data scientist Karl Schmitt spoke about the value of recruiting and retaining a diverse pool of students for the field of data science. “We want diversity in our future STEM workforce,” he said. “It’s just morally and ethically right that we do this.”

Studies have shown that diversity improves upon a wide variety of success measures. Diverse teams are more likely to experiment, leading to increased revenue, customers, and market shares. Tech start-ups are also more likely to find success with a diverse staff population, and teams that embody a variety of genders and ethnicities frequently outperform those that do not. “It really does produce better outcomes from students in the classrooms and in the workforce,” Schmitt said. “If we want to be growing in positive ways, we really need to be going for diversity.”

Schmitt drew many of his points from the National Center for Women & Information Technology (NCWIT), which connects academia and industry, conducts great research, and encompasses the K-12 level and higher education. “Most of their findings turn out to be applicable regardless of what field you’re in,” he said. The NCWIT Engagement Practices Framework outlines ways in which faculty can broaden participation in computing-based fields, both in the classroom and beyond. The first pillar, “Make it Matter,” encourages instructors to make classroom curricula relevant and meaningful to students. This involves making explicit and interdisciplinary connections to the field of computer science, addressing misconceptions about the field, and incorporating student choices in assignments. “A lot of students have this misconception that mathematics is either something you get or don’t get,” Schmitt said, adding that faculty members should accentuate the experimental nature of mathematics.

National Center for Women & Information Technology (NCWIT) Engagement Practices Framework.
The second pillar, “Build Student Confidence and Professional Identity,” highlights a growth mindset. “Students that come in as minorities often have less confidence,” Schmitt said. Leveling the playing field, so to speak, ensures that all students build self-assurance collectively. Teachers should offer both effective encouragement and student-centered assessment, and recognize that students are always growing. It is also important that minority students have ample opportunities to interact directly with faculty members, teaching assistants, and other individuals who are further along in their career tracks and can supply guidance and advice.

Another aspect of this subset involves the aggressive mitigation of stereotype threat. “Stereotype threat is the idea of behaving to what the expectations are, and unintentionally or intentionally saying things that indicate that there is a difference in stereotypes,” Schmitt said. “If you’re not aware of stereotype threat, it’s hard to actively mitigate it in your classroom.”

The NCWIT framework’s third and final pillar is “Grow a Positive Student Community,” which focuses on inclusivity. Instructors must avoid stereotypes, encourage student interaction (through group projects or interactive classroom seating, for example), and emphasize well-structured collaborative learning rather than independent tasks.

In addition to the Engagement Practices Framework, inclusive teaching and culturally responsive pedagogy serves as another valuable resource. This pedagogy is comprised of four principles: transparency, academic belonging, structured interactions, and critical engagement of difference. Schmitt addressed transparency first. Faculty members must explicitly state the purpose, tasks, and evaluation of assignments. Some students, particularly first-generation students, do not yet possess the mental framework to turn vague instruction into action. “It’s important to have thorough and clear transparency in everything you do,” Schmitt said. “The more transparent you are, the better your life gets. If you explicitly tell students what you want, you’re far more likely to get it.”

Schmitt then turned to the concept of academic belonging. Fostering academic belonging involves inviting all students to feel like they are a part of the larger academic community. This involves adjusting class projects and assignments based on students’ existing knowledge and providing opportunities that are designed for risk and failure. The latter strategy highlights the importance of the growth process, which is especially relevant in upper-level, experimental mathematics and data science.

Faculty members who wish to provide structured interactions must offer frameworks for student engagement — in the form of goals, protocols, and processes that ensure equitable access and contribution. For example, teachers should not simply assume that all students have a laptop. Schmitt suggested that classrooms have spare laptops so that everyone can participate equally on assignments.

Finally, the fourth dimension of inclusive teaching—critically engaging difference—encourages faculty members to recognize that different students bring diverse identities, needs, and strengths to the classroom setting, all of which have value.

Schmitt then offered several examples of inclusive learning from his time as an assistant professor at Valparaiso University (Valpo), which one can adapt to his/her university’s departments and capabilities. For instance, Valpo hosted an interdisciplinary speaker series that featured all female speakers—including a security expert and a museum designer—from a variety of cross-disciplines and computer science-based areas. Schmitt marketed this series with a campus-wide pathway so that it reached a wide audience. Some attendees even enrolled in computer science courses afterwards.

Schmitt also taught an introductory, project-driven data science class with an emphasis on social good. “All of my students left these projects and realized their value, even as just a learner,” he said. “They could see what kinds of spaces they needed to learn.” He carefully balanced teams for groupwork based on discipline, gender, and level of experience, so that each group had a diverse pool of students.

In addition, Valpo held a professional development workshop on stereotype threat for both students and faculty. The decision to invite students was intentional, because the school wanted students to see what types of issues they might personally encounter in the classroom (even if the professor is not aware). This workshop put students in a more informed, powerful position for future interactions, and empowered women and minorities to recognize and address stereotype-driven situations.

Schmitt concluded by reminding the audience that the number of women in computer science and statistics is quite low. While this is not necessarily the case for data science, which currently features a more equal gender breakdown, it is important to prioritize diversity to avoid a backslide in the field. “If we don’t take an active role in shaping the classes going forward, we’re going to lose the fact that we already have diversity,” he said.

 Lina Sorg is the managing editor of SIAM News.
blog comments powered by Disqus