Persistence in STEM: A Mixed-method Study of a Data Science Program for Underrepresented Students

• In Event: 40217 - Education and Inequality

Abstract

Data science is a relatively new field in STEM education, and it remains an open question
whether disparities in STEM education outcomes in existing STEM fields will be reproduced in
data science. This paper presents an impact and process evaluation of an intervention called
“Data Scholars” designed to support students underrepresented in data science both
academically and socially in an introductory data science course at UC Berkeley, with the goal
of encouraging student persistence in data science. Drawing on a matching design that
leverages both administrative and student survey data paired with qualitative interviews with
students, this study examines linkages between participation in two variants of Data Scholars
and both intermediate mechanisms and final outcomes. We find that the variant with the most
consistent impact improves sense of mattering, self-efficacy, and science identity, increases
interest in teaching data science and data science research, and increases future data science
coursetaking. Furthermore, these impacts seem to be driven by increases in student capacity to
navigate course and campus resources rather than improvements in grades and community
belonging. Implications for supporting underrepresented students in data science and STEM
education more broadly are discussed.

Authors

• David J. Harding, University of California-Berkeley

• Daniel Lobo, University of California-Berkeley

• Claudia von Vacano, UC Berkeley

• Byeongdon Oh, SUNY Polytechnic Institute

• Raquel Xitlali Zitani-Rios, University of California-Berkeley

• Jacqueline Brown, University of California-Berkeley

• Tiffany Chrysilla Hamidjaja, University of California-Berkeley

• Michael Ruiz, UC Berkeley

• Renee Starowicz,, UC Berkeley

• Rodolfo Mendoza-Denton, UC Berkeley