Computational Social Science Training Program

University Of California Berkeley, Berkeley CA

Investigators

Linked publications & trials

Abstract

Project Summary/Abstract The Computational Social Science Training Program (CSSTP) at UC Berkeley provides training in advanced analytics to predoctoral students in the social and behavioral sciences who investigate health topics covered by the Eunice Kennedy Shriver National Institute for Child and Human Development. CSSTP combines Berkeley’s long-standing strength in quantitative social and behavioral science with its nationally-recognized campus programs in data science education, practice, and research. It serves five entering trainees per year over five years. The training faculty includes 30 social scientists who have exemplary records of developing and applying novel statistical methods to health-related social/behavioral science problems, as well as 19 data scientists who are leading figures in the foundations of mathematics, statistics/biostatistics, and computer science. Trainees, who are drawn from a diverse pool of students in six social science doctoral programs, are provided with a rigorous and tailored program designed to teach a team science-based approach to problem solving and to emphasize the analysis of intensive or voluminous longitudinal data and high-density, large sample or population level agency databases. Each trainee is supported by a dual-preceptor model in which they are provided with a social sciences faculty mentor and a data science mentor who help to facilitate the trainee's progress through the program. CSSTP trainees are provided with community space at the Berkeley Institute for Data Science (BIDS), a dynamic multi-disciplinary data science research center, where trainees work alongside other data science fellows in residence. After completing their first-year course requirements in their home departments, trainees formally enter the program in their second year of graduate school, devise an individual development plan, and take a core two- semester course in computational social science, team-taught by training faculty. This course introduces students to essential data science methods and tools, including Python and R programming, data management, natural language processing, machine learning, causal inference, and responsible conduct and reproducibility of research. Instructional modes include lectures, in-depth discussion, and small group learning exercises. In the following year, students apply these skills through placements on collaborative health-related research teams or labs on campus and/or with external partners, thus putting skills in advanced analytics into practice through research involving the development and implementation of new methods. Additional training tailored to student needs and interests is provided through elective courses, a weekly computational social science workshop series, and ongoing working groups at BIDS and the Social Science D-Lab, a campus hub for data science training and research for social scientists. CSSTP’s benefits will extend to the greater campus and beyond by stimulating new faculty collaborations and by creating a critical mass of rigorously trained computational social science students who will be competitive and qualified for jobs in rapidly changing and evolving data intensive fields.