The October 3rd Machine Learning for Social and Behavioral Science Workshop brought together researchers eager to explore how emerging computational tools can help answer complex questions in their fields. Hosted by CSBS in partnership with the National Center for Supercomputing Applications (NCSA), the event was led by Dr. Jeff Levy from the University of Chicago Harris School of Public Policy, who shared his experience and perspective to a completely full auditorium of over fifty faculty, staff, and graduate students.
Throughout the day, Levy guided participants through key machine learning concepts including supervised learning, model evaluation, and cross-validation, while helping them situate these tools within the broader landscape of social and behavioral science. He emphasized the distinctions between causal inference and machine learning, and the valuable ways each can inform the other. “What is routine in one space may be a problem in another,” he shared, encouraging participants to think critically about how methodological choices shape outcomes.
Levy also underscored the practical value of contextual expertise in interpreting models. In one example, a satellite image of South Korea’s night lights might confuse someone unfamiliar with the area, but adding social science context reveals that the lights come from working fishermen. This connection between data and human context was a central theme of the day.
Hands-on exercises invited participants to experiment directly with machine learning tools in Python and R, and many felt confident enough to begin running machine learning operations with their own data that same day. The event brought together researchers from across disciplines, including economics, psychology, and education, who found creative parallels between their fields. One idea from natural resources or economics, for example, could inform a new approach in communications or public policy. The workshop provided a supportive environment for trying new methods, exchanging perspectives, and learning alongside others with similar experience levels.
As Levy noted in his presentation, “Machine learning, at its most basic, is the practice of using algorithms to parse data, learn from it, and then make a determination or prediction about something in the world.” The event demonstrated the growing opportunities for social and behavioral scientists to integrate computational methods into their research and highlighted how machine learning can extend the analytical reach of social and behavioral science.
Join us for our next methods event, AI Methods Series: Computer Vision and Beyond, on November 12. This session will explore how researchers can use computer vision tools to analyze visual data, expand their methodological toolkit, and uncover new dimensions of social and behavioral science research.
AI Methods Series: Computer Vision and Beyond
November 12 | 12:00 – 1:00 pm
NCSA Room 3100 | 1205 W. Clark St., Urbana