How do we measure ideological bias in media, in AI systems, and even in human judgment?
In this online session, Nicholas Beauchamp (Political Science, Northeastern University) examines how early computational models attempted to detect bias—and where they fell short. While many approaches rely on opaque “black-box” systems, his work introduces theory-informed structure to large language models in order to detect subtle, context-dependent bias while improving transparency and explainability.
This talk explores:
- How selection effects shape news coverage
- Why measuring AI bias often mirrors the same challenges as measuring human bias
- The limits of simple and complex models
As AI tools become central to social science research, understanding their limitations is just as important as using them. Register to learn how cutting-edge AI methods are being applied to complex social phenomena.
This hands-on workshop provides a structured introduction to tree-based methods for classification and regression, including decision trees, random forests, and boosting. Led by Jeff Levy of the Harris School of Public Policy, the session examines recent applications from the scholarly literature and guides participants through implementing these techniques in Python or R.
Why it Matters:
Tree-based methods offer a flexible alternative to traditional linear and classification models by capturing nonlinear patterns and complex interactions common in social science data.
Applications in Social and Behavioral Science:
- Predicting the changing prevalence of child marriage in South Asia. 1
- Predicting HIV testing uptake among substance users in substance use disorder treatment programs. 2
- Predicting depression among rural and urban disabled elderly. 3
- Predicting violent or criminal behavior of juveniles. 4
References:
- Dietrich, S., Meysonnat, A., Rosales, F., Cebotari, V., & Gassmann, F. (2022). Economic development, weather shocks and child marriage in South Asia: A machine learning approach. PLoS ONE, 17(9), e0271373.
- Pan, Y., Liu, H., Metsch, L. R., & Feaster, D. J. (2017). Factors associated with HIV testing among participants from substance use disorder treatment programs in the US: A machine learning approach. AIDS and Behavior, 21(2), 534–546
- Xin, Y., & Ren, X. (2022). Predicting depression among rural and urban disabled elderly in China using a random forest classifier. BMC Psychiatry, 22, 118.
- Oh, G., Song, J., Park, H., & Na, C. (2022). Evaluation of Random Forest in Crime Prediction: Comparing Three-Layered Random Forest and Logistic Regression. Deviant Behavior, 43(9), 1036–1049.