Sooahn Shin
Sooahn Shin
I am a Postdoctoral Associate in the Department of Political Science at MIT, working with Naoki Egami. My research develops statistical methods for studying causal effects and latent concepts, with applications to algorithm-assisted decision-making, LLM-generated labels, and the measurement of political ideology. Before joining MIT, I was a Data Scientist at Airbnb, where I worked on causal inference problems in marketing research. I received my PhD from the Department of Government at Harvard University in 2025, where I was advised by Kosuke Imai. In July 2027, I will join the Department of Political Science at the University of North Carolina at Chapel Hill as an Assistant Professor.
Publications
-
"Does AI help humans make better decisions? A statistical evaluation framework for experimental and observational studies"
Proceedings of the National Academy of Sciences, 2025. -
"ℓ1-based Bayesian Ideal Point Model for Multidimensional Politics"
Journal of the American Statistical Association, 2025. -
"Experimental Evaluation of Algorithm-Assisted Human Decision-Making: Application to Pretrial Public Safety Assessment"
(with discussion), Journal of the Royal Statistical Society, Series A (Statistics in Society), 2023. -
"Bayesian Methods in Political Science"
Luigi Curini and Robert Franzese eds., The SAGE Handbook of Research Methods in Political Science & International Relations, 2020.
Working Papers
Ideal Point Estimation Beyond a Single Dimension
Impact of AI on Human Decisions
-
"Triage Score: A Counterfactual Risk Assessment Instrument"
Causal Inference Using Panel Data with Missingness
-
"Difference-in-differences Design with Outcomes Missing Not at Random"
-
"Synthetic Control Methods with Missing Pre-treatment Outcomes"
Teaching Fellow
At Harvard, I served as a teaching fellow for PhD-level courses in causal inference, Bayesian statistics, and machine learning, all part of the Government Department Methods Sequence.
-
Causal Inference with Applications, Fall 2023 [course website]
Gov 2003,
-
Applied Bayesian Statistics for the Social Sciences, Spring 2023 [syllabus]
Gov 2017,
-
Data Science for the Social Sciences, Fall 2022 [course
website]
Gov 50 (undergraduate),
-
Introduction to Machine Learning, Spring 2022
Gov 2018,
-
Causal Inference with Applications, Fall 2021 [section notes]
Gov 2003,
Software
-
aihuman (Available on CRAN)
R package written in Rcpp for an experimental evaluation of causal impacts of algorithmic recommendations on human decisions developed by Imai et al. (JRSS, 2023) and Ben-Michael et al. (PNAS, 2025). -
issueirt
R package for issue specific ideal point estimation developed by Shin (working paper). It uses Stan. -
l1ideal
R package for ℓ1 norm based multidimensional ideal point estimation developed by Shin et al. (JASA, 2025). It uses multivariate slice sampling for the estimation.