Mathematics Research Seminar Series: Missing Data, Inference, and Causality: Applications to Law and Social Science

Mathematics Research Seminar Series
by Department of Mathematics in cooperation with Ateneo School of Government
Missing Data, Inference, and Causality: Applications to Law and Social Science

by Julian Morimoto

Date: Wednesday, 22 February 2023
Time: 4:00 - 5:00 pm
Venue: https://bit.ly/AdmuMathSeminar 

Abstract:
This talk discusses two papers at the intersection of statistics and law/social science. The first paper offers a new asymptotic theory on inference with missing data, which is more comprehensive than previous theories. It shows that as the sample size grows and the missing data decreases, inferences from partial data will converge to those obtained from complete data. Further, if the data are "Missing at Random," this convergence depends only on sample size. The implication is that missing data will have asymptotically no impact on parameter estimation and hypothesis testing. The prevalence of missing data in social science and legal datasets makes this result particularly valuable. This second paper presents a method for estimating a high dimensional, multi-factor Causal Risk Ratio. It builds on the Causal Risk Ratio developed by Zhao et al. 2022 by adding mechanisms that account for multiple treatments, different treatment orders, and high-dimensionality. The resulting method will be useful for legal researchers studying relative risk in high dimensional data, such as the relative risk of police violence among certain demographics or determining the most sympathetic forum for a legal dispute.

About the Speaker:
Julian Morimoto is currently a research fellow for the Fulbright Program and Harvard University, in collaboration with a local NGO, Initiatives for Dialogue and Empowerment through Alternative Legal Services (IDEALS). He is focusing on exploring human rights issues in the Philippines with a quantitative lens, and developing new quantitative methods to increase the capabilities of empirical legal scholarship. Most recently, he has published a new asymptotic theory of inference with missing data in Communications in Statistics — Theory and Methods. Prior to his fellowship, he was a Corporate Associate at Kirkland & Ellis LLP. He obtained his J.D. from Harvard Law School, where he worked on various empirical legal projects with faculty such as Kosuke Imai, Gary King, and Crystal Yang. Prior to law school, he obtained a B.A. in Mathematics from Case Western Reserve University