The Bayesian causal forest model: regularization, confounding, and heterogeneous effects
P. Richard Hahn, Arizona State University
January 29, 2021- 12:00-1:15pm
https://uconn-cmr.webex.com/uconn-cmr/j.php?MTID=mc19e545b14cc3a980ffc36760a5ce5f4
In this talk, I will describe recent work on Bayesian supervised learning for conditional average treatment effects. I will motivate the proposed Bayesian causal forest model in terms of fixing two specific flaws with previous approaches. One, our model allows for direct regularization of the treatment effect function, providing lower variance estimates of heterogeneous treatment effects. Two, by including an estimate of the propensity score as a control variable in our model we mitigate a phenomenon called "regularization induced confounding" that leads to substantial bias in previous approaches. I will conclude with a detailed discussion of designing simulation studies to systematically investigate and validate machine learning models for causal inference.
Note: Dr. Hahn may also talk about this tutorial a bit: https://math.la.asu.edu/~prhahn/xbcf_demo.html
Bio: Professor P. Richard Hahn has a B.A. in Philosophy of Science from Columbia University and earned his PhD in Statistics from Duke University in 2011. He taught at University of Chicago Booth School of Business for seven years before joining the School of Mathematical and Statistical Sciences at Arizona State University in 2017. His research lies at the intersection of machine learning and causal inference, specifically tree based regression methods for estimating heterogeneous treatment effects. Other research interests include latent variable models and statistical decision theory. He enjoys road trips in the mountain southwest with his family and riding and working on bicycles.
ONLINE INTERDISCIPLINARY SEMINARS ON STATISTICAL METHODOLOGY FOR SOCIAL AND BEHAVIORAL RESEARCH Talk
The online interdisciplinary seminars on statistical methodology for social and behavioral research is supported by the department of statistics and the department of education psychology in the University of Connecticut (UCONN), the Statistical and Applied Mathematical Sciences Institute (SAMSI) and the New England Statistical Society (NESS). The seminar is held online via WebEx and anyone in the world can join and it is scheduled monthly on Friday noon. The aims of the seminar are to promote the connection between the statistics and social/behavioral science communities and to encourage more graduate students to participate in the interdisciplinary research.
For announcements and WebEx live streaming links, please contact Tracy Burke (tracy.burke@uconn.edu).For questions related to the seminars, please feel free to contact organizers (Dr. Xiaojing Wang (xiaojing.wang@uconn.edu) and Dr. D. Betsy McCoach (betsy.mccoach@uconn.edu) ).
Full WEBEX Info for this talk: (Friday, Jan 29, 2021 12:00 pm)
https://uconn-cmr.webex.com/uconn-cmr/j.php?MTID=mc19e545b14cc3a980ffc36760a5ce5f4
Meeting number: 120 084 6708
Password: STATRMME
Join by video system
Dial 1200846708@uconn-cmr.webex.com
You can also dial 173.243.2.68 and enter your meeting number.
Join by phone
+1-415-655-0002 US Toll
Access code: 120 084 6708
For more information, contact: Betsy at betsy@uconn.edu