Abstract Number:
1375
Submission Type:
Invited Paper Session
Participants:
Xiao-Li Meng (1), Hanti Lin (3), Christian Martin Hennig (2), Conor Mayo-Wilson (4), Ruobin Gong (5), Konstantin Genin (6), Peng Ding (7), Hanti Lin (3), Xiao-Li Meng (1)
Institutions:
(1) Harvard University, N/A, (2) Department of Statistical Sciences, University of Bologna, Bologna, Italy, (3) Philosophy Department, University of California, Davis, Davis, CA, (4) DEPARTMENT OF PHILOSOPHY, University of Washington, N/A, (5) Rutgers University, N/A, (6) University of Tübingen, N/A, (7) University of California-Berkeley, N/A
Chair:
Co-Organizer:
Hanti Lin
Philosophy Department, University of California, Davis
Session Organizer:
Speaker(s):
Hanti Lin
Philosophy Department, University of California, Davis
Session Description:
The power of statistics is not only in its methods and applications, but also in the philosophical underpinnings that guide its principles and practices. Philosophers, especially those specializing in the philosophy of statistics, offer invaluable insights, probing the very essence of the methodologies we employ, and raising profound questions about their implications and limitations. They not only dissect the history and conceptual evolution of statistical ideas but also illuminate paths for future advancements. Furthermore, philosophical statisticians, a unique blend of both worlds, have made pivotal contributions by weaving intricate tapestries of thought, merging abstract philosophical concepts with concrete statistical ideas and methodologies. This session isn't just an interaction-it's an imperative union of these intellectual titans. We seek to harness the combined might of statistical philosophers and philosophical statisticians, offering a platform where they can explore, challenge, and reinforce the foundations of data science. By melding these two worlds, we aim to push the boundaries of knowledge, ensuring a robust, well-rounded, and forward-thinking future for data science.
The proposed session adopts a unique format, with each philosopher's presentation (20 minutes) followed by a 15-min segment where a statistician will provide commentary and further exploration as well as lead Q&A from the audience.
Conor Mayo-Wilson of University of Washington presents "The Value of Information, Randomization, and Dominance Principles". He dives deep into Bayesian decision theory's foundations, encompassing all rational choices from everyday decisions to epistemic dilemmas. Mayo-Wilson critically assesses the traditional axioms of Archimedean and complete preference, which are contentious in philosophy. Conor Mayo-Wilson proposes a new decision theory that is weak enough to drop those two axioms but still strong enough to prove a theorem that vindicates the rationality of acquiring free information. Robin Gong of Rutgers University, known for her expertise in imprecise probability and differential privacy, will offer the statistical arguments.
The discussion then shifts from Bayesianism to frequentism. Konstantin Genin of University of Tübingen, in his talk "Reconsidering the Foundations of Experimental Design," highlights a gap: the randomized controlled trial lacks a frequentist justification, despite its widespread use in causal inference. Genin introduces a theorem to address this. Peng Ding of Berkeley, renowned for his profound contributions to statistical experiments and causal inference and, will share his theoretical insights.
Hanti Lin of UC Davis, in "The Rubin Causal Model Rebooted: A Fully Nondeterministic Version", delves into the influential Rubin causal model, prevalent in health and social sciences. He addresses a key assumption about counterfactuals regarding untreated patients' outcome, which has stirred philosophical debate. Lin proves a theorem to show that the Rubin causal model has an alternative formulation that is weak enough to drop that controversial assumption and strong enough to preserve its successful applications, including especially the estimation of LATE (the local average treatment effect). Xiao-Li Meng of Harvard, deeply invested in foundational statistics, will provide a comparative statistical treatment.
Sponsors:
IMS 1
Section on Bayesian Statistical Science 2
Section on Statistics in Epidemiology 3
Theme:
Statistics and Data Science: Informing Policy and Countering Misinformation
No
Applied
No
Estimated Audience Size
Small (<80)
I have read and understand that JSM participants must abide by the Participant Guidelines.
Yes
I understand and have communicated to my proposed speakers that JSM participants must register and pay the appropriate registration fee by June 1, 2024. The registration fee is nonrefundable.
I understand