2024 Berkeley Statistics Annual Research Symposium (BSTARS)

April 10, 2024
1:00 p.m. - 6:00 p.m.
University Club, Memorial Stadium - Berkeley, CA

The Berkeley Statistics Annual Research Symposium (BSTARS) surveys the latest research developments in the department, emphasizing possible applications to statistical problems encountered in industry. The 2024 conference features a faculty keynote from Distinguished Professor Michael I. Jordan, talks by faculty, students, and industry members, a student poster session, and a networking reception. 

Schedule

1-1:20 p.m. Registration
1:20-1:30 p.m. Introductory Remarks
1:30-2 p.m. Faculty Research Talk + Q&A - Assistant Professor Sam Pimentel: Match-adaptive randomization inference for optimal propensity score matching
2-2:30 p.m. Student Research Talk + Q&A - Sizhu Lu:  Dealing with intercurrent events in randomized trial
2:30-2:45 p.m. Break
2:45-3:45 p.m. Student Poster Session and Recruiter Tabling 
3:45-4 p.m. Break 
4-5 p.m. Keynote by Distinguished Professor Michael I. Jordan: An Alternative View on AI: Collaborative Learning, Statistical Incentives, and Social Welfare 
5-6 p.m. Reception and Recruiter Tabling

Keynote Speaker, Michael I. Jordan 

Michael Jordan is the Pehong Chen Distinguished Professor in Electrical Engineering and Computer Sciences and a professor of Statistics and Computer Sciences. His research focuses on the relationships between computation, statistics, and economics. He also works on applications in molecular biology, natural language processing, signal processing and mechanism design. 

Title: An Alternative View on AI: Collaborative Learning, Statistical Incentives, and Social Welfare 

Abstract: Artificial intelligence (AI) has focused on a paradigm in which intelligence inheres in a single, autonomous agent.  Social issues are entirely secondary in this paradigm.  When AI systems are deployed in social contexts, however, the overall design of such systems is often naive---a centralized entity provides services to passive agents and reaps the rewards.  Such a paradigm need not be the dominant paradigm for information technology.  In a broader framing, agents are active, they are cooperative, and they wish to obtain value from their participation in learning-based systems.  Agents may supply data and other resources to the system, only if it is in their interest to do so.  Critically, intelligence inheres as much in the overall system as it does in individual agents, be they humans or computers. This is a perspective that is familiar in the social sciences, and a key theme in my work is that of bringing economics into contact with foundational issues in computing and data sciences.  I'll emphasize some of the mathematical challenges that arise at this tripartite interface.

Faculty Speaker, Sam Pimentel 

Title: Match-adaptive randomization inference for optimal propensity score matching

Abstract: Matching is an appealing way to design observational studies because it mimics the data structure produced by stratified randomized trials, pairing treated individuals with similar controls. After matching, inference is often conducted using methods tailored for stratified randomized trials in which treatments are permuted within matched pairs. However, in observational studies, matched pairs are not predetermined before treatment; instead, they are constructed based on observed treatment status. This introduces a challenge as the permutation distributions used in standard inference methods do not account for the possibility that permuting treatments might lead to a different selection of matched pairs (Z-dependence). To address this issue, we propose a novel and computationally efficient algorithm that characterizes and enables sampling from the correct conditional distribution of treatment after optimal propensity score matching, accounting for Z-dependence. We show how this new procedure, called match-adaptive randomization inference, corrects for an anticonservative result in a well-known observational study investigating the impact of hormone replacement theory (HRT) on coronary heart disease.

Student Speaker, Sizhu Lu

Title: Dealing with intercurrent events in randomized trials

Abstract: In randomized controlled trials (RCTs), there are multiple types of intercurrent events competing with each other, and patients dropped out of the RCT due to the first intercurrent event that occurred, which challenges the analysis. We focus on two types of intercurrent events, adverse events which are often related to the treatment, and loss of follow-up which is not, using different strategies. We adopt both the composite variable and the principal stratification strategies to deal with the missing outcome due to adverse events, and adopt the hypothetical strategy to deal with that due to loss of followup. We provide nonparametric identification and semiparametric estimation theory for these strategies, construct doubly robust estimators, and apply the proposed method to analyze data from an RCT study as a real-world application.