CISE Seminar: Ruoxuan Xiong, Emory University

Date: November 4, 2022
Time: 3:00PM-4:00PM
Hybrid location: 8 Saint Mary’s Street (PHO 203) or Zoom

Ruoxuan Xiong,
Assistant Professor
Emory University

Design and analysis of panel data experiments

In this talk, we focus on the problem of designing an experiment that is conducted on a set of units, such as cities, over multiple time periods, such as weeks or months. This type of experiment is particularly useful to study treatments that have instantaneous effects as well as effects that may accumulate or attenuate when the treatment is maintained over multiple periods; it can also be useful when outcomes have strong dependencies on time, for example, due to seasonality. The design problem involves selecting treatment times for each unit in order to most precisely estimate the instantaneous and cumulative effects of the treatment. Optimization of the treatment assignment decisions can increase the precision of treatment effect estimates for a given sample size, or alternatively, decrease the opportunity cost of the experiment by reducing the required size or duration of the experiment. We start by considering a fixed-sample-size design where all treatment assignment decisions and their timing are determined prior to the start of the experiment. For this case, the optimization problem is an NP-hard integer program for which we provide a near-optimal solution. Next, we study an adaptive experimental design problem where treatment assignments and the experiment duration are determined during the experiment, where these decisions are updated after each period’s data is collected. We propose a new algorithm, Precision-Guided Adaptive Experiment (PGAE) algorithm, that addresses the challenges at both the design stage and at the stage of estimating the treatment effects, ensuring valid post-experiment inference accounting for the adaptive design. PGAE combines ideas from dynamic programming and sample splitting. Finally, we talk about the application of panel data experiments to test changes to marketplace algorithms, such as pricing and matching, on a ride-hailing platform.

Ruoxuan Xiong is an assistant professor in the Department of Quantitative Theory and Methods at Emory University. She completed a Ph.D. in Management Science and Engineering from Stanford University in 2020. She was a postdoctoral fellow at the Stanford Graduate School of Business from 2020 to 2021. Her research is at the intersection of econometrics and operations research, focusing on causal inference, experimental design and factor modeling, and with applications in finance and healthcare. Her work was awarded the Honorable Mention in the 2019 INFORMS George Nicholson Student Paper Competition, and was among the finalists of the 2020 MSOM Student Paper Competition.

Personal website: http://web.stanford.edu/~rxiong/

Faculty Host: Jinglong Zhao
Student Host: Ahmad Ghandi