Events Calendar

Yearly View
View as List
Monthly View
By Month
Weekly View
By Week
Daily View
Download as iCal file
SDS Seminar: Cory Zigler
Tuesday, May 30, 2017, 11:00am - 12:00pm
Contact stat.admin[@]austin[dot]utexas[dot]edu 

Cory Zigler (Department of Biostatistics, Harvard T.H. Chan School of Public Health) 

Title: "Causal Inference with Interference for Evaluating Pollution Regulations"

Abstract: A fundamental feature of evaluating causal health effects of air quality regulations is that air pollution moves through space, rendering an individual’s health outcome a function of treatments (or exposures) originating from multiple locations. This gives rise to what is known in the statistics literature as interference, where potential outcomes for a given unit are defined in part based on treatments applied to other units. Virtually all formalizations of potential outcomes methods for causal inference assume that interference is not present. This talk introduces interference in studies of air pollution by briefly describing work on new causal estimators in settings of so-called partial interference, which occurs when interference is present within, but not between, distinct clusters of observations. It then outlines causal inference in a new type of interference setting on a bipartite network where interventions and outcomes are defined and measured on two distinct sets of observational units and interference arises due to complex exposure patterns such as those dictated physical-chemical atmospheric processes of pollution transport. We term this setting bipartite causal inference with interference. The methods are discussed in the context of evaluating health impacts of pollution emissions in the Northeastern US and in the context of evaluating effects of regulatory interventions implemented at coal-fired power plants. The discussion is designed to illustrate both the opportunities and the challenges to increasing the role of statistics for evaluating air quality regulatory policies.

Location: GDC 4.302