Causal inference under network interference: Methodological challenges and advances in public health applications

Vanessa McNealis (University of Glasgow)

Wednesday 14th May 14:00-15:00
Maths & Stats: 311B

Abstract

Much of the causal inference literature relies on the Stable Unit Treatment Value Assumption (SUTVA), which rules out interference between individuals. However, in many public health settings, this assumption does not hold. For example, in the context of a prevention program, one person’s vaccination status may indirectly influence the infection risk of their contacts within a social network. In addition to unmeasured confounding, estimating causal effects in these settings can be complicated by several factors, including homophily bias, unmeasured contextual confounding, autocorrelation, and uncertainty about the underlying network structure. In this talk, I’ll present recent methodological developments aimed at addressing these challenges, focusing in particular on work from my PhD and its application to data in education and public health. I’ll also touch on ongoing challenges and future directions in this area of research.

Add to your calendar

Download event information as iCalendar file (only this event)