Robust causal inference in networks: addressing misspecification, proxy observations, and transportability

A prominent assumption when studying treatment effects is the no-interference assumption, stating that treatment applied to one unit does not impact other units. However, interference -- an umbrella term for spillover, contagion, peer effects, and related phenomena -- is present in many settings. Relaxing the no-interference assumption is often accompanied by an assumed interference structure, commonly represented by a network. Various methods have been developed to address network interference under design-based, frequentist, or Bayesian perspectives.
A key assumption shared by recently developed methods is that the network is given and correctly specified. In this talk, I will present the implications of violations of these assumptions and offer solutions under varied probabilistic regimes.
To this end, if time permits, we will consider the following three cases: (1) the network is fixed but misspecified (2) the network might be random, but only proxy observations are available, and (3) the network is observed, but interest lies in causal effects in a new population under a different, possibly unknown, network.  

Joined work with Bar Weinstein and Nir Aviv