A Proxy Variable View of Shared Confounding

                         Yixin Wang 1 David M. Blei 2

             Abstract                  confounder) are difficult to obtain, e.g., requiring expensive
    Causal inference from observational data           real-time tracking of activities. Lifestyle is necessarily an
    can be biased by unobserved confounders.           unobserved confounder.
    Confounders—the variables that affect both the        Here we focus on the setting where multiple t ...