Biostatistics Seminar guest speaker, David Choi, Carnegie Mellon University, will present, "Estimation of Monotone Treatment Effects in Network Experiments".
Randomized experiments on social networks pose statistical challenges, due to the possibility of interference (such as peer influence) between units. To find rigorous confidence intervals on the average treatment effect in such settings, one typically must model the underlying social network -- "who can influence whom", how such effects might combine, and whether they can cascade over long distances. In many settings, this may be an unreasonable modeling burden. As an alternative, we propose new methods for finding confidence intervals on the attributable treatment effect. These methods do not make assumptions on the structure of the underlying social network, but instead require an identifying assumption that is similar to requiring nonnegative treatment effects; for example, assuming that a vaccine does not increase the risk of catching a disease, either directly or indirectly through vaccinated peers. Network or spatial information can be used to customize the test statistic; in principle, this can increase power to detect spillovers without making further assumptions on the data generating process.