A fundamental problem in network experiments is selecting an appropriate experimental design in order to precisely estimate a given causal effect of interest. In fact, optimal rates of estimation remain unknown for essentially all causal effects in network experiments. In this work, we propose a general approach for constructing experiment designs under network interference with the goal of precisely estimating a pre-specified causal effect. A central aspect of our approach is the notion of a conflict graph, which captures the fundamental unobservability associated with the casual effect and the underlying network. We refer to our experimental design as the Conflict Graph Design. In order to estimate effects, we propose a modified Horvitz--Thompson estimator. We show that its variance under the Conflict Graph Design is bounded as $O(\lambda(H) / n )$, where $\lambda(H)$ is the largest eigenvalue of the adjacency matrix of the conflict graph. These rates depend on both the underlying network and the particular causal effect under investigation. Not only does this yield the best known rates of estimation for several well-studied causal effects (e.g. the global and direct effects) but it also provides new methods for effects which have received less attention from the perspective of experiment design (e.g. spill-over effects). Our results corroborate two implicitly understood points in the literature: (1) that in order to increase precision, experiment designs should be tailored to specific causal effects of interest and (2) that "more local" effects are easier to estimate than "more global" effects. In addition to point estimation, we construct conservative variance estimators which facilitate the construction of asymptotically valid confidence intervals for the casual effect of interest.
翻译:暂无翻译