Modeling Interference Using Experiment Roll-out

Experiments on online marketplaces and social networks suffer from interference, where the outcome of a unit is impacted by the treatment status of other units. We propose a framework for modeling interference using a ubiquitous deployment mechanism for experiments, staggered roll-out designs, which slowly increase the fraction of units exposed to the treatment to mitigate any unanticipated adverse side effects. Our main idea is to leverage the temporal variations in treatment assignments introduced by roll-outs to model the interference structure. We first present a set of model identification conditions under which the estimation of common estimands is possible and show how these conditions are aided by roll-out designs. Since there are often multiple competing models of interference in practice, we then develop a model selection method that evaluates models based on their ability to explain outcome variation observed along the roll-out. Through simulations, we show that our heuristic model selection method, Leave-One-Period-Out, outperforms other baselines. We conclude with a set of considerations, robustness checks, and potential limitations for practitioners wishing to use our framework.