randomized experiments

This paper proposes a method to estimate spillover effects of a random treatment using a nonrandom sample of individuals for whom the analyst can observe their outcomes.