How would you design an experiment to measure the impact of a new driver incentive program when network spillover effects are present?
