Nova School of Business and Economics and NOVAFRICA
Working Paper No 1903
The use of lab in the field experiments has increased dramatically, given benefits of studying relevant populations. Conducted in environments where researchers must relinquish the control a standard laboratory offers, they raise the specter of communication from past to future participants, posing problems for inference. While researchers may take steps to avoid spillovers, little is known about the mechanics of such spillovers in lab in the field settings, nor to what extent they may bias inference. In rural villages participating in public goods games in Rwanda, I recover estimates of these spillovers by matching villages on all available pre-study observables, comparing those with and without communication opportunities. I find communication led to substantial unanticipated increases in cooperation, driven by conditional cooperators. I conclude with advice to manage potential bias from spillovers.