Abstract:
In a field experiment, I randomly assign college students to a low-cost behavioral intervention where they receive advice and prompts to make plans to study for an exam. Using detailed network data and exogenous variation in the exposure to the intervention, I causally estimate spillover effects on academic outcomes that are transmitted through study partners. I find positive spillover effects on untreated students' usage of an online learning applet. The average spillover effect on untreated is about 30% of the direct treatment effect. I use tie strength and network position to argue that a contagious use of the applet is a more plausible mechanism than information sharing. However, I find negative spillover effects on treated students' applet usage. Possible explanations can be preferences for information scarcity and coordination costs between treated students as they have already created plans. Despite finding spillover effects on study behaviors, I do not find spillover effects on performance outcomes measured by exam scores and course grades. This evidence highlights that spillover effects in naturally formed peer groups can interact with the intervention. Policy makers need to carefully engineer intervention targeting.