Doctoral Dissertation Research: Identifying Individuals' Causal Effects (Peer Effects) on Participation in Collective Action

Harvard University, Cambridge MA

Investigators

Abstract

General Abstract: Collective action, or cooperative behavior, is a central concept in economics and political science. Despite this significance, there is relatively little evidence regarding how individuals cause others to participate in cooperative ventures. This study analyzes cell phone data: the time and location of sent/received text messages and calls. The analysis tells us how individuals cause their peers to join in collective action; how this might encourage broader cooperative behavior; and how these processes differ from that of non-collective action areas. Additionally, this research introduces the field to cell phone metadata -- an exceptionally rich data source that can help provide insight into a range of political and social behavior. In particular, it contributes to studies of how technologically-enabled communication affects economic and political behaviors, and situates the research within one of the NSF's Ten Big Ideas, studying work at the Human-Technology Frontier. Technical Abstract: In this proposal, the PI seeks to investigate the causal mechanisms that lead to successful collective action. Prominent theories of collective action argue that individuals choose to participate in an instance of collective action based on two sets of factors: their own preferences and opportunity costs (independent from their peers) and/or on estimates of how many other people will participate, which is only marginally affected by peers' decisions. These approaches are contradicted by a growing empirical literature in behavioral economics, which finds strong evidence that individuals determine their immediate peers' economic choices in many other non-collective action domains. The PI contributes to this line of research in two important areas. First, the project proposes to develop a large-scale empirical study of how individuals lead others to participate in collective action. Second, the research will show how individual-level effects spread through communication networks, manifesting in widespread behaviors. This dissertation takes analyzes several country-years of anonymized cell phone metadata. This data records individuals' communication with other individuals before and during instances of collective action as well as their geographic movement (i.e. call/SMS sender, recipient, timestamp, and location). Using this data, the PI exploits a series of natural experiments to assess (1) individuals' causal effects on their peers' participation in collective action; (2) how these individual causal effects aggregate into collective action; and (3) how these causal processes differ from individuals? causal effects on their peers? economic decisions in non-collective action domains.