doi:10.1093/pan/mps016 Estimating the Causal Effects of Social Interaction with Endogenous Networks Jon C. Rogowski Department of Political Science, University of Chicago, 5828 S. University Ave., Pick Hall, 4th Floor, Chicago, IL 60637 e-mail: jrogowski@uchicago.edu (corresponding author) Betsy Sinclair Department of Political Science, University of Chicago, 5828 S. University Ave., Pick Hall 415, Chicago, IL 60637 e-mail: betsy@uchicago.edu Edited by Neal Beck Identifying causal effects attributable to network membership is a key challenge in empirical studies of social networks. In this article, we examine the consequences of endogeneity for inferences about the effects of networks on network members’ behavior. Using the House office lottery (in which newly elected members select their office spaces in a randomly chosen order) as an instrumental variable to estimate the causal impact of legislative networks on roll call behavior and cosponsorship decisions in the 105th–112th Houses, we find no evidence that office proximity affects patterns of legislative behavior. These results contrast with decades of congressional scholarship and recent empirical studies. Our analysis demonstrates the importance of accounting for selection processes and omitted variables in estimating the causal impact of networks. 1 Introduction Scholarly interest in social networks has increased dramatically in recent years. A particularly prominent line of inquiry centers on the capacity of networks to transmit information and influence, often shaping some behavioral outcome of interest among a target population. For instance, legislative scholars have invoked networks to explain how influence is transmitted between legislators. According to David Truman (1956, 1024), such networks are often ‘‘created by the proximity of offices or residences in Washington.’’ Young’s (1966) classic qualitative study of the nascent congressional years shows that living in congres- sional boardinghouses helped legislators form networks with other geographically dispersed members, and recent work by Masket (2008) shows that legislators in the California Assembly voted together more frequently when they shared desks on the chamber floor. Identification of causal effects that result from network processes, however, is often difficult. The endo- geneity that often accompanies network membership makes it difficult to separate the effects of the net- work from the ways in which network membership was determined. This problem is especially rampant in studies of political attitudes and behavior, in which researchers often have few alternatives but to use observational data to examine how one’s network peers affect her behavior. Without information about how individuals select into such networks, however, it is nearly impossible to make causal inferences about network effects (see Manski 1993). Authors’ note: We thank Sara Brady, James Fowler, Keith Poole, Keith Krehbiel, and Jonathan Woon for providing data used in this project. We thank Kevin Collins, David Darfomal, James Fowler, Rob Franzese, Donald Green, Greg Huber, Greg Koger, Seth Masket, Eric Oliver, Jennifer Victor, Nick Weller, and participants of the Fourth Annual Political Networks Conference and the Harris School Political Economy Lunch for lively discussion and helpful comments. The editors and two anonymous reviewers provided especially conscientious feedback. We are most indebted to John J. Pitney, who used his own social networks to help us locate the House lottery data. This research protocol (H10171) was approved by the University of Chicago IRB. Supplementary materials for this article are available on the Political Analysis Web site. Ó The Author 2012. Published by Oxford University Press on behalf of the Society for Political Methodology. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com Political Analysis Advance Access published May 9, 2012 by guest on June 30, 2012 http://pan.oxfordjournals.org/ Downloaded from