Birds of the Same Feather Tweet Together. Bayesian Ideal Point Estimation Using Twitter Data

Politicians and citizens increasingly engage in political conversations on social media outlets such as Twitter. In this paper I show that the structure of the social networks in which they are embedded can be a source of information about their ideological positions. Under the assumption that social networks are homophilic, I develop a Bayesian Spatial Following model that considers ideology as a latent variable, whose value can be inferred by examining which politics actors each user is following. is method allows us to estimate ideology for more actors than any existing alternative, at any point in time and across many polities. I apply this method to estimate ideal points for a large sample of both elite and mass public Twitter users in the US and five European countries. e estimated positions of legislators and political parties replicate conventional measures of ideology. e method is also able to successfully classify individuals who state their political preferences publicly and a sample of users matched with their party registration records. To illustrate the potential contribution of these estimates, I examine the extent to which online behavior during the  US presidential election campaign is clustered along ideological lines. *I would like to thank Jonathan Nagler, Joshua Tucker, Nick Beauchamp, Neal Beck, Ken Benoit, Richard Bonneau, Patrick Egan, Adam Harris, John Jost, Franziska Keller, Michael Laver, Alan Potter, Gaurav Sood, Chris Tausanovitch, Shana Warren, and two anonymous reviewers for helpful comments and discussions. e present work has been supported by the National Science Foundation (Award ) and the “La Caixa” Fellowship Program.  Introduction Measuring politicians’ and voters’ policy positions is a relevant, yet complex, scientific endeavor. Studies of electoral behavior, government formation, and party competition require systematic information on the placement of key political actors and voters on the relevant policy dimensions. e development of methods to estimate such positions, usually in a single latent dimension characterized as “ideology” (Poole and Rosenthal, ; Clinton, Jackman and Rivers, ; Shor, Berry and McCarty, ; Bonica, b; Jessee, ), represents one of the most important methodological contributions to political science in the past two decades. However, most studies estimate ideal points for legislators only. When the analysis also includes voters, it is done at the expense of strong “bridging” assumptions (Jessee, ), or only for selfselected population groups (Bonica, ). ere is also little work on cross-national ideological estimation (Lo, Proksch and Gschwend, ). Most importantly, given the sparse nature of the data (roll-call votes or contributions) and its costly collection (survey data), current measurement methods generate ideal points that are essentially static in the short-run. In this paper I show that using Twitter networks as a source of information about policy positions has the potential to solve these difficulties. Twitter has become one of the most important communication arenas in daily politics. Initially conceived as a website to share personal status updates, it now has more than  million monthly active users worldwide, including  of all online Americans. One distinct characteristic of this online social network is the presence of not only ordinary citizens, but also political actors. Virtually every legislator, political party, and candidate in developed democracies has an active Twitter account. Independent of their offline identities, they all interact within the same symbolic framework, using similar language in messages of identical length. Most importantly, they are embedded in a common social network. is opens the possibility of estimating ideological positions of all users on a common scale, which would allow for meaningful comparisons of voters’ and legislators’ ideal points. Source: Twitter’s Official Twitter Account, December , . [link] Source: e Pew Research Center’s Internet & American Life Project, August . [link]  e use of Twitter data presents three additional advantages over other sources of information about preferences. First, the large number of active users on this social networking site can be exploited to estimate highly precise ideal points for politicians, if we consider users as “experts” who are “rating” elites through their decisions of who to follow. Second, the structure of this network is far from static, which can facilitate the estimation of highly granular dynamic ideal points in real time. ird, it is possible to link Twitter profiles to other data through name identification, which provides interesting ways to examine differences between private and public political behavior. is series of advantages comes at the expense of one important limitation. Twitter users are not a representative sample of the voting age population. is can represent a difficulty in the context of studies about mass attitudes and behavior, but not for the method I present in this paper. Citizens who discuss politics on Twitter are more likely to be educated and politically interested, and that makes them a particularly useful source of information about elites’ ideology. is method relies on the characteristics of the social ties that Twitter users develop with each other and, in particular, with the political actors (politicians, think tanks, news outlets, and others) they decide to follow. I argue that valid policy positions for ordinary users and political actors can be inferred from the structure of the “following” links across these two sets of Twitter users. e decision to follow is considered a costly signal that provides information about Twitter users’ perceptions of both their ideological location and that of political accounts. Unlike other studies that estimate political ideology using social media data (Conover et al., ; King, Orlando and Sparks, ; Boutet et al., ), I am able to estimate ideal points, with standard errors, on a continuous scale, for all types of active Twitter users, across different countries. To validate the method, I estimate the ideological positions of legislators, political parties, and a large sample of active users in the US and five European countries. eir estimated ideal points replicate conventional measures of ideology. is method represents an additional measurement tool that can be used to estimate ideology, an important quantity of interest in political science, for a larger set of political actors and individuals than any other method before. To illustrate a potential use of these estimates, I examine the extent to which online behavior during the  US presidential election campaign is clustered  along ideological lines, finding support for the so-called “echo-chamber” theory and high levels of political polarization at the mass level.  Ideal Point Estimation Using Twitter Data . Previous Studies ere is a limited but increasing literature on the measurement of users’ attributes in social media, particularly in the field of computer science. Despite ideology being one of the key predictors of political behavior, its measurement through social media data has only been examined in a handful of studies. ese studies have relied on three different sources of information to infer Twitter users’ ideology. First, Conover et al. () focus on the structure of the conversation on Twitter: who replies to whom, and who retweets whose messages. Using a community detection algorithm, they find two segregated political communities in the US, which they identify as Democrats and Republicans. Second, Boutet et al. () argue that the number of tweets referring to a British political party sent by each user before the  elections are a good predictor of his or her party identification. However, Pennacchiotti and Popescu () and Al Zamal, Liu and Ruths () have found that the inference accuracy of these two sources of information is outperformed by a machine learning algorithm based on a user’s social network properties. In particular, their results show that the network of friends (who each individual follows on Twitter) allows us to infer political orientation even in the absence of any information about the user. Similarly, the only political science study (to my knowledge) that aims at measuring ideology (King, Orlando and Sparks, ) uses this type of information. ese authors apply a data-reduction technique to the complete network of followers of the U.S. Congress, and find that their estimates of the ideology of its members are highly correlated Ideology is defined here as themain policy dimension that articulates political competition: “a line whose le end is understood to reflect an extremely liberal position and whose right end corresponds to extreme conservatism.” (Bafumi et al., , p.) Each individual’s ideal point or policy preference corresponds to their position on this scale. See also Poole and Rosenthal (, ).