Do Birds of a Feather Vote Together, or Is It Peer Influence?

Abstract

Why and when do legislators vote together on policy? Do legislators decide to vote similarly mainly because of social influence or similarity? I contribute to the literature by offering a social network theory of voting. It is argued that close social ties help to establish reputations and to aid the flow of information among peer legislators, leading to increased influence and increased rates of co-voting among them. I examine the relative effects of information exchange networks and social exchange networks on rates of co-voting, controlling for similarities among legislators. The data include legislative votes from 2010 to 2014 in the Brazilian Chamber of Deputies. The results suggest that social influence stemming from social relationships exert substantial effects on rates of congressional co-voting.

Natural Language Decompositions of Implicit Content Enable Better Text Representations

“…Here, we model legislator behavior using their speech (here, tweets), asking: does the similarity between legislators' propositions help explain the similarity of their voting behavior? Traditional theories of homophily (McPherson et al, 2001) suggest that shared properties (e.g., electoral geography, Clark and Caro, 2013) increase the likelihood that two legislators vote the same way -and many research questions have centered around such co-voting (Ringe et al, 2013;Peng et al, 2016;Wojcik, 2018). In preliminary experiments we found when modeling co-voting, text similarity between legislator was a valuable predictor of co-voting (Goel, 2023).…”

“…At a high level, we operationalize legislator homophily by measuring the similarities of their embedded speech Twitter. Following Ringe et al (2013) and Wojcik (2018), we model the log odds ratio of the co-voting rate be-tween a pair of legislators i, j using a mixed effects regression model, controlling for the random effects of both actors under consideration. The covote rate λ is the number of times the legislators vote the same way -yea or nay -divided by their total votes in common within a legislative session.…”

“…x ij is an n-dimensional feature vector, where each element is a similarity score that captures a type of relationship between legislators i and j. While these features have traditionally included state membership, party affiliation, or Twitter connections (Wojcik, 2018) or joint press releases (Desmarais et al, 2015); we consider the language similarity between pairs of legislators based on our proposed method.…”

The Party Replies: Examining Local Party Responsiveness to Prospective Campaign Volunteers

“…At the same time, partisan activists are not representative of the median voter and local party chairs (almost all volunteers) are more likely to reflect the characteristics of “intense policy demanders” than the characteristics of a typical local citizen (Bawn et al 2012; Hassell Hans et al, 2020). In such cases, party leaders may seek out members who are demographically or ideologically similar (Wojcik 2018).…”