Conference Agenda

This study investigates the causal impact of holding a seat in a representative political body on individuals' social capital, utilizing the unique case of the Chilean Constitutional Convention (2021–2022). Unlike previous processes, this convention adopted an innovative electoral design that lowered entry barriers for independent candidates, many of whom came from social movements and civil society organizations. This institutional shift created a distinctive political setting to analyze how transitioning from activism to institutional politics affects delegates' social networks and political capital.

Using a two-mode bipartite dataset that tracks affiliations with up to six social organizations and political parties before and after the convention, the study evaluates changes in delegates' embeddedness, cohesion, and brokerage. The analysis employs a difference-in-differences (DiD) design, comparing delegates who secured a seat (treatment group) with candidates who narrowly missed election (control group).

The study hypothesizes that institutional participation increases embeddedness and cohesion for all delegates but reduces brokerage for independents due to the loss of their original bridging roles. Conversely, party-affiliated delegates are expected to experience increased brokerage capacity, leveraging their pre-existing partisan networks. Findings from this research contribute to the literature on political socialization, social capital, and network dynamics in emerging institutional contexts.

Recent research on political communication highlights how digital platforms facilitate policy discourse, yet much of this work focuses on ideological polarization rather than specific issues. This study builds on social network analysis (SNA) research by examining whether Twitter interactions among U.S. politicians provide information about cluster actors who coalesce around shared policy interests (i.e., policy constellations). While prior studies have investigated partisanship and influencer dynamics, less is known about how politicians form networks around specific issues, particularly in drug policy. Using a dataset of several hundred American politicians and over 7,000 cannabis-related tweets, I construct a two-mode network based on retweets, hashtag co-occurrences, and direct mentions ("@") to analyze patterns of interaction and discourse. The first mode comprises ties between politicians and organizations, while the second mode ties politicians to policies and political issues. Preliminary findings suggest that a small subset of politicians disproportionately drives the conversation, acting as key brokers who amplify and shape policy discussions. At the same time, distinct clusters emerge, with politicians and organizations aligning around specific policy stances, regulatory frameworks, and advocacy efforts. This study extends existing work on digital political networks by offering an analytic model to uncover policy constellations through communication ties. These findings contribute to research on political communication, digital advocacy, and the role of social media in shaping contemporary policy debates.

This paper examines the changing structure of elite conflict in Weimar Germany by analyzing interactions among political parties in the German Reichstag between 1920 and 1932. We introduce a new database derived from digitized parliamentary proceedings, capturing all speeches, interjections, and reactions to interjections recorded by parliamentary stenographers. Each interaction is assigned a politeness score based on the type of interaction (e.g., applause, agreement, laughter, shouting) and modifying descriptors (e.g., “tumultuous” or “lively”). Using these scores, we first employ network analysis to trace changes in the structure of deference. We then construct a party-party structural equivalence matrix based on shared interaction patterns and arrange parties in a two-dimensional space using multidimensional scaling. Our findings reveal that the resulting party structure closely aligns with one derived from roll-call vote similarity, whereas speech content alone fails to reproduce this pattern. By examining relationships formed through repeated interactions on the floor, this study provides novel insights into elite competition and polarization in the Weimar Republic, contributing to broader debates on institutional instability in democratic systems under stress.

Understanding political polarization from a social network perspective is critical for diagnosing how policy preferences diffuse, cluster, and shape opinion dynamics. We present a novel method, Neural Network Nominate (NNN), which extends bipartite network analysis to measure ideological positions from user–proposal ties. Building on Thurstone’s paired comparisons and the NOMINATE framework, NNN embeds both individuals and policy proposals in a low-dimensional latent space, learned by a neural network classifier that predicts the outcome of each user’s preference ties. We examine large-scale datasets from three contexts—Chile, France, and Brazil—collecting over five million user–proposal choices. The resulting bipartite networks are then mapped to ideological spaces where users cluster by political leaning, and proposals align with the left–right dimension.

Our approach consistently attains strong predictive accuracy, including 68–70% for pairwise choice and up to 91% for self-reported left–right classification. We highlight how these embeddings illuminate generational differences in polarization: for instance, younger cohorts in France and Brazil form tighter ideological clusters, whereas young right-wing Chileans align with left-leaning policies. By focusing on revealed preferences, NNN mitigates the social desirability biases often found in surveys and the echo-chamber distortions in social media data. The method offers a scalable, transparent means to identify bridging proposals that might reduce ideological divides. This contribution underscores the value of a network-based approach for analyzing mass political ideology, fostering new directions in understanding—and potentially intervening in—polarized political landscapes. Our findings reveal critical insights, opening pathways for digital democracy interventions.