The political information ecosystem has evolved significantly, with the internet and social media fundamentally transforming how political content is consumed and how beliefs are formed. Social media platforms have become central to political engagement, reshaping access to information and patterns of interaction. While these platforms democratize access, they also present challenges. Algorithmic filtering and personalized content creation contribute to echo chambers and belief polarization. Although only a small subset of individuals consumes highly distorted information, social media amplifies bias and misinformation, reinforcing users’ preexisting beliefs through content curation and recommendation algorithms. This study examines how these dynamics contribute to political polarization and the spread of misinformation within social media environments.

This research investigates how cognitive biases influence evaluations of information credibility and how social networks shape belief polarization and resilience against misleading information. It also explores whether social networks mitigate the risks posed by misleading information. The study employs agent-based modeling (ABM) to simulate and analyze individual behaviors, social networks, and information dissemination, emphasizing that political learning is nonlinear and shaped by complex interactions.

ABM effectively models the recursive nature of political learning, capturing emergent patterns from micro-level interactions among agents (e.g., citizens, media, and opinion leaders). This approach allows for an in-depth analysis of how cognitive biases and network structures influence political information processing. The study also examines how social networks may counter misinformation, demonstrating that strong networks with active communication can foster consensus, even in the presence of adversarial actors spreading disinformation.

The study begins by modeling individual-level political learning, incorporating the effects of cognitive biases on source credibility evaluation. It also examines how citizens navigate the 'exploration' versus 'exploitation' dilemma in the learning process. By placing agents within different social network structures, from elite-dominated information markets to homophilous citizen networks, the study explores conditions under which misleading information can be curtailed. Additionally, it introduces a novel type of disinformation provider, whose goal is to disrupt citizens’ understanding of political reality and hinder consensus-building.

Simulation results indicate that second-hand learning within social networks can mitigate belief polarization, particularly in homophilous networks, where individuals share similar political beliefs. When disruptive agents attempt to manipulate information, citizens within well-connected, resilient networks demonstrate greater resistance to disinformation. However, in fragmented networks, individuals are more vulnerable to misleading content. These findings underscore the importance of social structures in shaping belief systems and public opinion.

In conclusion, this study contributes to political communication and computational social science by demonstrating how agent-based simulations provide insights into the complex interactions of individual behaviors and macro-level network effects. It also highlights practical implications for designing interventions to counter misinformation and strengthen public discourse in the digital age. The research emphasizes the need to integrate psychological and social dimensions to foster a well-informed public capable of navigating the modern political information ecosystem.

That the international economic and political systems are interdependent, multilayered, dynamic networks is well understood—theoretically. However, the complexity of assembling such networks means that the research landscape remains fragmented, with focus on modeling the contributions of specific elements of the system, such as the geopolitical implications of the supply chain network for final goods (e.g., fabricated microprocessors) or of single classes of actors (e.g., firms local to countries or regions, states entering into alliances). These limitations are understandable: expanding the scope requires identifying, extracting, and modeling data that is historically outside the purview of social sciences. However, in a world of globe-spanning interconnected supply chains, the effort allows scholars to understand the emergent topology of the network of flows and dependencies.

We push the frontier of existing work by extracting multiple levels in the global supply chain of 15 materials central to a geopolitically important supply chain (advanced microelectronics). We build a multiplex data network by tracing these materials from extraction through multiple processing steps to trade of the final products. We use this data to produce a dynamic network model that captures the evolution of political leverage within the global microelectronics supply chain—a critical domain for international relations. In doing so, we identify how states exploit strategic positions within these networks over time in response to external political pressures and opportunities.

Our approach extends beyond the domain of supply chain mapping to model the relational processes through which geopolitical power is exercised and contested. By comparing our dynamic network measures against traditional models of interdependence, we demonstrate how advances in data accessibility and network methodology allow scholars to produce deeper insights into evolving patterns of political leverage. We can extend our methodological framework to other contexts in which complex political dynamics emerge from multiple interconnected networks.

I propose an interpretable measure of multi-group polarization for binary choice profiles from a first principle and minimum assumptions. The proposed measure is explicitly defined as the average probability that a single vote helps infer the voter's group label. This is in a stark contrast with current practices of using group polarization measures that are computed from estimates obtained by complicated estimation procedures with strong assumptions on data generating processes, relying on each researcher's discretion to determine an estimation model, the number of dimensions, and the scale of the measure. As a result, a direct interpretation of the level of polarization and comparison between data sets have been difficult. Thanks to its tractability, I derive the proposed measure's characteristics and examine its performance over axiomatic features of group polarization. The proposed measure uncovers aggregate patterns of party polarization in the United States House of Representatives during the modern and contemporary Congress.

Social media platforms like Twitter (now X) have provided an interesting arena for debates about topics like China, because producers of knowledge (political experts) and their subjects (politicians and governments) interact with each other on a relatively equal footing, limited and enabled by the same platform affordances. There has been an upsurge of research on Chinese government activity on social media lately, but no similar interest in the activity of China Watchers.

But the activity of China experts on Twitter may grant insights into how expert interactions and debates contribute to the knowledge production about authoritarian regimes and how political experts interact and compete with other actors who try to shape that discourse. This paper thus presents a description of the Twitter conversation of almost 1000 China experts worldwide since 2008, analyzing over 5 million posts. It reports on their online interaction with each other, with the general audience, but also with roughly 1000 accounts controlled or associated with government of the PRC. The China Watchers were selected through a snowballing nomination process, in which China experts were asked to nominate peers. The government accounts were included if 10 Chinese government accounts already in the list followed the given account.

A monthly sentiment analysis shows that sentiment towards China has steadily soured among China experts over time, with a predictable drop during the early months of the Covid-19 pandemic. The same measure among PRC government accounts, once very similar back in 2010, has soared. A diffusion analysis indicates that this sentiment towards China among China Watchers is contagious along the nomination ties.

Interactions between China Watchers and their government subjects on Twitter are surprisingly frequent. While it is more common for China Watchers to reply or mention Chinese government accounts around 16’000 times in this dataset, the latter do reciprocate. However, they appear to mention and reply to them mainly to express criticism with the China Watcher community: the sentiment expressed in these tweets is significantly more negative than those in their other tweets.

China Watchers that interact with the Chinese government are a specific group of China experts: they tend to be journalists, public intellectuals writing Op-Eds or individuals who have also held positions in the government or served as advisers. There is in general only a weak correlation between standing within the China Watcher community (i.e. the number of nominations received) and being mentioned or replied to by the Chinese government.

The paper also examines the role of accounts that appear to be ordinary Chinese or foreign citizens, but for some reason are being followed by at least 10 official accounts. They are retweeted frequently by official Chinese government accounts, who may try to pushed them as alternative China experts.

Finally, it examines changes since Elon Musk has taken over Twitter: have China Watchers remained on the platform or move on to other platforms? Do they post differently and interact with different individuals, and has their interaction in particular with the PRC accounts changed in any way?

Political communication on social media is shaped by networked interactions between elites and the public. While cross-partisan discussions have the potential to foster dialogue, they also risk reinforcing polarization through incivility and selective amplification. We leverage a large-scale longitudinal dataset of over 1.1 million interactions between U.S. legislators and their audiences on Twitter/X from 2020 to 2021 to examine the networked dynamics of cross-party engagement. By analyzing retweets, replies, and mentions as evolving relational structures, we reveal asymmetries in how political actors and audiences engage across party lines. Our findings indicate that Republican legislators experience increased engagement when amplifying Democratic content, while Democrats see a decline in visibility for similar actions, highlighting party-specific audience dynamics. We further analyze the posting styles associated with the visibility of cross-talks. For instance, incivility in cross-party replies increases visibility for Republicans by 12%. Furthermore, we reveal how engagement on cross-party interactions influences the likelihood of future bipartisan discourse, suggesting that political elites may adjust their strategies based on audience responses. Our study offers one of the first longitudinal examinations of cross-party interactions, providing new insights into the self-reinforcing nature of political networks. By integrating temporal analysis, engagement dynamics, and discourse quality, we contribute to the understanding of how political polarization manifests through networked interactions and identify key mechanisms that shape the long-term trajectory of bipartisan communication. Our findings have critical implications for platform governance, media strategies, and efforts to foster more constructive political discourse in online spaces.

Recent European elections have witnessed a rightward shift, with populist and far-right parties gaining ground. Many explain this shift through demographic polarization. It pertains to a growing divide between urban and rural voters, younger and older generations, and those with differing education levels, driven by economic and migration concerns. However, demographic polarization goes beyond mere group identity; it is shaped by lived experiences. Where and how people live, work, and whom they interact with influence their views on social change. This study expands the traditional sociodemographic lens.

We hypothesize that, beyond voters' sociodemographic profiles, it is the social networks they are embedded in, that shape political behavior, specifically turnout and polarization. The key characteristics of these networks include size, the degree of closure, and composition with respect to socioeconomic status, education, and migration background. Additionally, we examine the spatial dispersion of an individual's social environment. We focus this analysis on the 2023 Dutch general elections. We enrich neighborhood-level election results with insights into the social networks of residents. Social networks are sourced from Dutch registers covering the country's population and providing information on one’s socio-demographic profile as well as comprehensive mapping of social networks. The relational aspect of registers sheds light on formal links such as kinship, neighbors, classmates, colleagues, and household members. By integrating population-scale network analysis with electoral outcomes, we aim to uncover how social structures mediate the relationship between demographic factors and political behavior, offering a nuanced understanding of polarization dynamics in contemporary European democracies.

LGBTQ+ activism has a long history in the U.S., but it is not always clear who participates and why, especially given the risks involved. Recent work investigates individual drivers of participation in LGBTQ+ activism and, consistent with other groups, finds that experiences of victimization increase the likelihood that LGBTQ+ people participate in protest. Building on that work, this study identifies network drivers of participation in LGBTQ+ activism, including political marches, demonstrations, and protests. We use novel data from Wave III of the LGBTQ+ Social Networks, Aging, and Policy Study (QSNAPS), which includes 981 midlife and older LGBTQ+ adults who report on the characteristics and political behaviors of 13,412 network ties. Data were collected throughout 2023 from adults in 22 US states. We test the hypothesis that network characteristics and perceived alter behaviors are associated with respondent’s participation in LGBTQ+ activism. Specifically, we observe whether the respondent discusses politics or differs in political opinion from each alter and alters who the respondent thinks are also LGBTQ+, are kin, or who attended a protest, march, or political demonstration in the last year. Regression analyses show that greater discussion of politics is associated with a higher share of LGBTQ+ ties and a higher share of ties who know the respondent’s LGBTQ+ status in one’s network. Respondent’s participation in LGBTQ-related but not general protest is significantly associated with protest participation within one’s network. These results contribute to understandings of how networks shape political activism among a minoritized, politically targeted group in the US context.

Everyday political talk is key to deliberative democracies, and political talk solely between similar persons leads to echo chambers and polarization. Here we study how social networks influence the willingness to engage in political discussion with others. We do so with a focus on tie strength, similarity (opinions and identity dimensions), and extended network size and diversity. As such, we contribute a novel consequence of extended networks: whether discussing politics with others depends on how acquaintanceship networks are structured. We capture the willingness to engage in everyday political talk with a unique factorial survey experiment. We describe fictive situations and political discussion partners, and respondents then rate willingness to engage in political talk. This approach comes closer to a causal effect of extended networks compared to previous work. We introduce a multi-ethnic group perspective, scrutinizing how the impact of one’s extended network on engaging in political discussion depends on ethnic background, because, social network effects likely vary by ethnic group. We test our conjectures by analyzing novel data from 2022: the Netherlands Longitudinal Lifecourse Study (NELLS). This data sources includes an oversample of Turkish-Dutch and Moroccan-Dutch persons in the Netherlands. It includes an innovative survey experiment on political discussions, and a pre-tested Network Scale-Up (NSUM) instrument to tap into the size and composition of extended networks of acquaintances. We employ multilevel models with random intercepts and random slopes, interaction terms between our vignette attributes (e.g., background and political opinion of discussion partner), and respondent-level variables (e.g., extended network size/composition).

Intergovernmental organizations (IGOs) are believed to stimulate trade among their members, even without being explicitly designed to do so. Yet the relationship between IGOs and trade is frequently acknowledged to be bidirectional: IGOs stimulate trade, and trade stimulates joint IGO membership. This endogeneity is a fundamental characteristic of the social structure of the international system, as bilateral and multilateral relationships co-evolve in tandem with each other. Therefore, examining only one direction of the bidirectional relationship at a time may produce invalid estimates of effects. Implementing a co-evolutionary SAOM, this study demonstrates that these effects do operate in a co-evolutionary fashion, but differently than expected. Trade fosters joint IGO membership, but it is the similarity of IGO portfolios, rather than the raw strength of the IGO connection, that best predicts major trade relationships among countries.