Annie Waldherr
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Annie Waldherr has a total of 59 co-authors. Here are the top co-authors by number of shared publications:
- Aytalina Kulichkina
University of Vienna
6 shared publications - Jula Lühring
University of Vienna; Complexity Science Hub, Vienna
6 shared publications - Apeksha Shetty
Department of Communication, University of Vienna, Vienna, Austria
5 shared publications - Hannah Metzler
Complexity Science Hub Vienna, Vienna, Austria
5 shared publications - Barbara Pfetsch
Weizenbaum Institute for the Networked Society , Berlin, 10623
4 shared publications - Daniela Stoltenberg
Freie Universität Berlin, Germany
4 shared publications
Latest Publications
- Open Access
- 25 September 2026
- Published by Center for Open Science in SocArXiv
Why Politicians Retweet: News Trustworthiness, Emotion, and Partisan Asymmetry in Elite News Sharing among US Congress Members
On social media, politicians' retweets can determine which claims gain visibility in online and in wider political debate, yet the factors associated with these retweeting decisions remain poorly understood. Combining natural language processing measures of emotional tone and partisan cues with politician-level characteristics, we analyze the retweeting behavior of United States Members of Congress between 2021 and 2023 using multilevel logistic regression models. Using all news-related retweets during this period, paired with a negative sample of non-retweets, we examine how tweet content and legislators' ideological extremity predict the amplification of tweets linking to more and less trustworthy news domains. We find that partisan identity is a stronger predictor of retweeting than other factors. While politicians generally preferred retweeting content with links to trustworthy sources, this preference weakened when tweets contained party-related references. Anger and fear predicted higher retweeting likelihood when tweets referenced political opponents, while anger predicted lower retweeting likelihood when tweets referenced political allies. Furthermore, ideology conditioned how politicians responded to news trustworthiness: among Republicans, increasing ideological extremity was associated with weaker preference for trustworthy sources, with the most ideologically-extreme Republicans being especially likely to retweet tweets with links to less trustworthy domains. Together, these results suggest that politicians use retweets to reinforce partisan boundaries between 'us' and 'them'. This identity-centered pattern to retweeting explains how information from less trustworthy sources gains traction in polarized digital environments.
- 25 September 2026
- Published by Center for Open Science in SocArXiv
Open AccessWhy Politicians Retweet: News Trustworthiness, Emotion, and Partisan Asymmetry in Elite News Sharing among US Congress Members
On social media, politicians' retweets can determine which claims gain visibility in online and in wider political debate, yet the factors associated with these retweeting decisions remain poorly understood. Combining natural language processing measures of emotional tone and partisan cues with politician-level characteristics, we analyze the retweeting behavior of United States Members of Congress between 2021 and 2023 using multilevel logistic regression models. Using all news-related retweets during this period, paired with a negative sample of non-retweets, we examine how tweet content and legislators' ideological extremity predict the amplification of tweets linking to more and less trustworthy news domains. We find that partisan identity is a stronger predictor of retweeting than other factors. While politicians generally preferred retweeting content with links to trustworthy sources, this preference weakened when tweets contained party-related references. Anger and fear predicted higher retweeting likelihood when tweets referenced political opponents, while anger predicted lower retweeting likelihood when tweets referenced political allies. Furthermore, ideology conditioned how politicians responded to news trustworthiness: among Republicans, increasing ideological extremity was associated with weaker preference for trustworthy sources, with the most ideologically-extreme Republicans being especially likely to retweet tweets with links to less trustworthy domains. Together, these results suggest that politicians use retweets to reinforce partisan boundaries between 'us' and 'them'. This identity-centered pattern to retweeting explains how information from less trustworthy sources gains traction in polarized digital environments.
- Open Access
- 22 August 2026
- Published by Center for Open Science in SocArXiv
Agent-based Modeling and Simulations
Computational modeling and simulations have become increasingly relevant in the social sciences, offering powerful approaches for studying complex social phenomena. This chapter focuses on agent-based modeling (ABM) and its advantages for computational communication researchers, particularly for those who are just starting out. We introduce ABM, explain its core principles, and explore different types of models and their purposes. Through illustrative examples, we show how ABM has been and can be applied to address important research questions in communication science and related disciplines. The chapter then provides a practical step-by-step guide for conducting simulation research, outlining each step of the modeling cycle according to established methodological standards. We also highlight the importance of transparency and reproducibility through open science practices. Finally, we discuss emerging trends, such as the integration of large language models, which may shape the future of computational modeling and simulations in our field. This chapter aims to equip researchers, especially beginners in computational communication science, with a solid conceptual foundation and practical tools for incorporating ABM into their work.
- 22 August 2026
- Published by Center for Open Science in SocArXiv
Open AccessAgent-based Modeling and Simulations
Computational modeling and simulations have become increasingly relevant in the social sciences, offering powerful approaches for studying complex social phenomena. This chapter focuses on agent-based modeling (ABM) and its advantages for computational communication researchers, particularly for those who are just starting out. We introduce ABM, explain its core principles, and explore different types of models and their purposes. Through illustrative examples, we show how ABM has been and can be applied to address important research questions in communication science and related disciplines. The chapter then provides a practical step-by-step guide for conducting simulation research, outlining each step of the modeling cycle according to established methodological standards. We also highlight the importance of transparency and reproducibility through open science practices. Finally, we discuss emerging trends, such as the integration of large language models, which may shape the future of computational modeling and simulations in our field. This chapter aims to equip researchers, especially beginners in computational communication science, with a solid conceptual foundation and practical tools for incorporating ABM into their work.
- Open Access
- 14 August 2026
- Published by Center for Open Science in PsyArXiv
Emotions, Group Cues, and Algorithmic Interventions: An Agent-Based Model of Elite Retweeting on Twitter
Misinformation interventions are commonly targeted at ordinary users, despite political elites having an out-sized role in misinformation amplification, and in shaping public opinion. Individual-level interventions for ordinary users may be ineffective for political elites, whose interests and motives for sharing differ substantially. In this study, we develop an empirically grounded agent-based model of intra-elite retweeting behavior to evaluate content-reranking interventions aimed at reducing the spread of tweets linking to untrustworthy news websites. Using observational data from 2021 to early 2023, we estimate a retweeting submodel that captures how emotions, group references, partisan alignment, and domain quality influence amplification decisions among U.S. Members of Congress. We embed this decision model within a directed network of elite agents and their partisan audiences to model information diffusion. We then test seven interventions using two main criteria: multiplicative discernment (the relative amplification of trustworthy vs. untrustworthy information) and relative change in total retweet volume (changes in overall retweeting volume compared to the baseline condition). We find that only directly targeting tweets with links to low-quality news websites meaningfully reduced the spread of untrustworthy information. No indirect intervention, like down-ranking angry tweets or down-ranking politicians with an extreme ideology, was effective at improving multiplicative discernment. Moreover, the intervention that improved discernment also reduced the total retweet volume, revealing an inherent structural tension between platform incentives and public interest goals. This trade-off can help explain why effective interventions may remain unimplemented under current incentive structures for social media platforms.
- 14 August 2026
- Published by Center for Open Science in PsyArXiv
Open AccessEmotions, Group Cues, and Algorithmic Interventions: An Agent-Based Model of Elite Retweeting on Twitter
Misinformation interventions are commonly targeted at ordinary users, despite political elites having an out-sized role in misinformation amplification, and in shaping public opinion. Individual-level interventions for ordinary users may be ineffective for political elites, whose interests and motives for sharing differ substantially. In this study, we develop an empirically grounded agent-based model of intra-elite retweeting behavior to evaluate content-reranking interventions aimed at reducing the spread of tweets linking to untrustworthy news websites. Using observational data from 2021 to early 2023, we estimate a retweeting submodel that captures how emotions, group references, partisan alignment, and domain quality influence amplification decisions among U.S. Members of Congress. We embed this decision model within a directed network of elite agents and their partisan audiences to model information diffusion. We then test seven interventions using two main criteria: multiplicative discernment (the relative amplification of trustworthy vs. untrustworthy information) and relative change in total retweet volume (changes in overall retweeting volume compared to the baseline condition). We find that only directly targeting tweets with links to low-quality news websites meaningfully reduced the spread of untrustworthy information. No indirect intervention, like down-ranking angry tweets or down-ranking politicians with an extreme ideology, was effective at improving multiplicative discernment. Moreover, the intervention that improved discernment also reduced the total retweet volume, revealing an inherent structural tension between platform incentives and public interest goals. This trade-off can help explain why effective interventions may remain unimplemented under current incentive structures for social media platforms.
- 28 July 2026
- Published by Taylor & Francis in Digital Journalism
- p. 1-19
Relying on the Mainstream? Entity Networks in Alternative Media
Alternative media have become increasingly relevant in the contemporary high-choice media environment, because they cater to audiences who have lost trust in legacy media, and consequently also inf...
- 28 July 2026
- Published by Taylor & Francis in Digital Journalism
- p. 1-19
Relying on the Mainstream? Entity Networks in Alternative Media
Alternative media have become increasingly relevant in the contemporary high-choice media environment, because they cater to audiences who have lost trust in legacy media, and consequently also inf...
- Open Access
- 24 July 2026
- Published by Center for Open Science in SocArXiv
Relying on the Mainstream? Entity Networks in Alternative Media
Alternative media, characterized by their heterogeneous nature and self-perceived corrective role to legacy media, have gained increasing relevance in shaping public opinion. Despite their influence, large-scale analyses of their referential and actor-related reporting characteristics remain scarce. This exploratory study investigates named entity networks in German alternative and legacy media, focusing on their reporting on the Nord Stream 2 gas pipeline from 2011 to 2022. Using computational methods, we analyze news articles from 37 alternative and 98 legacy outlets, extracting named entities and constructing co-occurrence networks to identify distinct reporting patterns. Our findings challenge assumptions that alternative media inherently diverge from legacy media in their referenced entities. While alternative media exhibit distinctive network structures—such as lower modularity and higher transitivity—they also demonstrate substantial overlap in their choice of entities and patterns of media mentions. These findings suggest that alternative media frequently mention other media, which may reflect resource limitations and reactive reporting tendencies. This study contributes to a deeper understanding of alternative media by highlighting both their unique characteristics and shared practices with legacy media.
- 24 July 2026
- Published by Center for Open Science in SocArXiv
Open AccessRelying on the Mainstream? Entity Networks in Alternative Media
Alternative media, characterized by their heterogeneous nature and self-perceived corrective role to legacy media, have gained increasing relevance in shaping public opinion. Despite their influence, large-scale analyses of their referential and actor-related reporting characteristics remain scarce. This exploratory study investigates named entity networks in German alternative and legacy media, focusing on their reporting on the Nord Stream 2 gas pipeline from 2011 to 2022. Using computational methods, we analyze news articles from 37 alternative and 98 legacy outlets, extracting named entities and constructing co-occurrence networks to identify distinct reporting patterns. Our findings challenge assumptions that alternative media inherently diverge from legacy media in their referenced entities. While alternative media exhibit distinctive network structures—such as lower modularity and higher transitivity—they also demonstrate substantial overlap in their choice of entities and patterns of media mentions. These findings suggest that alternative media frequently mention other media, which may reflect resource limitations and reactive reporting tendencies. This study contributes to a deeper understanding of alternative media by highlighting both their unique characteristics and shared practices with legacy media.
- Open Access
- 18 June 2026
- Published by Center for Open Science in SocArXiv
Agent-based models of social media: A scoping review and research agenda
Social media research has advanced our understanding of individual-level processes in online communication, yet these insights are rarely linked to meso- and macro-level outcomes relevant to the field, including polarization dynamics, misinformation spread, and public opinion formation. Agent-based modeling (ABM), as a method of constructing theoretical representations of systems, offers a powerful approach to bridging this gap by linking micro-level interactions to meso-level processes and emergent macro-level phenomena. While ABM has been widely adopted in computer science, engineering, mathematics, and economics to model social media communication, such applications often rely on theoretical assumptions and empirical findings outside of social media research. To address this disconnect and encourage broader adoption of ABM in the field, we systematically review how the method has been applied across various disciplines to study social media phenomena and offer a comprehensive foundation for building on existing models. We identify common practices, research gaps, and opportunities for theoretical, methodological, and open science advancements. The resulting agenda provides a practical entry point for re-searchers new to modeling and recommends grounding relevant models in established theories and insights from social media research.
- 18 June 2026
- Published by Center for Open Science in SocArXiv
Open AccessAgent-based models of social media: A scoping review and research agenda
Social media research has advanced our understanding of individual-level processes in online communication, yet these insights are rarely linked to meso- and macro-level outcomes relevant to the field, including polarization dynamics, misinformation spread, and public opinion formation. Agent-based modeling (ABM), as a method of constructing theoretical representations of systems, offers a powerful approach to bridging this gap by linking micro-level interactions to meso-level processes and emergent macro-level phenomena. While ABM has been widely adopted in computer science, engineering, mathematics, and economics to model social media communication, such applications often rely on theoretical assumptions and empirical findings outside of social media research. To address this disconnect and encourage broader adoption of ABM in the field, we systematically review how the method has been applied across various disciplines to study social media phenomena and offer a comprehensive foundation for building on existing models. We identify common practices, research gaps, and opportunities for theoretical, methodological, and open science advancements. The resulting agenda provides a practical entry point for re-searchers new to modeling and recommends grounding relevant models in established theories and insights from social media research.
- Open Access
- 17 June 2026
- Published by Center for Open Science in SocArXiv
Using agent-based simulations to anticipate risks and mitigation measures in platform governance
Risk analysis and mitigation in platform governance are constrained by the difficulties of testing changes to platforms’ systems and accessing data. Other fields, ranging from public health to economics, as well as an emerging body of platform research, use agent-based modeling to overcome this obstacle by empowering researchers and regulators to test changes virtually. However, it is still unclear how and under what conditions agent-based modeling can effectively support platform governance. This article aims to reduce this uncertainty in two ways. First, it provides a theoretical framework on the benefits and risks of using agent-based modeling in platform governance, drawing on literature on platform risk assessment, the use of agent-based modeling in other fields, and anticipatory governance and evidence-based policymaking. Second, to validate this framework and clarify the conditions under which regulators can rely on evidence produced by agent-based models, the article reports the results of interviews with 18 regulators. The article finds that agent-based simulations have significant potential to complement other sources of evidence used in platform risk assessment. Their main added value is that they allow researchers and regulators to more easily, ethically, and independently explore the potential impacts of changes to platform design, and provide interactive and visual scenarios that concretize abstract platform dynamics.
- 17 June 2026
- Published by Center for Open Science in SocArXiv
Open AccessUsing agent-based simulations to anticipate risks and mitigation measures in platform governance
Risk analysis and mitigation in platform governance are constrained by the difficulties of testing changes to platforms’ systems and accessing data. Other fields, ranging from public health to economics, as well as an emerging body of platform research, use agent-based modeling to overcome this obstacle by empowering researchers and regulators to test changes virtually. However, it is still unclear how and under what conditions agent-based modeling can effectively support platform governance. This article aims to reduce this uncertainty in two ways. First, it provides a theoretical framework on the benefits and risks of using agent-based modeling in platform governance, drawing on literature on platform risk assessment, the use of agent-based modeling in other fields, and anticipatory governance and evidence-based policymaking. Second, to validate this framework and clarify the conditions under which regulators can rely on evidence produced by agent-based models, the article reports the results of interviews with 18 regulators. The article finds that agent-based simulations have significant potential to complement other sources of evidence used in platform risk assessment. Their main added value is that they allow researchers and regulators to more easily, ethically, and independently explore the potential impacts of changes to platform design, and provide interactive and visual scenarios that concretize abstract platform dynamics.
- 16 June 2026
- Published by Edward Elgar Publishing
- p. 148-162
Issue networks
Issue networks are constellations of loosely connected individual or collective actors who discuss a particular issue. The empirical study of issue networks has gained traction with the development of public spheres on the internet and on social media. Examples are civil society coalitions and social movements around social issues such as environmental policies, civil rights, or public health. Data collection of issue networks is usually either driven by an actor-centered or an issue-centered approach. In this chapter, we first discuss early actor-centered approaches operationalizing issue networks as snowball-sampled hyperlink networks and show how these have been adapted recently to study issue networks on social media such as Telegram and YouTube. Second, we describe the collection of hashtag networks from Twitter as the generic example for an issue-centered approach and illustrate how this principle can be applied to other platforms such as TikTok and to studying cross-platform issue networks.
- 16 June 2026
- Published by Edward Elgar Publishing
- p. 148-162
Issue networks
Issue networks are constellations of loosely connected individual or collective actors who discuss a particular issue. The empirical study of issue networks has gained traction with the development of public spheres on the internet and on social media. Examples are civil society coalitions and social movements around social issues such as environmental policies, civil rights, or public health. Data collection of issue networks is usually either driven by an actor-centered or an issue-centered approach. In this chapter, we first discuss early actor-centered approaches operationalizing issue networks as snowball-sampled hyperlink networks and show how these have been adapted recently to study issue networks on social media such as Telegram and YouTube. Second, we describe the collection of hashtag networks from Twitter as the generic example for an issue-centered approach and illustrate how this principle can be applied to other platforms such as TikTok and to studying cross-platform issue networks.
- Open Access
- 8 January 2026
- Published by Center for Open Science in SocArXiv
A Longitudinal Approach to the Analysis of Social Media Engagement: The Case of Anger-Driven Climate-Skeptic Message Propagation During the 2021 German Elections
This study examines the advantages of longitudinal modeling over cross-sectional approaches for analyzing social media engagement, specifically exploring how emotional reactions—particularly anger—influence content sharing. While cross-sectional methods are commonly used, engagement patterns are inherently temporal, making longitudinal approaches more appropriate. These methods offer a middle ground between simple cross-sectional analyses and complex time-series techniques, reducing omitted variable bias and accounting for time-varying factors like algorithmic amplification while accommodating irregular social media data. We empirically compare these approaches using 1,137 environmentally themed Facebook posts from German political parties during the 2021 federal election, focusing on Alternative für Deutschland (AfD) and Die Grünen (The Greens), parties with opposing environmental policies. Cross-sectional estimates using three sampling strategies were compared with longitudinal results via Bayesian multilevel regression with negative binomial specification.Findings reveal that longitudinal modeling produces more conservative and precise estimates, while cross-sectional methods tend to inflate effect sizes and interparty differences. Although all models show anger positively associates with sharing for AfD and negatively for Die Grünen, longitudinal analysis provides greater inferential stability by controlling for time-invariant confounders. Such differences have important theoretical implications, underscoring the value of incorporating temporal dynamics into social media research while acknowledging data access challenges.
- 8 January 2026
- Published by Center for Open Science in SocArXiv
Open AccessA Longitudinal Approach to the Analysis of Social Media Engagement: The Case of Anger-Driven Climate-Skeptic Message Propagation During the 2021 German Elections
This study examines the advantages of longitudinal modeling over cross-sectional approaches for analyzing social media engagement, specifically exploring how emotional reactions—particularly anger—influence content sharing. While cross-sectional methods are commonly used, engagement patterns are inherently temporal, making longitudinal approaches more appropriate. These methods offer a middle ground between simple cross-sectional analyses and complex time-series techniques, reducing omitted variable bias and accounting for time-varying factors like algorithmic amplification while accommodating irregular social media data. We empirically compare these approaches using 1,137 environmentally themed Facebook posts from German political parties during the 2021 federal election, focusing on Alternative für Deutschland (AfD) and Die Grünen (The Greens), parties with opposing environmental policies. Cross-sectional estimates using three sampling strategies were compared with longitudinal results via Bayesian multilevel regression with negative binomial specification.Findings reveal that longitudinal modeling produces more conservative and precise estimates, while cross-sectional methods tend to inflate effect sizes and interparty differences. Although all models show anger positively associates with sharing for AfD and negatively for Die Grünen, longitudinal analysis provides greater inferential stability by controlling for time-invariant confounders. Such differences have important theoretical implications, underscoring the value of incorporating temporal dynamics into social media research while acknowledging data access challenges.
- Open Access
- 31 October 2025
- Published by Center for Open Science in PsyArXiv
Quantifying the Causal Effects of Misinformation on Engagement and Emotions on Social Media
Misinformation drives engagement on social media platforms, yet the underlying emotional mechanisms remain poorly understood. Previous studies rely on biased samples and fail to distinguish between discrete emotions, while correlational approaches cannot disentangle the effects of misinformation from confounding factors such as political ideology. Whether misinformation causally affects online interactions therefore remains an open question. We analyze 9.9M German news tweets and 11M replies on Twitter (Oct 2020–Mar 2022) across news sources of varying trustworthiness. Using machine learning and causal inference methods, we examine whether news trustworthiness influences engagement and emotions in replies.Despite comprising only 6% of all news shared, untrustworthy sources receive 39% more retweets, 12% more quote retweets, but 14% fewer likes and 19% replies. Replies to untrustworthy content show 9% more anger, 22% more disgust, and 12% less joy, with anger potentially driven by self-selection. These findings show misinformation leaves distinct fingerprints online, with direct implications for platform design.
- 31 October 2025
- Published by Center for Open Science in PsyArXiv
Open AccessQuantifying the Causal Effects of Misinformation on Engagement and Emotions on Social Media
Misinformation drives engagement on social media platforms, yet the underlying emotional mechanisms remain poorly understood. Previous studies rely on biased samples and fail to distinguish between discrete emotions, while correlational approaches cannot disentangle the effects of misinformation from confounding factors such as political ideology. Whether misinformation causally affects online interactions therefore remains an open question. We analyze 9.9M German news tweets and 11M replies on Twitter (Oct 2020–Mar 2022) across news sources of varying trustworthiness. Using machine learning and causal inference methods, we examine whether news trustworthiness influences engagement and emotions in replies.Despite comprising only 6% of all news shared, untrustworthy sources receive 39% more retweets, 12% more quote retweets, but 14% fewer likes and 19% replies. Replies to untrustworthy content show 9% more anger, 22% more disgust, and 12% less joy, with anger potentially driven by self-selection. These findings show misinformation leaves distinct fingerprints online, with direct implications for platform design.