Abstract Network agenda-setting theory proposes that media transfer issue salience and structural attribute relationships to public and political agendas. However, this theory has been predominantly tested in stable democracies, leaving unclear how network-level agenda transfer functions during authoritarian transitions. This study employs a Triadic Network Agenda-Setting framework to examine how Hong Kong's transition from semi-democratic to authoritarian governance shifted agenda dynamics about asylum seekers. Analyzing asylum coverage across Hong Kong newspapers, online forums, and legislators’ Facebook pages between 2016 and 2024, the study employed structural topic modeling, multiple regression quadratic assignment procedure, and Granger causality tests to assess network-level structural alignment and attribute-level temporal dynamics. Within a corpus dominated by one pro-establishment newspaper, legislative-to-media structural alignment emerged in the authoritarian period but concentrated within that outlet rather than diffusing across the press system, while legislative temporal influence on attribute salience disappeared. This press-state alignment ran directly between legislators and the dominant outlet rather than transitively through public discourse, and engaged online publics remained structurally independent of media and legislative networks across politically heterogeneous communities. Given the outlet-selective character of press–state alignment, humanitarian counter-narratives may gain greater traction through outlets retaining structural autonomy than through direct contestation of the dominant outlet.
This paper investigates the association between Americans' lesbian, gay, bisexual, transgender (LGBT) social networks and their views on LGBT issues in education. We use nationally representative, probability-based survey data to examine Americans' exposure to LGBT individuals in their networks. Using item response theory and multiple regression, we explore the relationship between types of LGBT social networks and their support for LGBT-related scenarios in schools. Those who have LGBT individuals in their social networks express greater support for LGBT scenarios in schools, though the associations are stronger for individuals with LGBT friends than those with LGBT family members or other social relations. We also find that having a transgender individual in one's social network is more strongly associated with support for LGBT-related scenarios in schools than having an LGB-only network (this is true for both LGB and transgender-related scenarios). While correlational, our results point to the potential value of LGBT exposure in social networks for shaping views about LGBT issues in schools. The extremely low levels of exposure, particularly to transgender people, may be contributing to low levels of support for LGBT issues in schools.
This research note tests whether direct questions (DQs) overestimate negative attitudes toward the news media in Japan relative to an indirect list experiment (LE). Two survey-embedded randomized experiments assigned respondents to either a DQ condition or a double-LE with two control sets. The focal statement was that "the mass media are harmful to society." Across both studies, agreement was higher in the DQ condition than in the pooled LE estimate: 45.1% versus 29.7%, a 15.4-point gap. The double-list design showed no clear design effects, and the directional pattern was consistent across studies. These findings suggest that DQs may overstate negative media attitudes and highlight the directional nature of social desirability bias in surveys.
This study examines the sociopolitical determinants of North Korean defectors' attitudes toward nuclear weapons, reflecting their views while living in North Korea. Using original survey data from 1,241 defectors, the analysis tests how perceptions of external threat, domestic ideology, and social experience shape nuclear attitudes. Ordered logistic regression results show that perceiving South Korea as a threat, pride in Juche ideology, and trust in state media increase support for nuclear weapons, while exposure to South Korean media shows no consistent effect. These findings suggest that nuclear attitudes in authoritarian regimes are formed through the interplay of external threat perception, ideological indoctrination, and lived experience, offering insight into the microfoundations of proliferation politics.
Attitudes toward refugees (ATR) have become a pivotal concern in Germany following the substantial refugee influx of 2015. This study examines levels and changes in ATR in the German population between 2017 and 2021 across different sociodemographic groups. In contrast to previous studies investigating similar research questions, our approach captured within-individual changes across three dimensions of ATR-cognitive, affective, and behavioral-a perspective largely missing in existing research. Using the GESIS Panel and latent growth curve modeling, we found that after an initial increase in negative sentiments, attitudes gradually became more positive over time, with the strongest positive shift observed in the cognitive dimension, followed by the affective and behavioral dimensions. Women, West Germans, higher-educated, and politically left-leaning individuals held more positive ATR across all ATR dimensions. These sociodemographic differences remained largely stable without attenuating or intensifying over the 4 years. The uniform change patterns in ATR challenge assumptions that certain groups are more resistant to attitude change or that attitudes have polarized in Germany's post-2015 environment.
The global proliferation of "fake news" on social media poses significant challenges, yet research on its perceived effects in the Global South remains limited. This study addresses this gap by examining the third-person effects of "fake news" in Kenya, a context characterized by high social media usage and the prevalence of "fake news," especially during the politically charged 2022 general election. Drawing on an online survey (N = 883), we find evidence of a self-other discrepancy, with respondents perceiving greater "fake news" effects on others than on themselves. Furthermore, third-person effects positively correlate with support for censorship, a relationship that is moderated by the perceived spread of "fake news" in Kenya. These findings extend the third-person effect hypothesis to underexplored contexts, offering insights into how "fake news" shapes attitudes toward media regulation in hybrid democracies. Implications for theory and practice in the African sociopolitical landscape are discussed.
Social media have substantially changed the opportunities for political engagement, both in how citizens become informed and learn about politics, as well as in their political participation. On the one hand, citizens can encounter a large diversity of political content on these platforms, both in terms of format and topic. On the other hand, social media are increasingly used by citizens to express their views and mobilize others in highly personalized ways. While social media seem to hold unique opportunities for political learning and participation, it remains unclear how these processes might be linked to each other. Therefore, following communication mediation theory, we explore whether and how political knowledge about current affairs might mediate the association between exposure to political content on social media and participation on these same platforms. Data from a cross-sectional study in Flanders, Belgium (n = 2,520), reveal a complex participatory cycle. While citizens do seem to learn from social media political content, those who are more knowledgeable engage less in social media political participation. While unexpected, these findings highlight the importance of considering the unique social media affordances when researching political socialization and participation.
This study investigates how the number and labeling of response categories in survey scales affect respondent behavior, psychometric properties, and substantive conclusions. Using data from a web-based survey experiment with over 2,800 participants, we randomly assigned respondents to scales that varied in the number of response options and labeling formats. Engagement, response style tendencies, reliability and convergent validity were assessed through a combination of process data (e.g., response times, cursor movements), self-reports, and psychometric modeling. Our findings suggest that scales with a greater number of response categories tend to exhibit higher reliability, although the improvements are modest and depend on the specific scale. While scales with more response options required longer completion times and prompted more complex response behavior, the number of response categories did not influence substantive conclusions in simple regression models.
We experimentally evaluate the effect of media literacy tips intervention on discernment of COVID-19 and Russo-Ukrainian war fake news, utilizing a sample representative of general adult population from a highly polarized Slovak republic. In addition, we test the relative power of the intervention against cognitive virtues and already-held beliefs, while controlling for trust in established and alternative news sources. We find that media literacy tips significantly improve discernment by lowering the perceived trustworthiness of fake news. Individuals with higher analytic thinking, scientific reasoning, intellectual humility, and epistemic curiosity, and those who trust established media discern fake from real news better, while those who trust alternative media and those who possess more COVID-19 and pro-Kremlin disinformation beliefs, discern worse. Importantly, the positive effect of media literacy tips intervention remains significant even after controlling for cognitive virtues, already-held beliefs, and trust in news sources.
Despite extensive research on news consumption and political learning, the implications of emerging technologies, particularly artificial intelligence (AI)-driven smart media devices such as Amazon Echo and Google Home, remain underexplored. Moreover, little is known about how cognitive factors, such as trust in AI technologies, shape these dynamics. This study investigates the relationship between smart media use and political knowledge, emphasizing the mediating role of political discussion and the moderating influence of trust in AI. Drawing on survey data from three studies conducted across two countries, the findings reveal that news use via smart media devices is linked to both positive and negative outcomes for political knowledge, depending on the nature of individuals' political discussions. Specifically, smart media news use enhances political knowledge when discussions occur with strong social ties on social media, but it is negatively associated with knowledge when discussions involve weak ties. The magnitude of both positive and negative effects increases among those who exhibit greater trust in AI. By integrating smart media use and AI trust into existing models of political learning, this study offers novel insights into the evolving role of intelligent technologies in shaping citizens' political understanding in the digital age.
Although misinformation has been regarded as an urgent global risk, empirical evidence on its prevalence is scarce. Yet, news users across different countries believe that about half of all information they see is misinformation. Such heightened risk perceptions can have severe ramifications, as people who perceive that misinformation prevails may systematically distrust or avoid all established information. Yet, we currently lack evidence on how misinformation is perceived by news users, and how they distinguish it from accurate information. Therefore, we rely on the analysis of experienced misinformation versus credible information regarding the issue of climate change submitted by participants via hyperlinks in surveys across the US and the Netherlands (N = 1,396). Our main findings indicate that perceived misinformation encompasses a wide range of information disorders and sources whilst being infused with partisan biases. Perceptions of credible information are more consensual and echo ideal types of trustworthy information from credible, authoritative sources. Our findings illustrate how news users perceive the boundary between real and false information, which may offer a useful starting point for interventions aiming to raise awareness about the distinction between misinformation and reliable news.
Russia's invasion of Ukraine in February 2022 marked a turning point in European security politics. In this context, the population's willingness to fight can be seen as an indicator of societal inclination toward military self-defense and a precondition for strengthening defensive capacities. Yet, causal evidence on what shapes this willingness remains limited, and studies on the effects of Russia's invasion on public attitudes have been largely confined to European countries. I employ Russia's invasion within an "Unexpected Event During Survey" design to estimate its impact on respondents' willingness to fight for their own country. Drawing on data from the World Values Survey in Czechia and Uruguay, the analysis reveals a significant increase in willingness to fight among Czech respondents-a country geographically and historically proximate to Russia-but no comparable effect in Uruguay. These findings offer novel causal evidence that proximate interstate conflict can increase willingness to fight for one's country, which contributes to broader debates on conscription and military security.
During election campaigns, the media become crucial sources of information, driven by citizens' heightened interest in accessing political content. Although the relationship between political interest and media consumption has been extensively studied, important questions remain regarding the nature and direction of this causal link. This paper analyzes the relationship between these two variables during the 2018 and 2021 Mexican federal election campaigns, with the aim of identifying whether media effects, selection effects, or reciprocal effects were present. To this end, data from two panel surveys conducted among the adult population in Mexico during each electoral cycle were analyzed. These surveys measured political interest and media consumption at multiple points throughout the campaigns, allowing for an assessment of their causal dynamics. While some unidirectional effects were observed, the findings indicate that reciprocal relationships predominated across both electoral periods.
This study examines how explicitly political preferences, such as party support and issue positions, influence social interactions relative to lifestyle and cultural preferences that may be more immediate in people's daily lives. Using conjoint analysis, we assess the relative impact of factors such as political party preferences, issue stances, musical taste, restaurant choices, and football team fandom on social preferences in the Czech Republic. While political preferences (particularly agreement or disagreement on party support) are key drivers of social interactions among politically engaged individuals, cultural and lifestyle preferences also exert notable influence. Our findings suggest that although political factors often dominate, their salience varies depending on levels of political interest. By testing political preferences alongside everyday lifestyle and cultural factors, this study provides a more nuanced understanding of the diverse influences that shape social interactions in contemporary society.
The goal of our study was twofold. First, we aimed to recognize how concepts, factors, and artificial intelligence-driven challenges in research on public opinion vary across experts with different experiences in the field. Second, we examined how experts envision the future of the field in the evolving media landscape. We invited public opinion scholars from more than 30 countries to participate in the survey. Findings showed that public opinion scholars, regardless of their experience level and area of study, perceived opinion formation and respondent bias as 2 significant challenges to their studies. The thematic analysis highlighted a dual challenge for scholars and experts: addressing technical methodological hurdles while rethinking theoretical frameworks for public opinion research in the digital era.
The spread of online political misinformation on social media and messaging platforms during Malaysia's 15th General Elections raised significant concerns about its nature and causes. Dubbed the "Green Wave" phenomenon, the campaign period saw an increase in right-wing nationalist and religious content, intensifying fears of declining resilience to misinformation. While prior research has examined conservatism and religiosity separately in relation to misinformation, few studies have explored their combined effects across different platforms. Using a two-wave panel survey (Wave 1 N = 1,800; Wave 2 N = 901), this study examines how these factors shaped misinformation belief and sharing longitudinally during an election campaign. Consistent with prior research, conservatives were more likely to believe and share misinformation on Facebook and WhatsApp. Religiosity, however, showed no direct relationship with misinformation susceptibility, either in belief or sharing. Notably, religiosity moderated conservatives' behavior: conservatives with stronger religious outlooks were less likely to share misinformation on both Facebook and WhatsApp, though no similar moderating effect emerged for misinformation belief. These findings reveal a complex interplay between conservatism, religiosity, and misinformation, emphasizing the nuanced role of religiosity in being associated with weaker misinformation-sharing tendencies among conservatives. This study contributes to understanding how political and religious factors influence vulnerability to misinformation in diverse digital ecosystems.
At a time when the influence of external cues on opinion formation is gaining attention due to their relevance for preference formation, this research note investigates how external cues from political elites and close friends shape individual policy preferences. Building on prior studies by Barber and Pope (2019, 2024) regarding Donald Trump's impact on people's policy preferences during his 2017-2021 presidency, we examine whether this influence persists soon after Trump's first presidential tenure concludes in late 2021 through a survey experiment. We also assess the influence of a second external cue that has gained attention due to the growing role of social media in political processes: a close friend, who might also serve as a relevant figure for policy position formation. We find that Trump's liberal cues shift Republican preferences in a liberal direction, while close friend cues show diverse patterns, including some backlash effects where individuals adopt positions opposite to conservative friend advocacy. These effects vary with individuals' party identification, political knowledge, and social conformity levels. Individual characteristics moderate political elite influence, while peer influence operates through reactance mechanisms rather than conformity, revealing that social networks shape opinions through more complex pathways than commonly theorized.
Voters in the United States consistently express to pollsters that they would like a third-party option, yet third-party candidates rarely earn their votes. In this study, we demonstrate why expressed support for a third-party candidate does not match actual support because the inclination to vote for a third-party candidate does not align with a consistent political ideology. We utilize a novel experiment (N = 777) to demonstrate that, though there is a sizable number of people generally supportive of a third-party candidate, once a candidate is placed on the ideological spectrum, many of these voters are unlikely to support them. We then present a study of the 2024 U.S. presidential election (N = 556) in which we find robust desire to support a third-party candidate in general but a complete absence of specific candidate support. Overall, we find that third-party candidates face considerable ideological barriers to actual support at the ballot box despite voters' professed desire for a third-party candidate.
This article investigates how regional-level immigration of different origins and educational backgrounds predicts individual policy preferences across Western European countries between 2010 and 2020. Combining data from the European Social Survey and the European Union Labour Force Survey, we examine how changes in net migration affect support for restrictive immigration policies. We find that short-term increases in non-European immigration within a given region are associated with more restrictive migration policy views among natives, regardless of immigrants' education levels. Attitudes toward immigration are also relatively more negatively affected among natives with financial difficulties when immigrants are not tertiary educated, but not when those immigrants have a tertiary education. Our findings suggest that while origin-based heuristics dominate public responses to immigration, natives are not entirely blind to immigrants' educational attainment.
Anti-immigrant attitudes are one of the most important explanations for radical right voting. However, these attitudes are remarkably stable and, therefore, not well-suited to fully explain fluctuations in radical right parties' success. In this study, we tested whether perceiving immigration as an important problem is a condition under which anti-immigrant attitudes are translated into a radical right vote. The salience of immigration is more prone to fluctuations and may therefore contribute to our understanding of the dynamics of radical right voting. The Dutch Parliamentary Election Studies from 2002 to 2021 were employed to test our hypotheses. We found that the salience of immigration had an effect on radical right voting above and beyond anti-immigrant attitudes. Salience had a somewhat stronger effect among individuals with stronger anti-immigrant attitudes. We conclude that salience is not a necessary condition for individuals to translate their anti-immigrant attitudes into a radical right vote, but that salience does increase the probability of voting for the radical right, especially for those who already hold negative attitudes toward immigration. Future research should improve the measurement of what issues voters prioritize, and seek to understand which issues compel individuals to vote for the radical right.
