Search results

AbstractVoting Advice Applications (VAAs), which provide citizens with information on the party that best represents their political preferences, are often cited as evidence of the empowering capabilities of digital tools. Aside from the informational benefits of these voter guides, observational studies have suggested a strong effect on political participation and vote choice. However, existing impact evaluations have been limited by a reliance on convenience samples, lack of random assignment, or both. This raises questions about self-selection and the precise mechanisms underlying how voters learn about politics. Here, we provide evidence from a field experiment with survey outcomes conducted with a sample of over 1,000 German citizens in the 2017 federal election campaign. Using linked panel survey and digital trace data combined with a randomized encouragement to complete a VAA, we assess respondents' compliance with treatment and observe how the use of this tool affects political behavior, attitudes, media consumption, political knowledge, and even social media activity. Our findings reveal that the overwhelming consensus in favor of positive effects on turnout and vote choice should be treated with caution, as we find no such effects. Rather, the actual virtue of VAAs in a complex online information environment lies in increasing knowledge about parties' positions on issues—exactly the kind of information these tools were designed to provide.

Online political astroturfing—hidden information campaigns in which a political actor mimics genuine citizen behavior by incentivizing agents to spread information online—has become prevalent on social media. Such inauthentic information campaigns threaten to undermine the Internet's promise to more equitable participation in public debates. We argue that the logic of social behavior within the campaign bureaucracy and principal–agent problems lead to detectable activity patterns among the campaign's social media accounts. Our analysis uses a network-based methodology to identify such coordination patterns in all campaigns contained in the largest publicly available database on astroturfing published by Twitter. On average, 74% of the involved accounts in each campaign engaged in a simple form of coordination that we call co-tweeting and co-retweeting. Comparing the astroturfing accounts to various systematically constructed comparison samples, we show that the same behavior is negligible among the accounts of regular users that the campaigns try to mimic. As its main substantive contribution, the paper demonstrates that online political astroturfing consistently leaves similar traces of coordination, even across diverse political and country contexts and different time periods. The presented methodology is a reliable first step for detecting astroturfing campaigns.

Online political astroturfing-hidden information campaigns in which a political actor mimics genuine citizen behavior by incentivizing agents to spread information online-has become prevalent on social media. Such inauthentic information campaigns threaten to undermine the Internet's promise to more equitable participation in public debates. We argue that the logic of social behavior within the campaign bureaucracy and principal-agent problems lead to detectable activity patterns among the campaign's social media accounts. Our analysis uses a network-based methodology to identify such coordination patterns in all campaigns contained in the largest publicly available database on astroturfing published by Twitter. On average, 74% of the involved accounts in each campaign engaged in a simple form of coordination that we call co-tweeting and co-retweeting. Comparing the astroturfing accounts to various systematically constructed comparison samples, we show that the same behavior is negligible among the accounts of regular users that the campaigns try to mimic. As its main substantive contribution, the paper demonstrates that online political astroturfing consistently leaves similar traces of coordination, even across diverse political and country contexts and different time periods. The presented methodology is a reliable first step for detecting astroturfing campaigns.

Online political astroturfing—hidden information campaigns in which a political actor mimics genuine citizen behavior by incentivizing agents to spread information online—has become prevalent on social media. Such inauthentic information campaigns threaten to undermine the Internet's promise to more equitable participation in public debates. We argue that the logic of social behavior within the campaign bureaucracy and principal–agent problems lead to detectable activity patterns among the campaign's social media accounts. Our analysis uses a network-based methodology to identify such coordination patterns in all campaigns contained in the largest publicly available database on astroturfing published by Twitter. On average, 74% of the involved accounts in each campaign engaged in a simple form of coordination that we call co-tweeting and co-retweeting. Comparing the astroturfing accounts to various systematically constructed comparison samples, we show that the same behavior is negligible among the accounts of regular users that the campaigns try to mimic. As its main substantive contribution, the paper demonstrates that online political astroturfing consistently leaves similar traces of coordination, even across diverse political and country contexts and different time periods. The presented methodology is a reliable first step for detecting astroturfing campaigns.

Twitter provides a direct method for political actors to connect with citizens, and for those citizens to organize into online clusters through their use of hashtags (i.e., a word or phrase marked with # to identify an idea or topic and facilitate a search for it). We examine the political alignments and networking of Twitter users, analyzing 9 million tweets produced by more than 23,000 randomly selected followers of candidates for the U.S. House and Senate and governorships in 2010. We find that Twitter users in that election cycle did not align in a simple Right-Left division; rather, five unique clusters emerged within Twitter networks, three of them representing different conservative groupings. Going beyond discourses of fragmentation and polarization, certain clusters engaged in strategic expression such as "retweeting" (i.e., sharing someone else's tweet with one's followers) and "hashjacking" (i.e., co-opting the hashtags preferred by political adversaries). We find the Twitter alignments in the political Right were more nuanced than those on the political Left and discuss implications of this behavior in relation to the rise of the Tea Party during the 2010 elections.

Polls posted on social media can provide information about public opinion on a variety of issues from business decisions to support for presidential election candidates. However, it is largely unknown whether the information provided by social polls is useful or not. To enhance our understanding of social polls, we examine nearly two thousand Twitter polls gauging support for U.S. presidential candidates during the 2016 and 2020 election campaigns.
First, we describe the prevalence of social polls. Second, we characterize social polls in terms of the engagement they elicit and the response options they present. Third, leveraging machine learning models, we infer and describe several characteristics, including demographics and political leanings, of the users who author and interact with social polls. Finally, we study the relationship between social poll results, their attributes, and the characteristics of users interacting with them. Our findings suggest how and to what extent polling on Twitter is biased in terms of content, authorship, and audience. The 2016 and 2020 polls were predominantly crafted by older males and manifested a pronounced bias favoring candidate Donald Trump, whereas traditional surveys favored Democratic candidates. We further identify and explore the potential reasons for such biases and discuss their repercussions.

AbstractIn this note, we provide direct evidence of cheating in online assessments of political knowledge. We combine survey responses with web tracking data of a German and a US online panel to assess whether people turn to external sources for answers. We observe item-level prevalence rates of cheating that range from 0 to 12 percent depending on question type and difficulty, and find that 23 percent of respondents engage in cheating at least once across waves. In the US panel, which employed a commitment pledge, we observe cheating behavior among less than 1 percent of respondents. We find robust respondent- and item-level characteristics associated with cheating. However, item-level instances of cheating are rare events; as such, they are difficult to predict and correct for without tracking data. Even so, our analyses comparing naive and cheating-corrected measures of political knowledge provide evidence that cheating does not substantially distort inferences.

How populists engage with media of various types, and are treated by those media, are questions of international interest. In the United States, Donald Trump stands out for both his populism-inflected campaign style and his success at attracting media attention. This article examines how interactions between candidate communications, social media, partisan media, and news media combined to shape attention to Trump, Clinton, Cruz, and Sanders during the 2015–2016 American presidential primary elections. We identify six major components of the American media system and measure candidates' efforts to gain attention from them. Our results demonstrate that social media activity, in the form of retweets of candidate posts, provided a significant boost to news media coverage of Trump, but no comparable boost for other candidates. Furthermore, Trump tweeted more at times when he had recently garnered less of a relative advantage in news attention, suggesting he strategically used Twitter to trigger coverage.

Das von der VolkswagenStiftung geförderte Forschungsprojekt untersucht die Konsequenzen der Online-Medienpräsenz für politische Präferenzen und Verhaltensweisen. Die Studie wurde von YouGov USA durchgeführt. Im Erhebungszeitraum 23. April 2018 bis 15. Oktober 2019 wurden amerikanische Staatsbürger ab 18 Jahren mit Internetzugang in Onlineinterviews (CAWI) zu folgenden Themen befragt: Politische Präferenzen und politisches Verhalten, Nutzung sozialer Medien, Mediennutzung, Einstellungen zu bestimmten Themen, politisches Wissen, Meinungen zur Regulierung von Online-Belästigung. Die Auswahl der Befragten erfolgte durch eine Quotenstichprobe aus einem Online-Access-Panel.

Die Studie Media Exposure and Opinion Formation (MEOF) ist eine Mehrländer- und -wellen-Panelbefragung, die zwischen Juli 2017 und Oktober 2019 in Deutschland und April 2018 und Oktober 2019 in den USA durchgeführt wurde. Die Befragung ermöglicht es, politische Einstellungen und Verhaltensweisen, Wissen, Online-Medienkonsum und Einstellungen zu verschiedenen Themen zu untersuchen. Zusätzlich zu den Umfragedaten wurde eine passive Messtechnologie (Tracking-Software) auf den Desktop- und Mobilgeräten der Befragten eingesetzt, um mit Einverständnis der Befragten Echtzeitdaten über Webbesuche und die Nutzung mobiler Apps zu sammeln.

Das von der VolkswagenStiftung geförderte Forschungsprojekt untersucht die Konsequenzen der Online-Medienpräsenz für politische Präferenzen und Verhaltensweisen. Die Studie wurde von YouGov Deutschland durchgeführt. Im Erhebungszeitraum 13. Juli 2017 bis 14. Oktober 2019 wurden deutsche Staatsbürger ab 18 Jahren mit Internetzugang in Onlineinterviews (CAWI) zu folgenden Themen befragt: Politische Präferenzen und politisches Verhalten, Nutzung sozialer Medien, Mediennutzung, Einstellungen zu bestimmten Themen, politisches Wissen, Meinungen zur Regulierung von Online-Belästigung. Die Auswahl der Befragten erfolgte durch eine Quotenstichprobe aus einem Online-Access-Panel.