Does Politicized Media Distort Political Discourse? Evidence from U.S. Cable News Channels

2019, Elliott Ash and Elena Labzina

While previous work has shown that partisan media affects voter choices, an open question is whether and how partisan news messaging influences the language of political discourse. This paper provides evidence on this influence in the context of the U.S. Congress and major cable news networks for the years 2005 through 2008. We measure media influence using a measure of the similarity between language in Congressional speeches and language used by speakers on Fox News, CNN, and MSNBC. Exogenous variation in news exposure across congressional districts comes from relative channel numbering, which we use as instruments. We find Fox News has had the largest effect on the language used in the Congress, with MSNBC and CNN having little effect.



Assessing climate policy preferences and temporal frames with recent computational techniques

2019, Lukas P. Fesenfeld, Elena Labzina, and Adrian Rinscheid

Despite its importance in contemporary American politics, presidential scandal is poorly understood within political science. Scholars typically interpret scandals as resulting from the disclosure of official misbehavior, but the likelihood and intensity of media scandals is also influenced by the political and news context. In this article, I provide a theoretical argument for two independent factors that should increase the president’s vulnerability to scandal: low approval among opposition party identifiers and a lack of congestion in the news agenda. Using new data and statistical approaches, I find strong support for both claims. First, I estimate duration models demonstrating that media scandals are more likely when approval is low among opposition identifiers. Using exogenous news events as an instrumental variable to overcome the endogeneity of news congestion, I then show how competing stories can crowd out scandal coverage. These results suggest that contextual factors shape the occurrence of political events and how such events are interpreted.

Rewriting knowledge: Russian Political Astroturfing as an Ideological Manifestation


Abstract: As the most implicit part of online censorship, political astroturfing imposes considerable difficulties in terms of study because of identification challenges posed by its nature of concealment. Consequently, while being a prominent factor in today’s online political discourse, it is still vastly under-investigated. In particular, one perspective completely overlooked so far is how authoritarian regimes may use it to manifest their ideology and \emph{national role conceptions}. This paper studies the behavior of the most developed and institutionalized “political trolls” in the world – the Russian ones – on the largest collaborative knowledge network, Wikipedia. The paper employs a mixed-methods approach that combines Big Data and other qualitative methodology. First, making use of Wikipedia’s unique knowledge-centered structure, which distinguishes it from other popular online social networks, this work utilizes a novel strategy to identify political astroturfing by including the available geographical information in the analysis. Second, the conclusive findings of the following qualitative investigation affirm the existence of political astroturfing on the Russian Wikipedia and that its essential function stems from current Russian regime’s self-perception. Third, the study also looks for similar contributions from “genuine” people who happen to share parallel views with the Russian regime but who are not formally associated with it. Substantively, this work concludes that “a peaceful regional integrator” is the national role conception that best corresponds to the content produced by both identified political trolls and their “genuine” accomplices .link

Generalized Non-Inferential Approach to the Multinomial Logistic Regression for the Case of The Linear Spatial Model with the Valence Component

Labzina (2018) (Master Thesis in Statistics)

Abstract: Multinomial logistic regression model (MNL) is a powerful tool frequently employed in Political Science and other Social Sciences for studying the relative probabilistic impact of the input factors on the output obtained from a discrete and bounded set. One of the MNL’s core assumptions is the identical sets of the potential outcomes for all studied subjects. Meanwhile, real socio-political settings may easily violate this condition and, consequently, the independence of the irrelevant alternatives (IIA) assumption as well. For example, this is the case in those electoral systems where parties do not have to run in all electoral regions to obtain parliamentary seats. So is it still possible to apply MNL in this situation? This paper presents an applied solution along with some coding examples. By refining the method first proposed by Yamamoto (2014), this work suggests a neat Bayesian approach to estimate the MNL even if the availability of the outcomes varies across the set of the observations. Furthermore, the article presents two applications of the method to the Spanish and British General Elections.

Fighting Terrorism in the Electronic Marketplace of Ideas: Assessing the Anonymous campaign against Islamic Extremists on Twitter with Machine Learning

Labzina, Elena and George Yin (2017)

Abstract: The Islamic State (IS) is adroit at running online campaigns to gather support for its causes. In 2015, Anonymous, an international hacker collective, initiated a campaign to expose and report jihadist Twitter accounts to disrupt the terrorist organization’s recruitment and propaganda drive online. Is a ragtag group of trolls, casual volunteers, and shady hackers match for one of the world’s deadliest terrorist organizations? To assess the effectiveness of the Anonymous campaign against IS, we built an original dataset on 16,286 unique suspicious jihadist Twitter accounts based on more than 450,000 reports from Anonymous volunteers. We then apply machine learning techniques to demonstrate a clear connection between Anonymous reporting, IS affiliation, and suspension of suspicious jihadist Twitter accounts. Furthermore, we estimate that 86% of the Anonymous reported accounts is highly likely to be affiliated with IS. This study highlights how machine learning models can help social scientists further their understanding of religious extremism online and cyber-activism. link

State-controlled Media and Foreign Policy: Analyzing Russian-language News

Labzina, Elena and Mark Neiman (2018)

Authoritarian regimes frequently employ state-owned media to frame and explain their domestic and foreign policies, often referencing national markers and invoking social identities to justify their actions. State-sponsored media accounts of ongoing events, therefore, are expected to conform to and be consistent with government-approved narratives. We take advantage of recent advances in textual analysis to test several predictions regarding news coverage of neighboring states, particularly in the build-up to military intervention. We analyze the vocabulary structure of Russia-24 broadcasts, a state-owned news channel, and Dohzd, an independent news source and identify shifts in coverage using a change-point model. Using a placebo approach to separate event-driven coverage from state-directed propaganda, we find that Russian state-owned media significantly increased its coverage of Georgia and Ukraine, in the months preceding
Russia’s military interventions. This increased coverage was often predicated with an increased discussion of traditional Russian geopolitical rivals, such as the US. link

Environmentalism as the Channel of Dissidence in China

Huang, Yixin and Elena Labzina (2017)

Abstract:  Do health and environmental issues serve as a channel to express the dissident attitudes against political institutions in Chinese society? Can we say that the Chinese government is more democratic and responsive regarding its healthcare institutions and environmental policies than in general? From the existing literature, we know that the Chinese online censorship mostly centers around possible political mobilization, not the general critique related to politics and economics. As Gary Kings shows in his work (2013), the Chinese censors are prompt to delete online messages and posts calling for action, but mostly ignore abstract complaints. Are the censors relatively tolerant to calls for mobilization when they are related to health and environment? In our work, first, we show that Chinese people are more open while discussing the environment and health related issues in particular. Second, we find that the censorship is less active regarding these issues even if they contain certain mobilization references. We achieve this by looking at 226,601,446 messages from the major Chinese social network Sino Weibo for the period from 1/2012 to 12/2012, among which 211,538 and 2,335,395 are environment and health related. These are a very significant result showing the tolerant reaction to discourse related to health and environmental institutions may be an opening to the overall democratization of the Chinese regime.