Politicizing the Medium and the Message: Media’s Role in Partisan Divisions and Mass Polarization

Abstract

To gain novel insights into ideological divisions within the media institution and assess their role in mass partisan polarization, this dissertation develops a comprehensive framework to examine U.S. media coverage of polarizing issues and its effects on audience engagement and news dissemination at the national scale. This framework integrates a global, longitudinal analysis of coverage of three highly polarizing issues by U.S. media outlets—COVID-19, the war in Ukraine, and the latest war in Gaza—with an investigation into the most engaging and widely disseminated news content. In the case of COVID-19, the framework also models the relationship between news dissemination and public health outcomes across American communities.Using advanced computational and statistical methods, this work presents arguably the largest, most systematic and comprehensive analysis of media coverage of polarizing issues to date. More than 10 million news reports from 131 U.S. media outlets, spanning up to 4.8 years and disseminated across three major platforms—media websites, Facebook, and Twitter (X)—were analyzed. The findings reveal that every analyzed dimension of media coverage—including source selection, coverage volume, topic framing, emotional tone, construal level, and politicization through partisan references—was profoundly shaped by ideological bias. Notably, substantial differences were observed in how news was crafted for different digital platforms, highlighting the pivotal role of platform design in shaping media content. The study demonstrates that media coverage correlates strongly with both ideological orientation and degree of partisanship, with highly partisan outlets producing more extreme narratives than their moderate counterparts. These outlets relied more heavily on socio-psychological signals—such as out-group partisan references and abstract representations—to craft their narratives. Interestingly, some aspects of coverage showed non-linear relationships with partisanship, suggesting that the ideological dynamics influencing news framing are more complex than previously assumed. Network analysis enabled the construction of the first comprehensive “connectomes” of media coverage on polarizing issues, exposing deep ideological cleavages within the U.S. media ecosystem. These connectomes revealed how media outlets with differing ideological leanings prioritized distinct individuals and sources and showed the dominance of partisan over nonpartisan outlets in both source selection and coverage emphasis. Audience engagement analysis revealed stark differences in how Americans interacted with and disseminated polarizing news. Audiences of liberal and conservative media were drawn to different topics, and within each ideological sector, patterns of engagement further distinguished the audiences of moderate versus strongly partisan outlets—suggesting divisions within ideological groups. Unexpectedly, during the COVID-19 pandemic, audiences of conservative media exhibited a progressive increase in preferential engagement with and dissemination of COVID-19-related news across Facebook and Twitter (X), indicating that polarized media coverage can drive shifts in audience engagement patterns at a national scale. Geo-located analysis of Twitter data revealed that the dissemination of news containing partisan references, as well as accurate information countering COVID-19 misinformation, predicted local patterns of vaccination and mortality, even after accounting for socioeconomic and political variables. These findings provide more direct evidence that media narratives influenced shifts in attitudes, beliefs, and behaviors, potentially contributing to the polarization of health outcomes during the pandemic. The dissertation also identifies socioeconomic pressures as highly significant predictors of media coverage. National indices related to COVID-19—including intensive care unit admissions, deaths, hospitalizations, and vaccination rates—as well as economic indicators such as unemployment and the S&P 500 index, were significant predictors of COVID-19 media coverage, with differential effects across partisan media outlets. Socioeconomic stress also interacted with patterns of partisan news dissemination to help explain the polarized COVID-19 outcomes across U.S. communities. Building on these findings, this dissertation proposes the Media Accelerated Polarization (MAP) model, which posits that ideological fragmentation within the media ecosystem, combined with structural and algorithmic digital factors, interacts with socioeconomic pressures to systematically produce polarized news coverage. This coverage, in turn, shapes audience engagement and persuasion patterns across digital platforms, fueling the selective amplification of politicized content. These dynamics then interact with local socioeconomic conditions to deepen ideological divisions and intensify mass polarization. The conceptual framework developed in this work—together with the MAP model—contribute to the foundation of the emerging field of institutional media polarization, and may advance our understanding of how ideological, structural and other biases within the media ecosystem shape news production and dissemination in ways that reinforce and exacerbate partisan divisions. By linking institutional media polarization to audience polarization, as well as to the initiation, persistence, and escalation of mass partisan divides, this research highlights key mechanisms through which the media may influence societal fragmentation. A deeper understanding of these dynamics could inform the development of strategies aimed at mitigating polarization and fostering greater social cohesion in an increasingly divided public sphere.