A Digital Hypodermic Needle? Essays On The Impact Of Misinformation, Framing, And Images On American Public Opinion In The Internet Age

Misinformation is en vogue. In recent years, misinformation has gotten blamed for seemingly everything—from the ascendancy of unlikely politicians to the thousands of unvaccinated Americans who have perished due to COVID-19. The term "misinformation" has become part of the American imaginary, bringing along with it serious normative concerns about the mass public being susceptible to unhealthy variants of political persuasion. This worldview, however, presupposes that misinformation operates like the proverbial hypodermic needle of the early twentieth century models of public opinion formation, ever-powerful to influence and persuade the public at a massive scale. But does empirical evidence bolster such a view? This dissertation engages with both this question and the longstanding scholarly debate between those who think the media has inculcation-like effects on public opinion and those who see the media as having limited effects that are only detectable among those with key predispositions. I extend this debate into the digital realm with a focus on misinformation. To do so, I employ a three-paper approach in this manuscript. In the first two chapters, I review the relevant literature and provide a broad, data-driven overview of how the American media, political elites, academia, and mass public have been discussing misinformation, disinformation, and fake news. Although the terms have become ubiquitous in everyday parlance, scholars are barely beginning to understand how consumption of this type of content influences political attitudes. My next two chapters contribute to that enterprise. Chapter 3 explores the differential impact of misinformation and another, more well-studied form of political communication: framing. Using an externally-valid survey experiment, I find that framing, not misinformation, drives _general_ attitudes about certain aspects of the 2017 Unite the Right rally in Charlottesville, VA. In addition to replicating findings from the framing literature, which show that framing is powerful at persuading audiences, I find significant heterogeneities in treatment effects across partisan identities in this chapter. This suggests that my theorized mechanism—congruence—moderates the effect of both misinformation and framing in persuading audiences. One of the limitations of this chapter is that it focused only on text-based misinformation—yet, the existence of multimedia-driven online platforms means that individuals encounter misinformation "in the wild" that is enhanced, so to speak, with images. To address this limitation, Chapter 4 expands the nascent visual politics literature to ask whether text-based misinformation that is enhanced with misinformation-containing visuals impacts attitudes more than text-based misinformation without them. Using an externally-valid survey experiment with a factorial design about the January 6th, 2021 riots at the US Capitol, this chapter finds that images matter and can interact with textual misinformation to showcase significant effects among the mass public—but these effects are limited to very specific situations (at least in this experiment). This experiment further shows heterogeneous treatment effects among partisans. I conclude this dissertation with some thoughts on future research and discuss the implications of my findings for American democracy writ large.