UCLA

A recent wave of studies on political advertising has demonstrated the small (and sometimes absent) persuasive effects of television and digital advertising in American elections. However, less is known about how the types of messages and dosage of political advertisements affect people and even less about how dosage interacts with messaging. This dissertation reports the results of an individual-level randomized field experiment on 778,882 registered voters conducted over Facebook during the 2018 Republican gubernatorial primary in Texas. The ads, produced by the incumbent governor’s campaign, were designed to increase turnout among likely Republican voters in the state. I find small but discernible increases in voter turnout, with some message strategies resonating more than others. In particular, a social pressure message emphasizing civic duty outperformed a partisan identity appeal. I further demonstrate that among the more effective messages, delivering a higher volume of advertisements yields larger returns in voter turnout (up to about 1.2 percentage points increase for 20 exposures of a social pressure ad). Finally, when considering race as a moderating variable, the effects were not significantly different across racial groups when the advertisements emphasized partisan identity instead of racialized policies, suggesting that non-racial appeals had uniformly modest effects across demographics. These findings contribute to our understanding of digital campaign effects by highlighting the roles of message framing and exposure frequency in voter mobilization.