Audience costs, which are defined as political costs for a leader generated when the leader fails to follow through on international commitments, are an important concept in international relations to understand the causes of war and peace. However, despite its prominence, evidence for audience costs is mixed at best. In this dissertation, by conducting a series of survey experiments in the United States, I examine a variety of causal mechanisms and observable implications of audience cost models. In Chapter 2 I explore the role of the media. The result of a survey experiment shows that the media's negative framing of a leader's backing down from a threat increases audience costs, but only after subjects are informed of the leader's justification. Chapter 3 tests the idea that the magnitude of audience costs depends on an individual's personality traits. A pilot experiment using a sample of undergraduate students does not support this expectation. Lastly, I compare the credibility of public versus private threats in Chapter 4. A conjoint survey experiment demonstrates that as many proponents of audience cost theory suggest, public threats are perceived as more credible, and the sources of their credibility are domestic audience costs.