Prior research suggests that narratives may be an effective persuasive device. Despite the scholarly progress made in this area, uncertainty regarding the contexts in which this message format holds an advantage over other types of persuasive appeals persist. More specifically, statistical and narrative message formats have been compared extensively with mixed results. This dissertation aims to help mitigate these disparate findings in addition to answering a call for including a third, mixed format in the analysis. The three message formats---narrative, statistical, and combined---were tested in two studies with two health contexts. In the first, undergraduate students' intention to obtain HIV screening were examined; in the second, older adults' intentions to receive testing for Hepatitis C were analyzed. Data were collected from an undergraduate convenience sample for study one (N = 393) and Amazon's Mechanical Turk for study two ( N = 378). The data were analyzed with planned contrasts as well as hierarchical linear regression; supplemental analyses included coding open-ended responses and structural equation modeling. The studies' findings indicate an advantage for the combined format that embeds statistical information within the narrative. Implications for scholars and practitioners are discussed.