Past health campaigns were rarely research-driven; consequently, publicly funded campaigns have failed, leading to no effects or even shifts in attitude counter to those intended. Recent success with research-driven campaigns illustrates the importance of further research into the mechanisms of message processing among audiences. At the same time, media theory has shifted over the last several decades from "effects" models to "processing" models, yet most communication research still neglects to examine processes as such---over time. In this dissertation, the process of persuasion in response to anti-drug public service announcements is examined in a young-adult sample over time. Three studies are used ultimately to test theoretically generated models of argument and audiovisual-feature effects on reported persuasion in audience members, provided over time during message receipt. Study 1 examines static arguments, developing a method for estimating the strength of an isolated argument for an individual audience member; findings suggest that argument strength is contextual and sensitive to individual differences. Study 2 concerns how these arguments combine in a multi-argument message, and how time and order affect persuasive outcomes. Changes in belief depended on the discrepancy between the current argument strength and the current belief, and evaluations were subject to a recency effect. Study 3 combines the effects of overt arguments with audiovisual features in actual advertisements. Interactions between message features and individual characteristics dominated the results, suggesting that the mechanisms of action for arguments and other message features are complex.

Conclusions can be classified into 4 themes: the contextual nature of argument strength, the role of involvement in biased processing, the relationship between information density cognitive requirements, and the effects of time and order on persuasive outcomes. Practically, these results imply that, while using scholarship to design messages is superior to atheoretical work, reliance on main-effect relationships between message features and persuasion is unlikely to lead to desired attitude change---i.e., designing messages with high "argument strength" and/or message-sensation value, without consideration of how these features will interact with one another and with characteristics of the audience, is not enough to produce a message with a high likelihood of success.