Although there is a large body of research on the effectiveness of a variety of strategies for designing health messages, much of this research consists of experimental studies concerned with the immediate persuasive effects of message exposure on isolated individuals. Thus, these studies typically do not measure talk as an outcome of health message exposure. However, research suggests that talk about health messages can influence message effects. Specifically, the nature of talk in response to health messages (i.e., whether it is supportive or unsupportive of the message) and the degree of message elaboration generated by talk have important implications for the effects of messages on attitudes, beliefs, and behavior. Research also suggests that intense emotion plays an important role in generating talk about health messages.

The purpose of this dissertation was to investigate the effects of health messages containing shifts in emotional valence on the likelihood of talk, features of the talk generated, as well as psychological and behavioral outcomes, when compared to messages that were singular in their emotional valence. This research was the first to conduct an experiment to compare the effects of two persuasive health messages containing shifts in emotional valence -- from a positive to a negative emotion (pos-neg) or from a negative to a positive emotion (neg-pos) -- with two single-valence emotional messages, one negative and one positive, on outcomes related to talk and persuasion. Attitudes, beliefs, and behavioral outcomes were measured at two time points, immediately following message exposure and again one week later. Information about conversations related to the message and/or behavior were also collected at the one-week follow-up.

Emotional shift messages were found to generate more talk than single-valence messages because they were more emotionally intense and elicited greater message processing. These mediating variables were also responsible for the effect of emotional shift messages on persuasive outcomes at Time 1 and Time 2. This demonstrates the potential of emotional shift messages to promote both talk and persuasion. Additionally, messages ending with negative emotion were more effective in influencing beliefs and attitudes at Time 1 as well as generating not only more talk but more positive talk compared to messages that ended with positive emotion. This research also revealed that intention to avoid texting while driving immediately following message exposure had a greater influence on beliefs and behavior at Time 2 than talk. This suggests that the characteristics of talk may be a more important determinant of message effects than whether talk occurred.