Since the beginning of the 20th century, persuasion researchers have been concerned with answering questions regarding the major determinants of message effectiveness (e.g., Boster & Mongeau, 1984; Chaiken, 1980; McCroskey, 1969; O'Keefe, 1990; Petty & Cacioppo, 1986a; Stiff, 1986). Much of this research has explored the important components of the persuasion process by positing and testing research questions and hypotheses to reveal the conditions under which messages are most persuasive.
In their quest to extend the persuasion literature, researchers sought to address questions centered generally around knowing two things: what kind of persuasive message is most effective (e.g., Allen et al., 1990; Hale, Mongeau, & Thomas, 1991; McCroskey, Young, & Scott, 1972), and what factors contribute to the persuasiveness of a message (e.g., Arnold & McCroskey, 1967; Harte, 1976; McCroskey & Young, 1981; Petty & Cacioppo, 1984; Slater & Rouner, 1997). In order to answer these questions, the dominant research approach has been to vary the content of a persuasive message in experimental conditions and to examine subsequently its persuasive outcome. In doing this, scholars have examined systematically the persuasiveness of a variety of message types such as refutational (e.g., Allen et al., 1990; Hale et al., 1991; McCroskey et al., 1972; O'Keefe, 1998; Winkel, 1984), narrative (e.g., Ah Yun & Massi, 2001; Kazoleas, 1993; Kopfman, Smith, Ah Yun, & Hodges, 1998), and statistical appeals (e.g., Ah Yun & Massi 2000; Dickson, 1982; Yalch & Yalch, 1984).
The research focusing on persuasive messages has increased our understanding of various message effects in three important ways. First, we know that various message types successfully elicit the desired attitude (see, for example, McCroskey, 1967, 1969; Nadler, 1983; Reinard, 1988). With respect to refutational messages, for example, Allen et al. (1990) compared the persuasive effect of three different messages. One message presented only positive statements, the second presented a two-sided refutational message, and the third was a two-sided nonrefulational message. Allen and his colleagues found that the refutational message was more persuasive than the one-sided message, which, in turn, was more persuasive than the two-sided nonrefutational message. Research such as this has enabled scholars to determine the types of messages that are most persuasive and elicit desired attitudes.
Second, the research has provided explanations as to why various message types are persuasive (e.g., Harte, 1976; Rosenthal, 1971; Stiff, 1986). For example, in an examination of fear appeals, Witte (1992) developed a model to explicate the processes involved when individuals are exposed to fear appeals. Her model suggests that fear appeal messages that induce a perceived threat in a receiver will promote a positive health behavior change so long as that person believes that a behavioral change is possible and that it will successfully avert the threat presented in the message.
Finally, previous research has explained that any number of intervening factors may explain the effect of a given message. For example, Ah Yun and Massi (2001) found that sympathy, happiness, and perceived vividness mediated the relationship between content variations of a narrative message and a person's attitude toward signing an organ donor card.
This body of research suggests that understanding which types of messages are most persuasive, why specific messages are persuasive, and which factors mediate the relationship between the type of message and its outcomes are important components of understanding persuasive messages. Although a great deal of the persuasive message effects research has answered the above questions in regards to many specific message types-for example, refutational and narrative messages-scholars have not undertaken such an examination of statistical appeals to determine why they are persuasive and which factors mediate the relationship between statistical messages and their effects. …