Nevins, A. (1947). Fruits of Manifest Destiny, 1847–1852. New York, NY: Charles Scribner’s Sons.
Nichols, M. H. (1952). Kenneth Burke and the ‘new rhetoric.’ Quarterly Journal of Speech, 38, 133–144. Perelman, C., & Olbrechhts-Tyteca, L. (1969). The new rhetoric: A treatise on argumentation. Notre Dame, IN: University of Notre Dame Press.
Portolano, M. (2009). The passionate empiricist: The eloquence of John Quincy Adams in the service of science. Albany: State University of New York Press.
Postman, N. (1985). Amusing ourselves to death: Public discourse in the age of show business. New York, NY: Penguin Books.
Putnam, R. D. (2000). Bowling alone: The collapse and revival of American community. New York, NY: Simon & Schuster.
Putnam, R. D. (1997). The decline of civil society: How come? So what?” Optimum: The Journal of Public Sector Management, 27, 27–37.
Ray, J. W. (1978). Perelman’s universal audience. The Quarterly Journal of Speech, 64, 361–375.
Reid, R. F., & Klumpp, J. F. (2005). American rhetorical discourse (3rd ed.). Long Grove, IL: Waveland Press.
Rossiter, C. (Ed.). (1961). The federalist papers. New York, NY: New American Library.
Scott, R. L., & Smith, D. K. (1969). The rhetoric of confrontation. Quarterly Journal of Speech, 55, 1–8. Skocpol, T. (2003). Diminished democracy: From membership to management in American civic life. Norman: University of Oklahoma Press.
Sutton, E. W., & Rackham, H. (Eds.). (1983). Cicero in twenty-eight volumes: De oratore. (Vol. 3). Cam-bridge, MA: Harvard University Press (Loeb Classical Library).
Thonssen, L., & Baird, A. C. (1948). Speech criticism: The development of standards for rhetorical appraisal. New York, NY: Ronald Press.
Tracy, K. (2010). Challenges of ordinary democracy: A case study in deliberation and dissent. University Park: Pennsylvania State University Press.
Tulis, J. K. (1987). The rhetorical presidency. Princeton, NJ: Princeton University Press.
Vaughn, S. (1980). Holding fast the inner lines: Democracy, nationalism, and the Committee on Public Information. Chapel Hill: University of North Carolina Press.
Walzer, A. E., & Beard, D. (2009). Historiography and the study of rhetoric. In A. A. Lunsford (Ed.), The SAGE handbook of rhetorical studies (pp. 13–33). Thousand Oaks, CA: Sage.
Ward, J. (1759). A system of oratory. London: J. Ward. Whately, R. (1963). Elements of rhetoric: Comprising an analysis of the laws of moral evidence and of persuasion, with rules for argumentative composition and elocution (D. Ehninger, Ed.). Carbondale: Southern Illinois University Press. (Original work published 1828)
Windt, T. O., Jr. (1972). The diatribe: Last resort for protest. Quarterly Journal of Speech, 58, 1–14. Wood, G. S. (1969). The creation of the American republic, 1776–1787. New York, NY: W. W. Norton.
Wood, G. S. (1974). The democratization of the American mind. In Leadership in the American Revolution. Washington: Library of Congress.
Zappen, J. P. (2009). Kenneth Burke on dialectical-rhetorical transcendence. Philosophy and Rhetoric, 42, 279–301.
Zimmerman, J. (1989). The New England town hall meeting: Democracy in action. Westport, CT: Praeger.
CHAPTER 2
The Effects of Message Features
Content, Structure, and Style
Lijiang Shen and Elisabeth Bigsby
Consider a generic model of the persuasion process. Affective, cognitive, and behavioral processes mediate the relationship between the message-related variables (source, message, recipient, and context) and the desired outcome variables (attitude, intention, and behavior; Petty & Wegener, 1998). Although communication scholars are interested in the mediating processes of persuasion, as reflected in the chapters on theories and models in this handbook, the study of message features distinguishes communication research in persuasion from that in other disciplines such as psychology. The study of message features refers to aspects of communication itself; and as pointed out by Dillard and Pfau (2002), “questions concerning how messages might be designed to produce the greatest suasory impact lies at the very center of persuasion research” (p. xvi; see also Miller & Burgoon, 1978). In this chapter, we strive to present a review of research findings regarding the effects of message features on persuasion with a discussion of both practical and theoretical implications.
Persuasive messages are often broken into parts for study and analysis with researchers examining the persuasive influence of specific message features. Message construction also encourages this type of conceptualization and often addresses one message component at a time. From a campaign design perspective for example, decisions on the message topic or theme are made prior to any work on arguments or visuals. The topic, theme, or story being told (including plot and characters) is the content of the message (Lang, 2000; Stephenson & Palmgreen, 2001). Essentially, what the message is about. Closely aligned with the content is the presentation or structure of the message’s arguments. The number of arguments the message contains, the order of the arguments, and whether or not points of opposition are acknowledged and/or addressed. The final major message component, style, generally includes language use like word choices and figure of speech. In mediated messages, style also refers to features like edits, music, and pacing (Geiger & Reeves, 1993; Lang, 2000; Morgan, Palmgreen, Stephenson, Hoyle, & Lorch, 2003). Each message feature influences the persuasion process in unique ways. The following sections will outline important research in this area, with an emphasis on reviewing meta-analytic studies when possible, and attempt to provide an overview of how message features may work together to influence persuasion.
Message Content
* * *
Inherent in the theme of a message is the supporting evidence and whether or not opposing viewpoints will be acknowledged. For example, the theme of a political campaign advertisement is often built on the underlying goal of the message: support for our candidate or attacking an opponent. The following section focuses on the results of meta-analyses on the effects of two content-related features that have received substantial attention in persuasion research, type of evidence and message sidedness.
Type of Evidence
All persuasive messages advocate a particular position with the goal of getting the receivers to think or behave a particular way. The advocacy, therefore, more or less states a claim that the source expects the receivers to accept as true (e.g., indoor tanning damages your skin), or actions for them to enact (e.g., you should not tan indoors). Toulmin (1969) argues a claim should be backed by data to make an argument strong. Data is therefore the basis for persuasion and consists of evidence such as factual information and the reasoning behind the claim. Data are linked to the claim by a warrant, which legitimizes the claim by demonstrating that the data are relevant. This model of argument suggests that the strength, as well as type of evidence presented to support the message advocacy, may directly impact message effectiveness (see also Reinard, 1998; Reynolds & Reynolds, 2002).
In general there are four types of evidence. Statistical evidence presents statistics such as frequencies and percentages to support the claim. For example, “More than 1 million skin cancer cases are diagnosed annually in the U.S.”, or “Between 40–50% of Americans who live to age 65 will at least once have basal cell carcinoma (BCC) or squamous cell carcinoma (SCC).” Testimonial evidence uses a person’s personal experience, eye-witness account, or personal opinion to support the claim (including expert testimony). For example, “It was hard, but I quit smoking so you can do it, too.” Anecdotal evidence is evidence that is based on a person’s observations of the world. It is a personal interpretation of or opinion toward a target and is often subjective in nature. Analogical evidence use analogies to support a claim, comparing one idea/situation to another. Analogies are mainly useful when dealing with a topic that is novel.
Neither anecdotal nor analogical evidence are as widely used or studied as statistical and testimonial evidence, likely because the latt
er are considered strong types of evidence. A meta-analysis examining the effectiveness of testimonial evidence provides support for this claim. The average effect size was r = .23, k = 16, N = 2,800 for studies with group interactions, and r = .25, k = 14, N = 1,920 for studies without group interactions (Reinard, 1998). However, Allen and Preiss (1997) conducted a meta-analysis comparing the persuasive impact of testimonial and statistical evidence and their results suggested that statistical evidence is more pervasive than testimonial evidence (r = .10, k = 15, N = 1,760), with homogenous effect sizes. Hornikx (2005) hypothesized that the inconsistent findings may be a result of the differences in evidence types and their conceptualizations and operationalizations. In a comparison of studies, Hornikx tentatively claimed anecdotal evidence to be the least effective; however, a formal statistical analysis of the data was not conducted. Hornikx was focused on describing the methodological differences between the studies, and instead provided a count of the number of studies that found certain results. For example, studies that found (1) statistical evidence to be more persuasive than anecdotal evidence (k = 6), (2) anecdotal evidence to be more persuasive than statistical evidence (k = 1), and (3) no significant difference between statistical and anecdotal evidence (k = 5; Hornikx, 2005).
Message Sidedness
In addition to having a strong argument, the context of the argument presentation is important to the persuasion process. That is, whether or not the message will reference the opposition. Messages that only include arguments that support the position of the persuader are one-sided messages. A one-sided message does not make statements about the opposition’s view or even acknowledge the existence of an opposing point of view. Two-sided messages include both supportive arguments and an acknowledgement or mention of the opposition’s arguments. A two-sided message can be non-refutational, that is, it does not provide counter-arguments against the opposing view; or can be refutational and provide counter-arguments to demonstrate the superiority of their own arguments and advocacy over the opponents’.
Clearly, the question of which is more effective, one- or two-sided messages, has gained much attention in persuasion research. In fact, three meta-analyses, which report inconsistent conclusions, have been published on this topic (Allen, 1991; 1998; O’Keefe, 1999). Allen (1991) found a slight advantage for two-sided messages (r = .04, k = 26, N = 7,547) regarding persuasive effectiveness. With more data (primary studies), Allen (1998) confirmed the conclusion of his previous study: r = .03, k = 70, N = 10,580. On the contrary, O’Keefe (1999) found no difference between one- and two-sided messages regarding their overall persuasive effect (r = −.00, k = 107, N = 20,111). Although it appears that as more and more recent studies were included, the difference in effectiveness disappears, a closer examination of these meta-analyses reveals a new picture, an understanding of instances when a one-sided message will be more persuasive than a two-sided message and vice versa. Ultimately, there was more consensus than inconsistency between Allen and O’Keefe’s analyses when potential moderators are considered.
Potential Moderators
Type of two-sided message (i.e., refutational vs. non-refutational) has been found to be a significant moderator. Allen (1991) found that one-sided messages were actually more effective than two-sided messages when they are non-refutational (r = -.06, k = 6, N = 1,819); while the pattern is reversed when the two-sided messages are refutational (r = .08, k = 19, N = 5,624). Again, this finding is confirmed in the 1998 piece. One-sided messages were more effective than non-refutational two-sided messages (r = -.09, k = 26, N = 3,159); and the pattern is reversed when the two-sided messages are refutational (r = .07, k = 43, N = 7,317). With more data, O’Keefe’s (1999) findings were very similar. One-sided messages were again found to be more persuasive than non-refutational two-sided messages (r = -.05, k = 65), but less persuasive than refutational two-sided messages (r = .08, k = 42). This shows that the inconsistency in overall conclusion was due to the fact that O’Keefe’s analysis included more studies that compared one-sided and non-refutational two-sided messages, which privileged one-sided messages (65 in O’Keefe, 1999, vs. 26 in Allen, 1998), and canceled out the advantage of refutational two-sided messages over one-sided messages.
It is generally believed that refutational two-sided messages are more effective because the representation of opposing statements reduces counter-arguing by the recipients. Instead of trying to think of the opposition’s arguments to combat the message advocacy, they are presented in the message. This in turn may lead to more positive cognitive responses than would occur in non-refutational messages (McGuire, 1985).
Audience Favorability
Allen (1991, 1998) and O’Keefe (1999) both found audience favorability to be a significant moderator. Allen (1991) found that one- and two-sided messages did not differ in their persuasiveness when the audience’s pre-existing attitude was favorable, that is, toward the message advocacy (r = .00, k = 8, N = 2,952); while two-sided messages were more effective when the audience’s pre-existing attitude was unfavorable (r = .08, k = 9, N = 1,195). Allen (1998) suggested the interaction between type of two-sided message and audience favorability could be a moderator because the sample was heterogeneous. However, because of the small number of studies available, a formal test of this potential moderator was impossible. On the other hand, while O’Keefe (1999) also found audience favorability appeared to have some influence on the persuasion process, his conclusion was not exactly consistent with that of Allen (1991). O’Keefe found one-sided messages were significantly more effective than two-sided messages when the audience had an initial attitude toward the topic, whether it was initially favorable (r = −.14, k = 10) or unfavorable (r = −.11, k = 9). There was no such difference between one- and two-sided messages when the audience was initially neutral (r = −.02, k = 36).
Advertisement Versus Non-advertisement
One key difference between Allen (1998) and O’Keefe (1999) was the number and types of primary studies included in the meta-analysis. More advertising research was included in O’Keefe’s study, which appears to be a moderator as well. O’Keefe found that whether the message was an advertisement or non-advertisement did not make a difference in the effectiveness of one- vs. two-sided messages (r = .00, k = 35 for advertising messages, and r = −.00, k = 72 for non-advertising messages). However, when combined with type of two-sided message, whether or not the message was an advertisement did appear to have some influence. For non-advertising messages, there was a pattern in which refutational two-sided messages were more effective than one-sided messages (r = .08, k = 33), which in turn were more effective than non-refutational two-sided messages (r = −.07, k = 39). This pattern did not hold when only advertisements were examined; there was no significant difference in persuasion between two-sided messages of either kind and one-sided messages.
O’Keefe (1999) also assessed the impact of one- and two-sided messages on credibility. Overall, there was a significant advantage in perceived credibility for two-sided messages (r = .09, k = 56, N = 6,937). However, this advantage was only significant for advertising messages (r = .15) and not for non-advertising messages (r = .04). O’Keefe suggested credibility may be jointly influenced by topic and type of two-sided message, but there was not sufficient data to test that possibility.
Including only studies that used advertisements, Eisend (2006) conducted a meta-analysis examining differences between one- and two-sided messages based on message structure, the persuader (marketer), and audience variables. Unlike O’Keefe, Eisend found two-sided advertisements were more effective than one-sided advertisements (r = .07, k = 217); although the effect size was small. Two-sided messages also significantly increased source credibility (r = .22, k = 32, N = 1,554), perceived novelty (r = .35, k = 4, N = 185), and positive cognitive responses (r = .09, k = 10, N = 465); they decreased negative cognitive responses (r = −.18, k = 13, N = 615) and resulted in more favorable attitudes toward the message (r = −.05, k = 56, N = 3,305) and
the brand (r = .12, k = 65, N = 3,152). Several moderators, conducted with two-sided messages only in regression models, were found to significantly impact one or more of the outcome variables.
For example, greater amounts of negative information presented increased favorable attitudes toward the brand (unstandardized regression coefficient B = .47, k = 40, p < .05), and increased purchase intention (B = .63, k = 22, p < .001), but it also increased negative attitudes toward the advertisement (B = −.39, p < .001). When negative information was placed first, source credibility (B = −.64, k = 32, p < .001) and favorable attitude toward the brand (B = −.12, k = 48, p < .001) decreased. When negative information was placed last, favorable attitude toward the brand increased (B = .15, k = 40, p < .001). Eisend’s results suggest that negative information should be included at the end of an advertisement; but this may not be true with other types of messages (i.e., health and political).
O’Keefe’s (1999) meta-analysis included more advertising studies (n = 35) than Eisend’s (2006) meta-analysis (n = 29), and between the two authors, only 13 articles overlapped. The analyses for the most part also examined different outcome variables, so it seems likely the differences between the meta-analyses account for some of the differences among their results. One result, however, was consistent between them. Both found two-sided advertisements to be more credible than one-sided advertisements.
Message Structure
* * *
Message structure concerns primarily with how either the data or the claim are presented in persuasive communication. Two features related to message structure have received some attention in the persuasion literature: (1) climax versus anticlimax order of arguments and (2) conclusion explicitness.
The SAGE Handbook of Persuasion Page 5