Aristotle's (trans, 1984) theory of persuasion centered on three elements of the rhetorical situation: the speaker, the speech, and the audience. He held that each of these elements was the source of a particular form of artistic proof.
The speaker accomplished persuasion by means of ethos, or credibility. For Aristotle, credibility was generated by means of the speech, not by the rhetor's prior reputation. This is intrinsic credibility: in the course of presenting the speech and publicly reasoning through the issues, the speaker displays who he or she is. A similar idea is Lake's (1983, 1990) concept of the enacted argument. From the performed speech itself the audience will come to estimate the speaker's expertise, trustworthiness, and good will. Aristode said that high levels of these three components of ethos led to persuasion and constituted the "speaker" contribution to the process. Contemporary persuasion theories continue to feature the importance of credibility (Chaiken, 1980; McCroskey & Tevan, 1999; Petty & Cacioppo, 1986; Petty, Cacioppo, & Goldman, 1981), but the modern focus is on extrinsic credibility, such as is created by explicitly stated qualifications in an introduction. Here, intrinsic credibility is our interest.
For Aristotle, the key element of the speech was its argument, its logos. Logos does not mean logical proof, as it is often glossed. Logos means "word" in Greek, and the Greeks considered that their language was a sort of encyclopedia of all that was knowable (Jaeger, 1939). By manipulating words, they believed they could discover new knowledge (this may have been an impulse to their development of formal logic). So Aristotle defined logos as proof drawn from facts about the subject matter, facts expressed in words. The facts (or evidence) controlled the arguments. Rhetorical arguments, he said, are of two sorts, what today we would call argument schemes. The enthymeme is a somewhat systematic word manipulation, a rhetorical approximation to logic's syllogism. Enthymemes rely on the audience to supply part of the content so that not everything need be said (Bitzer, 1959), and this feature generates their natural informal presentation, making them appear loose in comparison to the forms of academic logic. The other sort of rhetorical argument was the example, a simple approximation of logic's induction. Where a philosopher might require a large sample for generalization, Aristotle noticed that audiences could be moved by a single example or a handful of them. Whether the rhetor argued ethymematically or by example, he or she needed to express evidence in words, and this led to the persuasive power of logos.
The role of evidence in argumentation has not been well theorized. Hample (2006; Hample et al., 2000) has criticized the common conceptualizations of evidence on the grounds that they consist of poor typologies (e.g., narratives, statistics, analogies, causal evidence, testimonies), category systems that are not mutually exclusive or mutually exhaustive. Another telling critique has been that in empirical studies evidence has sometimes been classified as merely being a quotation or not, and this left the evidentiary content intact when the quotation marks and attributions were removed (Kellermann, 1980). Here we adopt a non-typological approach to the assessment of evidence. Hample (2006; Hample, et al., 2000; Skubisz, 2010, 2011) has developed a series of scales on which recipients rate the evidence that they receive. This research program has converged on four dimensions on which ordinary actors perceive evidence: clarity (e.g., clear, hard to understand), morality-effectiveness (e.g., wise, true), artistic quality (e.g., humorous, artistic), and whether or not it seems prejudiced (e.g., racist, sexist). These scales will be used in the current study to operationalize this aspect of logos.
Our particular implementation of Aristotle's theory permits us to make one clear point about the ordering of logos, ethos, and persuasion effects. Since we are viewing credibility as intrinsic, it must be created by the message content. Credibility might possibly generate a sort of halo effect, causing retrospective or biased evaluations of message content, but we theorize that the effects of evidence will be mediated by credibility rather than the other way around. (If we were working with extrinsic credibility, we would have to consider the possibility that a source's qualifications could affect judgments of the message content.) On one hand, the idea that content generates intrinsic credibility might suggest that the effects of evidence quality should exert themselves only by means of credibility's power, and therefore be completely mediated by ethos. On the other hand, Aristotle treated both logos and ethos as sources of persuasion, suggesting that evidence should directly affect both credibility and persuasion. So our first hypothesis tests both views:
H1: Evidence will affect credibility, credibility will influence attitude change, and the effects of evidence on attitude change will be at least partly mediated by credibility.
Our design and analysis permit us to test between complete mediation (no direct effect of evidence on persuasion) and partial mediation (evidence directly results in both credibility and persuasion). We consider that Hypothesis 1, if successful, will display the core relationships among speaker, content, and outcome.
Aristotle's third source of artistic proof was the audience, and more particularly the audience's emotional state. Pathos refers to the speaker's ability to use elements of the speech to connect with the listeners' values and feelings. A successful pathetic appeal engaged the audience's emotions to make them receptive to the speaker's position. The Rhetoric contains an early survey of the psychology of affect, indicating which emotions were felt by which people and for what reasons (in 4th century BCE Athens, of course). We have included several constructs that respect the persuasive energies that originate in the audience, particularly the possibility of fear.
Our procedures involve messages proposing that audience members should address various health-related risks. Two variables, severity of risk and susceptibility to risk, reflect audience members' own estimates of the danger they face. Severity and susceptibility were key elements of the health belief model (HBM) of persuasion, and have continued to play important theoretical roles in the intellectual descendants of the HBM (Bandura, 1998; Janz & Becker, 1984; Rimal & Real, 2003; Turner, Rimal, Morrison, & Kim, 2006; Witte, 1992; Witte & Allen, 2000). People's estimates of risk are predictive of their health- related behaviors, beliefs, attitudes, and information-seeking interests.
Our next research objective is to clarify the way that risk perceptions fit into the model developed to test Hypothesis 1. If risk perceptions do belong in the model, that means that the models described by H1 are misspecified and should be regarded as nothing more than comparisons to earlier, equally misspecified, models. Risk perceptions should be endogenous, that is, they ought to result from other elements of the persuasion event. Each of our experimental messages indicated that the public was at some risk--of supergerms, of meningitis, or of inadequate nutrition. Awareness of risk is held to cause adherence to any recommendations designed to ameliorate it (Fischhoff, Slovic, Lichtenstein, Read, & Combs, 1978). We follow Fischoff et al. and many others (Bandura, 1988;Janz & Becker, 1984; Rimal & Real, 2003; Turner, Rimal, Morrison, & Kim, 2006; Witte, 1992; Witte & Allen, 2000) in considering that risk perception has two elements, perceived severity of threat and personal susceptibility to the danger. The degree to which each respondent felt at risk varied naturally.
We theorize that risk perception is intrinsic to the messages--that is, it is developed or suppressed by the information and arguments provided to recipients. (In this way, it is like intrinsic credibility.) Consequently, our theory is that evidence will influence risk perception, and that risk perception will affect persuasion. Therefore,
H2: Risk perceptions will be due to message evidence, and will then affect attitudes.
We will test Hypothesis 2 by adding risk perception to the best model developed to address Hypothesis 1. In fact, our thinking about risk perceptions implies that the model described in...