Affect and Persuasion

CHAPTER 10

James Price Dillard and Kiwon Seo

ognitive approaches to understanding design and message effects as they pertain to human thought and action have embraced affect. It aims to illustrate the issues that are most C a model of mind-as-computer (Miller, central to research on emotion and persuasion. 2003). Indeed, in areas such as artificial intelli- The third and final portion considers what future gence, one of the primary tools of the trade is research might look like by focusing attention on computer simulation. The mind-as-computer practices and ideas that have retarded research as metaphor pointed the way to understanding per- well as those that are likely to yield the greatest suasive processes in terms of input processes, benefit as we move forward. comprehension, associative memory, decision making, depth of processing, and recall: All of which led to improved understanding of how Conceptions of Affect messages create change in individuals. But, as data accrued and theoretical boundaries were Affect is an umbrella term that is meant to encom- tested, it became apparent that something crucial pass feelings of all sorts. Beneath that umbrella lie was missing: Computers didn’t feel. In sharp some important distinctions that implicitly and contrast, humans are often, perhaps always, explicitly guide the questions that researchers ask experiencing some kind of affect. Some research about affect as well as their corresponding answers. has shown that pre-existing moods caused Gaining an appreciation of the research requires a research participants to respond to persuasive closer examination of the various ways in which messages in very different ways. Other lines of affect has been conceptualized. inquiry established that the feelings produced by persuasive messages undergird purchasing behaviors, voting, and health decisions. Bipolar Valence What is meant by affect? The first section of this chapter reviews how that term has been used. One approach to thinking about affect is in The theoretical perspectives that correspond terms of valence, where valence means a contrast with different usage are also considered. The sec- between good and bad or pleasant and unpleasant. ond segment explores questions of message The strongest version of the valence perspective is

150

(c) 2013 Sage Publications, Inc. All Rights Reserved. Chapter 10. Affect and Persuasion——151 bipolar model. On this view, affect is invariably and fear, both of which occupy the same space in located at some point on a continuum that ranges the unpleasant–high arousal quadrant of the cir- from positive at one end of the scale and negative cumplex. They have also been justly criticized as at the other. Here, the underlying theoretical con- atheoretical in that they are inductive findings cept of affect is hydraulic: For every unit of good derived entirely from the application of dimen- feeling that accrues to an individual, one unit of sion reduction algorithms to judgments of the bad feeling is displaced. frequency or similarity of affect words. Despite these shortcomings, the PA models have main- tained a place in the research literature for 40 or Categorical Valence more years.

A less stringent assumption can be seen in studies in which good and bad feelings are con- Discrete Emotions and sidered separately, as when research participants Appraisal Theories are asked how good they feel and, in a separate question, how bad they feel (e.g., Watson & In simple terms, the discrete emotions approach Clark, 1994). In these cases, the conceptual rela- casts affect as a set of qualitatively distinct states tionship between valenced categories is one of that vary in intensity. Individuals may experience conceptual independence. Studying the effects of more or less happiness, sadness, or jealousy, but emotion on persuasion typically means using each emotion is categorically different from every positive and negative feelings as predictors. other emotion. One way in which they are distinct is their causes. Cognitive appraisal theories are frameworks Valence Plus Arousal for understanding discrete emotions. There are several such theories, but they all agree that emo- Another tradition of affect research expands tions arise from a particular form of cognition on the valence-only models by adding an arousal known as appraisals (Scherer, Schorr, & Johnstone, dimension. These pleasure-arousal (PA) models 2001). At the broadest level, appraisals are judg- view valence as the hedonic tone of the experiments of the extent to which relevant aspects of ence and arousal as the subjective experience of the environment are configured so as to promote energy versus lassitude. Together, pleasure and or inhibit an individual’s goals, where goals range arousal are thought to define “core” or elemental in abstraction from survival to arriving at work aspects of all affective experience (Russell & on time. When the person-environment relation- Feldman Barrett, 1999). Some theorists see value ship is seen as relevant and goal congruent, posi- in asserting the existence of an affective circum- tive emotions follow. Conversely, perceptions of plex. Pragmatically, this means that pleasure and relevance and goal incongruence yield negative arousal are orthogonal to one another. Any and emotions. But, there is a host of other appraisals all affects are arrayed in a circle defined by terms too that, in combination, create unique constella- that are roughly equidistant from the point at tions of appraisals that define different emotions. which the two dimensions cross. Thus, all affects For instance, joy/happiness derives from the can be understood in terms of their location. belief that (1) some event has transpired that is In this view, depression is defined in terms of (2) compatible with a previously existing goal unpleasantness and low arousal. Joy is both (e.g., Roseman, 2001). Guilt is the product of pleasant and energetic. One inarguable problem knowledge that one has failed to meet some per- for the PA models lie in the placement of anger sonally relevant standard for behavior. Other

(c) 2013 Sage Publications, Inc. All Rights Reserved. 152——PART II. Theories, Perspectives, and Traditions possible appraisals might be expressed as the the motivations associated with particular answers to questions such as: How certain is the emotions are best conceptualized at a more event (e.g., past vs. future)? What is the cause specific level. Disgust causes people to (e.g., self vs. other)? Can I control it (e.g., high recoil, hope prompts engagement, and vs. low)? Was it fair/legitimate (e.g., high vs. team pride encourages embrace. In other low)? Emotion theorists disagree about how the words, each emotion is functionally differ- content and number of appraisals that are ent and behaviorally specific form of required for a theory of emotion. However, they approach or avoidance. agree on the larger point that each of the emotions results from an exclusive pattern of ante- One important question that has been raised cedent judgments. about appraisal models is the order in which the Emotions themselves can be thought of as cognitive judgments occur. Lazarus (1991) draws distinctive, patterned responses that are observ- a line between two groups of judgments: Primary able in six domains. appraisals—relevance and goal congruity—have to do with the nature and significance of the • One of them is subjective experience. Fear event. Secondary appraisals, which include feels different from anger, which feels dif- notions of accountability and coping potential, ferent from elation. have more to do with defining the options for • Emotions are also represented in the physi- behavioral response. Another writer, Scherer ological domain, which includes changes in: (1984), suggests that appraisals are sequenced blood flow to different areas of the body, such that they move from rudimentary evalua- blood pressure, heart rate, nervous system tions, such as novelty and the intrinsic pleasant- activation, and muscle tension. ness of the event, through more cognitively • Neurological activity in distinct brain regions complex judgments, including cause, power, and is associated with different emotions (Phan, legitimacy. His multistep model does not require Wager, Taylor, & Liberzon, 2002). that every appraisal take place. Steps can be • Emotions correspond with alterations in skipped. But, the overall process is thought to expression. The most obvious of these alter- occur in a strict order. Whether one supposes ations are facial expressions, but other two-steps or several in the appraisal process, the aspects of behavior, such as gait and body assumption that appraisals are ordered implies lean, are expressive as well. This point questions concerning the speed at which the pro- underscores an often unappreciated fea- cess unfolds. ture of emotion: They are not merely inter- An answer can be had by turning attention to nal phenomena. Rather, emotional states the function of emotions. From an evolutionary produce behaviors that are seen and inter- perspective, emotions are decision-making pro- preted by others. grams that accept input in terms of appraisal • Emotions bring about changes in cognition. information, then output directions for behavior. Fear, for example, narrows the perceptual They are designed, evolutionarily speaking, to field and focuses attention on the threaten- provide adaptive solutions to problems that ing stimulus. Happiness enables associa- occurred with regularity in the physical and tions across conceptual categories. social environment that produced homo sapiens • Emotions prompt change in the realm of (Tooby & Cosmides, 2008). A key part of the motivation. Although all emotions energize argument here is that those problems were recur- behavioral tendency in terms of approach or rent. Though they might take many different avoidance (except, perhaps, contentment), forms, threats to individuals’ well-being occurred

(c) 2013 Sage Publications, Inc. All Rights Reserved. Chapter 10. Affect and Persuasion——153 with reliable frequency. And, obstacles to goal well-developed and widely accepted. But, theories achievement presented themselves on a regular of appraisal are not theories of persuasion. To basis. Due to the persistence of these problems, render them applicable to the issues that are the humans developed emotions as a standardized focus of this volume, persuasion researchers have means for recognizing and addressing them. One attempted to extend the framework at both of the primary advantages of emotions over cog- ends of the appraisal-emotion sequence (see nition is their speed. Relatively speaking, emo- Figure 10.1). For example, messages may be tions are very fast. This can be a distinct benefit viewed as representations of past, current, or when circumstances require an immediate future environments such that they instigate response. It may prove disadvantageous when an appraisals. Indeed, manipulations of message immediate impulse calls for behavior in one content to achieve variations in valence and prob- direction, but a more considered examination ability of an outcome are remarkably similar to of the situation suggests the reverse. Because appraisals of goal congruence and relevance persuasive messages so often prompt several (i.e., Lazarus’s primary appraisals). However, emotional and cognitive responses, studying per- where appraisal research emphasizes the overall suasive effects requires that researchers recognize content of the circumstances, persuasion inquiry and test for multiple, possibly conflicting, reac- has recognized the importance of noncontent tions to messages. features such as style and structure (e.g., Shen & Bigsby, this volume). The model in Figure 10.1. also extends the appraisal process by considering Understanding standard persuasion outcomes—beliefs, attitudes, Persuasion and Emotion intentions, and behavior—as potential consequences of emotion (see chapter 4 in this volume An Appraisal Model of Persuasion for a discussion of the relationships among various persuasion outcomes). These additions have As previously described, the focus of appraisal allowed researchers to pose theoretically engaging theories is the link between a specific type of cog- questions such as: What message features provoke nition (i.e., appraisals) and a particular type of emotional response? How exactly do emotions affective response (i.e., emotions). The theoretical influence persuasion? The next two sections machinery designed to explain that linkage is now examine those questions in more detail.

Figure 10.1 An Appraisal Model of Persuasion

Message Appraisals Emotion Persuasion Content Relevance Experience Beliefs Style Congruence Physiology Attitudes Accompaniments Certainty Neurology Intentions Cause Expression Behavior Control Cognition Legitimacy Motivation

Message Features unlikely to be resolved in the near future. Even so, That May Evoke Emotions probably there is agreement that the table is missing some elements that might be important Content to the study of persuasion and affect. For example, a more exhaustive account would include the As Shen and Bigsby (chapter 2 of this volume) self-conscious emotions of guilt, shame, pride, note, content can be thought of as the topic of a and embarrassment. Because these feelings may message, the theme of an appeal, or, in some figure prominently in consequential contexts cases, the story that is being told. Appraisal the- such as democratic action (Iyer, Schmader, & ory is especially useful for identifying message Lickel, 2007) and the radicalization of terrorists content that might arouse emotion. Table 10.1. (McCauley & Moskalenko, 2008), they warrant presents a summary of appraisals and their cor- the attention of persuasion researchers. Another responding emotions at two levels of abstraction. set of missing entries in Table 10.1. is those emo- Molar appraisals are global summaries of the tions that are not experienced, but only antici- cognitions that precede emotion. They are often pated. Lindsey, Yun, and Hill (2007) offer a case helpful shorthand, but only insofar as they are in point when they demonstrate that anticipated used with awareness that they summarize a more guilt may provide the motivation for helping precise and complex set of judgments. The col- unknown others (see also Wang, 2011). Thus, umn labeled molecular illustrates this point by Table 10.1. is intended as an illustration, one that offering a more detailed description of their is helpful primarily for its heuristic value. underlying components. Both are useful for The distance between events and appraisals understanding the antecedents of emotional also merits mention. Whereas it is tempting to arousal and, hence, for message design. However, assume that audience members generally grasp there are at least two respects in which Table 10.1. the intended meaning of messages, such a pre- is limited. sumption can be wildly off the mark. One classic For one, controversy regarding what should example can be seen in the literature on fear be counted as an emotion and what should not is appeals, which for many years simply assumed

Table 10.1 Emotions and Their Cognitive Antecedents

Appraisals

Emotions Molar Molecular

Fear Danger High probability of severe harm

Anger Social offense Unwarranted obstruction of goal

Sadness Loss Irrevocable failure to meet goal

Happiness Progress Acute movement toward goal

Hope Potential progress Change in the likelihood of goal achievement

Contentment Satisfaction Goal has been met

Source: Adapted from “The persuasive influence of emotion in cancer prevention and detection messages,” by J. P. Dillard and R. Nabi, 2006, Journal of Communication, 56, p. 123-139. Copyright 2006 by Wiley. Adapted with permission.

that message content that was considered scary by of compassion, pity, anger, and disgust (relative the message designer would be fear-inducing in to thematic framing). The same pattern, for the the message recipient. When attention was given same emotions, has been reported for persuasive to the gap between message content and emo- messages that make an explicit case against man- tional response, it became clear that threat datory minimum sentencing (Gross, 2008). appeals, rather than fear appeals, would be a more Major’s (2011) data show heightened anger and appropriate title for that research literature diminished happiness for thematically framed because that terminology avoids conflating mes- messages about lung cancer and obesity. More sage features with their expected effects (see complex findings appear in Gross and chapter 12 in this volume). As appraisal theory D’Ambrosio (2004), who describe framing effects makes plain, any given message may evoke dis- on responses to the 1992 Los Angeles riots that tinct emotional effects in different recipients. The are contingent on person factors such as political presentation of specific message content does not orientation (liberal vs. conservative). After read- guarantee evocation of particular emotions. ing a message that emphasized individual responsibility, conservatives experienced more anger Style than liberals. The reverse pattern held the two groups consumed a message that highlighted the Whereas message content refers to what the situational causes of the riots. These statistical message is about, style is how that content is interactions underscore the appraisal theory expressed. Classically, this has meant a focus on notion that emotional responses are the result of language. The advent of mediated messages, the interplay between stimulus and person. however, has broadened the meaning of style to Hence, while message features may produce main include ideas such as editing, point of view, effects, theory and research indicate that it would and pacing. be ill-advised to expect simplicity as a regular With regard to language, one of the most pro- occurrence. lific areas of inquiry has been that of framing. The term framing has also been used to dis- This is the idea that stylistic variations shape tinguish messages that contrast the potential understanding by selecting, then making salient, for favorable outcomes with those that high- certain aspects of a perceived reality to the exclu- light undesirable consequences. More specifi- sion of other elements (Entman, 1993). The cally, gain-framed messages emphasize the concept is broad enough that it now refers to benefits of compliance with the advocacy, several distinct lines of research. One of these which may include avoiding harm. Appeals that draws primarily from the tradition of journalism are loss-framed highlight the unattractive con- and mass communication. To wit, news coverage sequences of failure to comply or the potential is thematically framed when it emphasizes con- loss of attaining wanted outcomes. Although text broadly by presenting evidence that is collec- the genesis of this distinction lies in Kahneman tive, abstract, and general. Episodically framed and Tversky’s (1979) prospect theory, its’ appli- news stories focus on concrete events and specific cation to persuasion is appreciably different cases (note the parallels here with research on from their work on decision-making (O’Keefe narrative; see chapter 13 in this volume). & Jensen, 2006). Although most existing research on news Only a few studies have investigated the ability framing hews to a cognitive perspective, more of gain- and loss-framed appeals to evoke affect. recent work highlights emotional reactions. For Millar and Millar (2000) report that gain frames instance, Aarøe’s (2011) analysis of news cover- produced more favorable reactions than did loss age of a controversial Danish law revealed that frames. However, their bipolar measure of affect episodic framing produced stronger expressions rendered it impossible to tease out the effects of

(c) 2013 Sage Publications, Inc. All Rights Reserved. 156——PART II. Theories, Perspectives, and Traditions frame on emotions of positive and negative feel- story is sufficient to elevate perceptions of risk of ings independent of one another. Similarly, melanoma, farm failure, roller coasters, and tick- Schneider et al. (2001) assessed only negative borne diseases (the findings are summarized in affects, fear and anxiety, which precluded the Zillmann, 2006). Although many persuasive possibility of testing for an effect of gain-framing efforts target outcomes other than beliefs about on positive emotions. Studies that utilized a dis- risk, Zillmann’s data imply the potential for a crete emotions approach present a clearer pic- broader array of effects, including emotional ture. Shen and Dillard (2007) describe positive arousal. Indeed, Banerjee, Greene, and Yanovitzky’s effects of gain (vs. loss) framing on happiness (2011) research shows a dosage effect of images and negative effects on disgust, anger, fear, and on surprise such that before-and-after images of sadness. Cho and Boster (2008) found function- cocaine users provoked more surprise as a func- ally identical results for the same set of emotions. tion of the number of images. At first blush, this overall pattern of results might There are features of accompaniments that appear to be most parsimoniously interpreted by may themselves be considered stylistic. For exam- the valence-only model of affect. However, to do ple, the size of an image may influence emotional so would be to overlook important variation arousal. Research using clips of people engaged in with the categories of positive and negative, as potentially arousing activities, such as piloting a well as the unique effects of discrete emotions on fighter plane and driving in the Indianapolis persuasion. 500, reported that viewers were more excited and perceived more danger when watching the clips Accompaniments on larger versus small screens (Lombard, Reich, Grabe, Bracken, & Ditton, 2000). All persuasive messages contain or imply content. Even seemingly vacuous advertisements Research Designs and that present no more than a brand name on a the Elicitation of Affect blank page are understood to be promoting a particular product or service. There are other Generally speaking, there are two kinds of message features, such as imagery, that may research paradigms used to study persuasion and accompany message content even though they do affect. In the first, messages are designed with the not stand on their own. The choice of the term intention of provoking particular emotions. For accompaniments is not intended to imply trivial- example, fear/threat appeals may contain content ity. In fact, it may be that they are capable of that describes the likelihood and severity of some powerful effects on the elicitation of emotion. hazard if the message recipient fails to comply Nonetheless, they are mainly used to support or with the advocacy. Research designs that vary amplify the verbal portions of messages. message features for the purpose of eliciting Probably the most extensive theoretical analy- emotion may be referred to as message-induced sis of imagery is offered by Zillmann’s (1999) affect approaches. These designs and their corre- exemplification theory, which posits that indi- sponding theory focus on pathos as a rhetorical viduals form and maintain beliefs about phe- proof or, equivalently, the study of emotional nomena based on samplings of direct or indirect appeals. experience. Exemplars are defined as informa- Of course, even very brief messages vary on a tional units that are representative of some phe- multitude of dimensions including content, nomena. For instance, a news story about a par- structure, pacing, language, imagery, and so on. ticular smoker may exemplify the category of Such complexity greatly complicates the problem persons who smoke. A series of studies demon- of causal inference. Which message features are strates that the addition of a picture to a news responsible for evoking the intended emotions?

And, when other feelings are created, which mes- processing of a subsequent persuasion effect. For sage features brought them about? Some traction example, Chang (2011) examined how affect on the inference problem can be gained by induced by a magazine editorial influenced reac- including multiple messages of any given sort in tions to an advertisement that was presented the research. If they comprise a representative set, after the editorial. Because of its correspondence one should expect some consistency of effects with real-world media contexts, the use of mes- across the set for the message feature under sages to evoke feelings in message-irrelevant study. To the extent that the other message fea- affect designs enhances external validity. How- tures are really unimportant, they should show ever, they also present challenges to internal little or no consistency of effect. validity as a result of the complexity of the affect- A different research paradigm separates the inducing stimulus. message from the affect induction altogether (e.g., Anghelcev & Sar, 2011). One common instantiation of this approach is the Life Event The Effects of Affect Inventory Task (Schwarz & Clore, 1983), an experimental procedure in which participants Evidence of Effects are asked to recall and write briefly about a past event that caused them to experience a particu- For at least two millennia, students of suasory lar emotion. Subsequently, participants are pre- discourse have embraced the idea that emotions sented with a persuasive message. The phrase can persuade (Aristotle, 2007). Evidence of this message-irrelevant affect has been used to point can be seen in the analysis of emotional describe these research designs in order to appeals by early rhetoricians, including Aristotle, emphasize that the induced emotions have no and can be traced through history to the present logical connection to the message. Indeed, one time. But, the more nuanced question—how key feature of many investigations is an effort should the emotion-persuasion link be analyzed to obfuscate in the minds of the participants and understood?—is still under debate. There are any perceived linkage between the affect induc- several options to consider. From the perspective tion and the persuasive message. This is usually of the dimensional valence model, the affects achieved by informing participants that they vary along a single dimension (positive vs. nega- are taking part in two unrelated studies, one tive) and should be tested accordingly. That is, that deals with emotional memories and one researchers should measure affective responses that focuses on something else, such as message on a series of semantic differential scales, then evaluation. correlate the sum of those scales with the persua- From the standpoint of causal inference, pro- sion outcome of interest. The categorical valence cedures like the Life Event Inventory Task are model directs researchers to similar options, attractive in that specific emotions clearly occur except that positive and negative affects are mea- prior to message processing. One can then pre- sured independently of one another and treated sumably draw uncluttered inferences concerning as two separate predictors. The pleasure-arousal how emotions influence message processing. model, also a dimensional model, ordains the use This is an interesting question, but it is different of two sets of bipolar scales and, thus, two con- from that of how message-induced emotions tinuous predictors: Pleasant (vs. unpleasant) and might bring about persuasion. And, of course, arousal (vs. subdued). The discrete emotions many communication researchers are less inter- position insists that each emotion is defined by a ested in supposedly pure manipulations of emo- different function and different response pat- tion and more concerned with how the affect terns. Thus, emotions must be measured indi- that is induced by one message might shape vidually, and their effects on persuasion must be

(c) 2013 Sage Publications, Inc. All Rights Reserved. 158——PART II. Theories, Perspectives, and Traditions evaluated using methods that allow for unique and, in the same analysis, a third negative emo- effects to manifest themselves. tion may show no significant effect at all. In To date, there are many existing studies that further illustration of this same general point, test the relationship between affect and persua- Nan (2009, p. 437) presents results showing that sion. Only a subset has produced evidence that anger toward public service announcements is can be used to differentiate the various models of counter-persuasive, while guilt seems to gener- affect. Studies that assess emotions on only bipo- ate agreement with the message. These findings lar scales must be excluded because the measure- too are at odds with any model of affect that ment procedures assume, rather than test, the anticipates uniform effects within valence- structure of affect. If affects are measured on a defined categories, that is, the bipolar valence positive versus negative scale, researchers cannot model, the categorical valence model, and the then unscramble the differences between sadness pleasure dimension of the PA model. and anger or hope and happiness because those 2. There is variation in the effects of positive distinctions are lost at the moment of data col- emotions. Participants in the Kim, Park, and lection. The same problem extends to pleasure- Schwarz (2010) were asked to list three life arousal models, except that there are two events that were either peaceful or exciting. dimensions rather than one. In fact, the only data Then, under the guise of a separate study, data that can be used to empirically evaluate the were gathered on their evaluations of a message worth of the different models come from investi- that promoted a vacation to Japan. There were gations that utilize discrete emotion procedures. two versions of the advertisement. One of them It is possible to aggregate specific emotions into emphasized the possibilities for adventure and larger categories, such as pleasant and unpleas- excitement via exposure to anime/manga, elec- ant, but impossible work in the other direction, tronics, and Japanese drumming, while the other that is, to disassemble global affective judgments promised a serene and tranquil experience into discrete elements. focused on hot springs, tea ceremonies, and There are four types of evidence in the per- religious traditions. Responses to the question suasion literature that distinguish the discrete “I would like to take a vacation in this country” emotions model from the alternatives. Each is were most favorable when message content illustrated next using investigations from differ- matched the pre-existing emotional state. In ent contexts: other words, the effects of these two positive emotions were reversed as a function of message 1. There is variation in the effects of negative content. Nan (2009, Table 2) also provides evi- emotions. Although public opinion remains dence of the independent, and generally oppos- divided about the wisdom of a Western-led war ing, effects of two other positive emotions. In with Iraq, there is no question that a sizeable her data, happiness is associated with greater portion of the U.S. population is critical of their persuasion, and contentment with less. government’s action. In a survey of undergrad- uate U.S. citizens, Iyer, Schmader, and Lickel 3. Discrete emotion models exhibit better predic- (2007) examined the associations of anger, tive power than the alternatives. In one of the few guilt, and shame with intentions for political direct comparisons of affect models, Dillard and action. They found that both anger and shame Peck (2001) examined emotional reactions to predicted intentions to advocate withdrawal public service announcements on a variety of pro- from Iraq, while the association with guilt was social topics, such as exercise, charitable giving, nonsignificant. The results are instructive in that and avoiding contraction of HIV. They conducted they illustrate how two negative emotions can two series of regression analyses, each of which exert independent effects on a single intention, used perceived effectiveness of the messages as the

criterion variable. In the discrete emotions analy- rather, necessary for understanding the effects of ses, anger, fear, sadness, guilt, happiness, and con- emotion on persuasion (e.g., Huddy, Feldman, tentment were used as separate predictors. For the Taber, & Lahav, 2005; Lerner, Gonzalez, Small, & valence analyses, anger, fear, sadness, and guilt Fischhoff, 2003; Lerner & Keltner, 2000). were combined to form a single, negative-affect predictor; happiness and contentment were com- Mechanisms of Effect bined to form a single, positive affect variable. A ratio of the R2 change for each set of equations The appraisal model of persuasion (Figure 10.1.) permitted a judgment of the relative predictor depicts movement from emotion to persuasion power of each affect model. The discrete emotions outcome with a simple, single arrow. However, analysis was superior to the valence analysis by a because emotions are complex states that impli- factor of two. cate a host of perceptual, cognitive, physiological, 4. Different emotions produce different persua- and motivational systems, they might influence sion outcomes. Appraisal theory presupposes that the persuasion process through multiple means emotions are means of solving functionally dis- (Dillard & Nabi, 2006). tinct types of problems. From that premise, it Perhaps the simplest claim regarding the seems reasonable to infer that different emotions mechanism is that emotions can have a direct would produce different persuasive outcomes. effect on persuasion outcomes (Dillard & Nabi, An illustration of these variations is offered by 2006). This idea traces back to the assumption Brader (2005), in an experiment conducted in that stimulation of an action tendency is a defin- the context of a Democratic primary race for ing feature of any emotional response. Each governor. Two messages were designed, both of emotion has some implications for action that which manipulated various nonverbal cues to can be understood as approach/engagement or create either enthusiasm or fear. More concretely, avoidance/withdrawal. These motivational ten- one message paired uplifting music and images dencies are guides to behavior with direct impli- of children with message content that described cations for attitudes and intentions. However, conditions that were “good and getting better” because emotions are evolutionarily designed to (p. 392). The other appeal, which argued that solve specific problems, the action tendency asso- things were “bad and getting worse,” (p. 392) was ciated with each emotion is a more specific accompanied by tense music and pictures of vio- variation on one, the other, or both of those two lence. In a crossed design, both messages were broad themes. Table 10.2. provides a summary attributed to both candidates. The data revealed of emotions and their predominant action ten- that the enthusiasm message solidified pre-existing dencies. The qualifier predominant is important voter preferences and dramatically increased in this application because action tendencies intention to vote (a 29% increase). In contrast, vary as a function of context. For example, the the fear message was more effective at changing predominant action tendency for fear is acquies- voter preferences, but had no significant influ- cence. But, under some circumstances, fear ence on intention to vote. motivates aggression. The table also attempts to elaborate on the point that action tendencies can The research cited in preceding paragraphs is be analyzed at both the general and specific lev- illustrative: Much more evidence exists. The els. At the general level, anger and happiness are research that appears in the literature prior to both approach emotions in that they prompt 2001 is summarized by Dillard and Meijnders engagement with the emotion-inducing stimu- (2002). And, accrual of data since that time has lus. However, at the specific level, they are quite only supported the conclusion that a discrete different from one another. The logic of the emotions perspective is not merely preferable, but direct effects mechanism predicts that action

Table 10.2 Emotions and Their Action Tendencies

Emotions Generala Specific

Fear Avoid Acquiesce

Anger Approach Attack

Sadness Avoid Review plan/convalesce

Happiness Approach Bask/bond

Hope Approach Renew effort

Contentment Avoid Immobility a. These should be understood as approach/engage and avoidance/withdrawal. tendency effects will be moderated by various In parallel, inducing anger or happiness—both aspects of the advocacy or the context in which of which are approach emotions—activates the the message is presented. behavioral approach system and causes gain- Another class of mechanisms reflects height- framed messages to be evaluated more favorably ened sensitivity to appraisal-relevant message (Yan, Dillard, Shen, in press). features. The most commonly observed type of Emotions can also influence the way in which effect is matching, in which emotion-consistent messages are processed. Specifically, emotions of beliefs are amplified by the evocation of the cor- different sorts might propel individuals toward or responding emotion. Just as perceived risk away from a message topic, a phenomenon that induces fear, fear increases estimates of risk. Con- Nabi’s (1999) cognitive-functional model (CFM) versely, prior instigation of anger attenuates per- characterizes as motivated attention. In the same ceived risk (Lerner et al., 2003). Kim et al. (2010) vein, message consumers may try to regulate the describe matching effects for pre-message emo- emotion via motivated processing (Nabi, 1999). tions and subsequent issue judgments that are Based on the type of emotion experienced, moti- mediated by expectancies, a specific form of vated attention sets a baseline attention level that belief. Turner (2007) reports findings in which will either impede (for avoidance emotions, like matching effects for anger carry over to politi- fear) or facilitate (for approach emotions, like cally oriented action tendencies, including sign- anger) subsequent information processing. ing a petition, talking to friends, and organizing Expectation of reassurance from the message an event. might then further shape style of processing. For There is also evidence of emotional induc- example, given anger’s nature, the CFM predicts tions that can activate the behavioral approach that it is likely to promote closer information and avoidance systems, which, in turn, render processing. Fear is likely to promote more mes- audience members sensitive to noncontent message scrutiny only when reassurance cues are sage features. Using the Life Event Inventory unavailable. Huddy and Gunnthorsdottir (2000) procedure, Yan, Dillard, and Shen (2010) dem- suggest that message-induced emotion effects onstrate a variation on the sensitization phe- may be limited to message recipients who are nomenon such that when sadness or fear are used knowledgeable about and involved with the issue. to stimulate the behavioral inhibition system, Although the data on matching effects are loss framed messages become more persuasive. appealing, Agrawal and Duhacheck (2010) suggest

(c) 2013 Sage Publications, Inc. All Rights Reserved. Chapter 10. Affect and Persuasion——161 that they may be overridden, at least in some cases, via judgments of good or bad, favorable or unfa- by motivated processing. In their research, shame vorable, and like or dislike, all of which capture or guilt was evoked by describing others either the cognitive aspect of evaluations. In contrast, observing or suffering from the negative conse- measures that index responses such as anxious, quences of binge drinking. The others-as-observers pleasant, tense, happy, and sad may be appropri- perspective produced shame because message ate for the assessment of emotion, but they are recipients were presented as behaving in way that invalid indicators of attitude. Some investiga- yielded unfavorable judgments of their friends tions continue to intermingle measures of affect and family. In contrast, when friends and family and attitude, thereby diminishing the interpret- were described as bearing the costs of the research ability of their results. participants irresponsible behavior, the participants reported feeling guilt. Antidrinking mes- The Affect Heuristic sages that were matched on induced emotion, yielded mismatched persuasive effects presumably Heuristics are short-cut decision-making rules because participants were motivated to counter- that come into play when a more detailed analy- argue against appeals that further threatened their sis of persuasive circumstance is unwarranted or sense of self. impossible. For instance, when faced with a non- functional computer, many people defer to the recommendation of their IT specialist rather Other Themes in than spend time themselves learning the technol- ogy of personal computers, disassembling their the Research Literature machine, then locating and fixing the problem Affect and Attitudes themselves. Pertinent to this chapter is the claim that individuals often rely on an affect heuristic: The concept of attitude is inextricably bound A feeling that captures the positive or negative up with that of evaluation. And, of course, so is aspect of a stimulus (Slovic & Peters, 2006). This affect. The fact that the two concepts both draw definition seems to carry on the tradition of con- on valence as an underlying concept has meant flating affect and attitude. However, the termi- that, at certain points in history of social science, nology and theorizing are more contemporary. researchers made little or no distinction between Experiments conducted on the affect heuristic them. For example, the original version of the manipulate variables, such as time pressure, that theory of reasoned action viewed affect and atti- are held to govern the activation of other cogni- tude as essentially synonymous (Fishbein & tive heuristics (Slovic, Finucane, Peters, & Mac- Ajzen, 1975). However, in subsequent years, the- Gregor, 2002). Indeed, the results show parallel orists developed meaningful grounds on which effects. However, the affect heuristic is defined in to differentiate the two. One of the most compel- terms of feelings rather than cognition. This ling expositions is offered by Zanna and Rempel bears some similarity to the notion of signal (1988) who show that the utility of treating value, that is, the idea that subjective experience (1) attitude as a summary cognitive judgment is the means by which conscious awareness is regarding the goodness-badness of a mental informed of the change in the physiological, object and (2) affect as one or more feeling states expressive, cognitive, and, motivational domains that may influence attitude. Their position aligns (described in more detail below). But, work on perfectly with the appraisal model of persuasion the affect heuristic seems to reject the conception given in Figure 10.1., and it has clear implica- of specific emotions. Despite its application to tions for the measurement of both attitude and risk assessment, which is normally associated affect. To wit, attitude is appropriately assessed with fear, research on the affect heuristic makes

(c) 2013 Sage Publications, Inc. All Rights Reserved. 162——PART II. Theories, Perspectives, and Traditions only a distinction between positive and negative assessment of the intrinsic pleasantness of stim- affect. In the final analysis, the notion of an affect uli (Scherer, 1984). And, as theories such rea- heuristic has been successful at stimulating soned action tell us, even many initially complex research. However, it seems to bring together decisions ultimately reduce to good versus bad concepts from appraisal theory, cognitive theory, judgments (chapter 8 in this volume). Jointly, and the valence approach in ways that are funda- these facts of life may bring valence judgments to mentally incompatible with one another. Thus, it the forefront of human consciousness, where is difficult to know how to locate the research in they are as likely to bias the views of social scien- a more general framework. tists as they are lay persons. Another reason for the endurance of the valence perspective may lie with the methods Issues for Future Research that are commonly employed in studies of affect. Atheoretical applications of exploratory factor The Regrettable Persistence of analysis to emotion data typically produce one Valence Models positive emotion factor and one negative emotion factor. To generate an example, undergradu- Taken together, the reasons to prefer a discrete ate students at Penn State were asked to react to a emotions perspective over the alternative con- one-page print advertisement regarding partici- ceptions are quite strong (Nabi, 2010). It seems pation in an education program that would clear that continued dependence on valence render them eligible for a free tablet personal models is a mistake. Nonetheless, the valence computer (Seo & Dillard, 2012). The message model still appears in the literature with remark- recipients provided information regarding their able regularity—a circumstance that seems to emotional reactions as well as estimates of their call for explanation. intention to take part in the program. Next, the One way of accounting for this curiosity is to emotion data were submitted to a principle axis recognize that valence judgments are salient factor analysis followed by oblique rotation. throughout life. They are learned early, then Table 10.3. gives the loadings for the pattern practiced regularly. Rudimentary processes such matrix, which are reasonably clear and perfectly as the startle response are quickly followed by an typical: There are two factors, one composed of

Table 10.3 Pattern Matrix of a Principal Axis Analysis of Emotion Data

Emotion Factor 1 Factor 2

Happiness .63 -.12

Contentment .73 .12

Anger -.18 .73

Sadness .01 .77

Fear .08 .46

Note: N = 455. Emotion judgments were made on 0 = None of this feeling to 4 = A great deal of this feeling using multiple items for each affect: Happiness (happy, excited, eager, and cheerful; α = .90), contentment (content, calm, satisfied, and tranquil; α = .71), anger (angry, irritated, annoyed, and aggravated; α = .88), sadness (sad, dreary, and dismal; α = .50), and fear (fearful, afraid, and scared; α = .85).

positive emotions and the other consisting of of use. But, what that might be is not evident. It negative emotions. This simple, intuitive solu- would be far more useful to reject this simplifica- tion looks as if it conforms perfectly with the tion of emotional influence and get on with the valence model of affect. In line with the factor business of refining theory and conducting research analytic findings, we created two variables to that is equal in nuance to human experience. represent positive and negative affect (i.e., happiness + contentment and anger + sadness + fear). These two variables were then used as predictors The Value of Research on Style of intention to participate in the education pro- and Accompaniments gram for personal computer (sample item: If I can, I will take part in the Free Tablet PC Pro- Most of the time, content is king: Genuine gram). The regression analysis yielded standard- differences in message content evoke different ized coefficients of .40, p < .05, for positive affect cognitive and emotional responses, as well as and -.23, p < .05 for negative affect. Anyone varying degrees of persuasion (although, of looking at these results might conclude that course, these differences must be perceived). It is positive affect promotes persuasion, while nega- not the least surprising that many applied per- tive affect has an inhibitory effect. suasion campaigns undertake formative research But, there is a fly in the ointment. The purpose with the goal of identifying substantive argu- of factor analysis is to group things together that ments to support the goal of changing beliefs, are similar. If the factor analytic findings consti- attitudes, and behavior (chapter 17 in this vol- tute a useful description of reality, then the ele- ume). There is little reason to question the desir- ments of each factor must show similar effects. ability of this aim. However, there is at least one To test for this pattern, a second regression analy- significant barrier to achieving it. sis was conducted in which each of the emotions Argument novelty has been posited to a nec- was used as separate predictors of intention. The essary condition for persuasion (Morley & results showed that the emotions that made up the Walker, 1987). Minimally, new arguments carry positive emotion factor did not show consistent more weight than those that have been heard relationships with behavioral intention. Happi- before (Prochaska, Johnson, & Lee, 1998, p. 68). ness was directly associated with intention In some persuasive domains, argument novelty (β = .62, p < .05), but contentment showed the may be an infinite resource. Political campaigns opposite relationship (β = -.20, p < .05). Thus, the come quickly to mind: They are endlessly cre- positive emotions did not exhibit similar effects. ative because they are so frequently unbeholden Attention to the negative emotions yielded a to facts. In other areas, practical or ethical con- complementary conclusion: Anger was signifi- siderations constrain the available pool of novel cantly and inversely related to intention (β = -.14, arguments. Health-related persuasion surely p < .05), but the weights for sadness and fear were ought to be limited to the facts as they are functionally zero (β = -.03 and β = -.04, both non- understood by the community of science. This significant). Hence, the negative emotions did not means that the claim that smoking causes cancer exhibit similar effects. Overall then, the explor- has been repeated so many times that it may be atory factor analysis encourages grouping the data virtually impossible to find an audience for in a way that produces inaccurate conclusions and whom the claim is novel. Hence, persuasion obscures the true nature of the data. In contrast, practitioners who aim to reduce risky health allowing theory to direct the analysis provides a behaviors, diminish crime, or encourage physi- justifiably variegated approach to the analysis. cal activity may have few alternatives other than Given their ubiquity and endurance, it seems style and accompaniments in the production of that valence models must be capturing something effective campaigns.

This is not to suggest that efforts to devise monitoring their feelings and recognizing the compelling arguments should be abandoned. causes of those affects. Communication Rather, it implies that, if campaigns are to maxi- researchers have the opportunity to create the- mize their impact, they must consider message ory that links messages features with emotional features in addition to argumentative content. To response and emotional response with impor- the extent that those features mold parts of the tant persuasion outcomes. The development of persuasive process, it is likely that their influence a scientific account of emotion and persuasion derives from emotional arousal. The graphic is a worthy aim in its own right. To the extent imagery that appears on cigarette packages in that it enables audience members to better some non-U.S. countries, is an opportunity to understand and manage the relationship understand the interplay of emotion and per- between persuasive appeals and emotion suasion, as well as a possible means of improv- impact, it furthers the goals of creating more ing public health. In this same vein, the well- effective citizens, more critical consumers, and established tendency for human facial expression more successful decision makers. to create emotional contagion has not generated a body of persuasion research that is commensu- rate with the power of the concept. Anthropo- References morphic inclinations that produce imposition of Aarøe, L. (2011). Investigating frame strength: The human faces on nonhuman objects offers case of episodic and thematic frames. Political intriguing possibilities (Landwehr, McGill, & Communication, 28, 207–226. Hermann, 2011). Not to be overlooked are the Agrawal, N., & Duhachek, A. (2010). Emotional com- more subtle aspects of the visual experience, such patibility and the effectiveness of anti-drinking as color and script style (Wakefield, Germain, & messages: A defensive processing perspective on Durkin, 2008) or camera angle (Giessner, Ryan, shame and guilt. Journal of Marketing Research, Schubert, & van Quaquebeke, 2011). 47, 263–273. In short, efforts to enhance our understanding Anghelcev, G., & Sar, S. (2011). The influence of of stylistic issues are needed. Such research can pre-existing audience mood on message rele- be intellectually provocative and theoretically vance and the effectiveness of health PSAs: stimulating. In some applied settings, it may be Differential effects by message type. Journalism & Mass Communication Quarterly, 88, 481–501. the only available means of enhancing persuasive Aristotle. (2007). On rhetoric: A theory of civic discourse effectiveness. (George A. Kennedy, Trans., 2nd ed.). New York, NY: Oxford University. Banerjee, S. C., Greene, K., & Yanovitzky, I. (2011). Conclusion Sensation seeking and dosage effect: An explora- tion of the role of surprise in anti-cocaine mes- The artist Vincent van Gogh said “Let’s not sages. Journal of Substance Abuse, 16, 1–13. forget that the little emotions are the great cap- Brader, T. (2005). Striking a responsive chord: How tains of our lives and we obey them without political ads motivate and persuade voters by realizing it.” He was certainly correct in identi- appealing to emotions. American Journal of Polit- ical Science, 49, 388–405. fying emotion as a powerful determinant of Chang, C. (2011). The influence of editorial liking and human action. Indeed, much of this chapter editorial-induced affect on evaluations of subse- has reviewed research that accords perfectly quent ads. Journal of Advertising, 40, 45–58. with van Gogh’s assertion. Yet, on his last Cho, H., & Boster, F. J. (2008). Effects of gain versus point—that “we obey them without realizing loss frame antidrug ads on adolescents. Journal of it”—van Gogh may have overstepped. Message Communication, 58, 428–446. consumers do have the ability to defuse emotion- Dillard, J. P., & Meijnders, A. (2002). Persuasion induced judgments and action tendencies by and the structure of affect. In J. P. Dillard &

M. W. Pfau (Eds.), The persuasion handbook: Lerner, J. S., Gonzalez, R. M., Small, D. A., & Fischhoff, Developments in theory and practice (pp. 309–328). B. (2003). Effects of fear and anger on perceived Thousand Oaks, CA: Sage. risks of terrorism: A national field experiment. Dillard, J. P., & Nabi, R. (2006). The persuasive influ- Psychological Science, 14, 144–150. ence of emotion in cancer prevention and detec- Lerner, J. S., & Keltner, D. (2000). Beyond valence: tion messages. Journal of Communication, 56, Toward a model of emotion-specific influences s123–s139. on judgment and choice. Cognition and Emotion, Dillard, J. P., & Peck, E. (2001). Persuasion and the 14, 473–493. structure of affect: Dual systems and discrete Lerner, J. S., & Tiedens, L. Z. (2006). Portrait of the emotions as complementary models. Human angry decision maker: How appraisal tendencies Communication Research, 27, 38–68. shape anger’s influence on cognition. Journal of Entman, R. M. (1993). Framing: Toward clarification Behavioral Decision Making, 19, 115–137. of a fractured paradigm. Journal of Communica- Lindsey, L. L. M., Yun, K, A., & Hill, J. B. (2007). tion, 43, 51–58. Anticipated guilt as motivation to help unknown Fishbein, M., & Ajzen, I. (1975). Belief, attitude, inten- others: An examination of empathy as a modera- tion, and behavior. Reading, MA: Addison-Wesley. tor. Communication Research, 34, 468–480. Giessner, S. R., Ryan, M. K., Schubert, T. W., & van Lombard, M., Reich, R. D., Grabe, M. E., Bracken, C. C., Quaquebeke, N. (2011). The power of pictures: & Ditton, T. B. (2000). Presence and television: Vertical angles in power pictures. Media Psychol- The role of screen size. Human Communication ogy, 14, 442–464. Research, 26, 75–98. Gross, K. (2008). Framing persuasive appeals: Episodic Major, L. H. (2011). The mediating role of emotions in and thematic framing, emotional response, and the relationship between frames and attribution policy opinion. Political Psychology, 29, 169–192. of responsibility for health problems. Journalism Gross, K., & D’Ambrosio, L. (2004). Framing emotional and Mass Communication Quarterly, 88, 502–522. response. Political Psychology, 25, 1–29. McCauley, C., & Moskalenko, S. (2008). Mechanisms of Huddy, L., Feldman, S., Taber, C., & Lahav, G. (2005). political radicalization: Pathways toward terror- Threat, anxiety, and support of anti-terrorism ism. Terrorism and Political Violence, 20, 415–433. policies. American Journal of Political Science, 49, Millar, M., & Millar, K. (2000). Promoting safe driving 610–625. behavior: The influence of message framing and Huddy, L., & Gunnthorsdottir, A. H. (2000). The per- issue involvement. Journal of Applied Social Psy- suasive effects of emotive visual imagery: Superfi- chology, 30, 853–866. cial manipulation or the product of passionate Miller, G. A. (2003). The cognitive revolution: A his- reason? Political Psychology, 21, 745–778. torical perspective. Trends in Cognitive Sciences, 7, Iyer, A., Schmader, T., & Lickel, B. (2007). Why indi- 141–144 viduals protest the perceived transgressions of Morley, D. D., & Walker, K. (1987). The role of impor- their country: The role of anger, shame, and guilt. tance, novelty, and plausibility in producing belief Personality and Social Psychology Bulletin, 33, change. Communication Monographs, 54, 436–442. 572–587. Nabi, R. L. (1999). A cognitive-functional model for Kahneman, D., & Tversky, A. (1979). Prospect theory: the effects of discrete negative emotions on infor- An analysis of decision under risk. Econometrica, mation processing, attitude change, and recall. 47, 263–291. Communication Theory, 9, 292–320. Kim, H., Park, K., & Schwarz, N. (2010). Will this trip Nabi, R. L. (2010). The case for emphasizing discrete really be exciting? The role of incidental emotions emotions in communication research. Communi- in product evaluation. Journal of Consumer cation Monographs, 77, 153–159. Research, 36, 983–991. Nan, X. (2009). Emotional responses to televised PSAs Landwehr, J. R., McGill, A. L., & Hermann, A. (2011). and their influence on persuasion: An investiga- It’s got the look: The effect of friendly and aggres- tion of the moderating role of faith in intuition. sive “facial” expressions on product liking and Communication Studies, 5, 426–442. sales. Journal of Marketing, 75, 132–146. O’Keefe, D. J., & Jensen, J. D. (2006). The advantages Lazarus, R. S. (1991). Emotion and adaptation. New York, of compliance or the disadvantages of noncom- NY: Oxford University Press. pliance? A meta-analytic review of the relative

persuasive effectiveness of gain-framed and Slovic, P., Finucane, M. L., Peters, E., & MacGregor, D. G. loss-framed messages. Communication Yearbook, (2002). The affect heuristic. In T. Gilovich, D. Griffin, 30, 1–43. & D. Kahneman (Eds.), Heuristics and biases: The Phan, K. L., Wager, T., Taylor, S. F., & Liberzon, psychology of intuitive judgment (pp. 397–420). I. (2002). Functional neuroanatomy of emotion: New York, NY: Cambridge University Press. A meta-analysis of emotion activation studies in Slovic, P., & Peters, E. (2006). Risk perception and PET and fMRI. NeuroImage, 16, 331–348. affect. Current Directions in Psychogical Science, Prochaska, J. O., Johnson, S., & Lee, P. (1998). The 15, 322–325. transtheoretical model of behavior change. In Tooby, J., & Cosmides, L. (2008). The evolutionary S. A. Schumaker, E. B. Schron, J. K. Ockene, & psychology of the emotions and their relation- W. L. McBee (Eds.), The handbook of health ship to internal regulatory variables. In M. Lewis, behavior change (pp. 59–84), 2nd ed. New York, J. M. Haviland-Jones, & L. F. Barrett (Eds.), NY: Springer. Handbook of emotions (3rd ed., pp. 114–137.) Roseman, I. J. (2001). A model of appraisal in the emo- New York, NY: Guilford. tion system: Integrating theory, research, and Turner, M. M. (2007). Using emotion in risk commu- applications. In K. R. Scherer, A. Schorr, & nication: The Anger Activism Model. Public Rela- T. Johnstone (Eds.), Appraisal processes in emotions Review, 33, 114–119. tion: Theory, methods, research (pp. 68–91). Wakefield, M. A., Germain, D., & Durkin, S. J. (2008). New York, NY: Oxford University Press. How does increasingly plainer cigarette packag- Russell, J. A., & Feldman Barrett, L. (1999). Core affect, ing influence adult smokers’ perceptions about prototypical emotional episodes, and other things brand preference? An experimental study. Tobacco called emotion: Dissecting the elephant. Journal of Control, 17, 416–421. Personality and Social Psychology, 76, 805–819. Wang, X. (2011). The role of anticpated guilt in inten- Scherer, K. R. (1984). On the nature and function of tions to register as organ donors and to discuss emotion: A component process approach. In organ donation with family. Health Communica- K. R. Scherer, & P. Ekman (Eds.), Approaches to tion, 26, 683–690. emotion (pp. 293–317). Hillsdale, NJ: Erlbaum. Watson, D., & Clark, L. A. (1994). The PANAS-X: Scherer, K. R., Schorr, A., & Johnstone, T. (Eds.). Manual for the positive and negative affect sched- (2001). Appraisal processes in emotion: Theory, ule-Expanded Form. Iowa City: University of Iowa. methods, research. New York, NY: Oxford University Yan, C., Dillard, J. P., & Shen, F. (2010). The effects of Press. mood, message framing, and behavioral advocacy Schneider, T. R., Salovey, P., Apanovitch, A. M., Pizarro, on persuasion. Journal of Communication, 60, J., McCarthy, D., Zullo, J., et al. (2001). The effects 344–363. of message framing and ethnic targeting on Yan, C., Dillard, J. P., & Shen, F. (in press). Emotion, mammography use among low-income women. motivation, and the persuasive effects of message Health Psychology, 20, 256–266. framing. Journal of Communication. Schwarz, N., & Clore, G. L. (1983). Mood, misattribu- Zanna, M. P., & Rempel, J. K. (1988). Attitudes: tion, and judgments of well-being: Informative A new look at an old concept. In D. Bar-Tal & and directive functions of affective states. Journal A. W. Kruglanski (Eds.), The social psychology of Personality and Social Psychology, 45, 513–523. of knowledge (pp. 315–334). Cambridge, UK: Seo, K., & Dillard, J. P. (2012). (Unpublished data). Cambridge University Press. University Park: Pennsylvania State University. Zillmann, D. (1999). Exemplification theory: Judging Shen, L., & Dillard, J. P. (2007). The influence of the whole by some of its parts. Media Psychology, behavior inhibition/approach systems and mes- 1, 69–94. sage framing on the processing of persuasive Zillmann, D. (2006). Exemplification effects in health messages. Communication Research, 34, the promotion of safety and health. Journal of 433–467. Communication, 56, S221–S237.