MASTER’S THESIS
The influence of discrete emotions in a public health message: a quantitative study on “me” versus “them”
Michelle Penning Student ID: 10858768
Universiteit van Amsterdam Graduate School of Communication Master’s programme Communication Science
Supervisor: Stephanie Welten
Abstract
The harmful psychological and physical health effects of excessive noise show an obvious necessity for the development and implementation of specific intervention strategies. Emotional appeals have been used in health campaigns, although the focus has been on valence and fear. However, different emotions can have different effects on persuasive outcomes. Moreover, in order to be better able to change an individual’s behavior, it is necessary to learn how we can bypass the third-person effect, which is currently associated with negative appeals. In the current study it is tested to what extent positive versus negative and basic versus self-conscious emotions differ in their effect on persuasion outcomes and the third-person effect. Participants were presented with an advertisement on wearing hearing protection including a fear, shame, happiness or pride appeal. Participants in the fear condition showed a higher intention to wear hearing protection. Further, participants in a basic emotion condition showed a greater third-person effect than people in a self-conscious emotion condition. The results do not only extend current knowledge, but also give increased ability to create applicable interventions and show that it is too simplistic to think that
intervention strategies should be based on only one emotion.
Keywords: discrete emotions, valence, type, third-person effect, persuasion outcomes,
The influence of discrete emotions in a public health message: a quantitative study on “me” versus “them”
According to the World Health Organization 1.1 billion youngsters are at a risk of permanent hearing loss due to listening to loud music (“1.1 miljard jongeren,” 2015; Vogel, Brug, van der Ploeg, & Raat, 2007). The GGD estimates that 450.000 youngsters in the Netherlands between 15 and 25 have a hearing loss of 10 decibels or more. Every year, 22.000 more youngsters are added to this number (“Gehoorschade door,” nd.). It is apparent that this noise induced hearing loss crisis requires urgent attention. Research has shown that youngsters see hearing loss as a low priority, even though there are several negative
consequences ascribed to the problem, such as a poorer quality of life – related to isolation, reduced social activity, increased symptoms of depression and limited career options – and a decreased ability to monitor the environment for warning signs and other sounds (Arlinger, 2003; Chung, Des Roches, Meunier, & Eavy, 2005; Vogel et al., 2007). A plausible reason for this low priority is that the perceived seriousness of damage to the hearing and the perceived benefits of wearing hearing protection may be outweighed by the cultural pressures to
conform to stylistic norms and youthful images of attractiveness and healthy bodies (Crandell, Mills, & Gauthier, 2004). However, the harmful psychological and physical health effects of excessive noise show an obvious necessity for development and implementation of specific intervention strategies.
Already in the 4th century B.C.E., Aristotle argued that it is necessary for an orator to include emotional appeals in persuasive messages, since manipulations of emotions could sway opinions most effectively, as emotions have specific causes and effects (Miceli, Rosis, & Poggi, 2006; O’Shaugnessy & O’Shaughnessy, 2004). However, there has been an ongoing discussion whether rational or emotional appeals are in fact more effective (O’Shaugnessy & O’Shaugnessy, 2004). So far no consensus has been reached, although researchers do suggest
that emotional appeals are more effective when it comes to remembering and catching and holding attention (Angie et al., 2011; Asbeek, Brusse, Neijens, & Smit, 2010; Stephenson et al., 2005). Moreover, research has shown that emotional responses might not only be an important consequence of social and moral behavior, but they also might form these behaviors (Pham, 2007; Agrawal, Menon, & Aaker, 2007). Therefore, the current study aims to learn more about the influence of discrete emotions in a public health message on wearing hearing protection.
Several other theoretical reasons can be identified as well. Emotional appeals have been widely used in persuasive messages for a long period of time, however, it is only for a couple of years now that it has become the subject of scientific research (Angie, Connelly, Waples, & Kligyte, 2011; Pham, 2007). And that knowledge is, to say the least, limited. It is largely defined by research on fear appeals (Agrawal et al., 2007; Nabi, 2002) and by studies that define affect as a unidimensional and bipolar construct (Angie et al., 2011; Stephenson et al., 2005). However, an extensive body of literature has yielded complex and often
inconsistent results regarding the extent to which fear appeals change attitude and behaviors. Still though, many public health campaigns use fear as a persuasive strategy, as such to change attitudes, intentions and behaviors (Lewis, Watson, White, & Tay, 2007). Recently, research has begun to examine the role of specifically defined emotional states, as it is shown that different emotions can have different effects on persuasive outcomes (Agrawal et al., 2007; Griskevicius, Shiota, & Neufeld, 2010; Nabi, 2010). Not only is a difference in outcomes of negative versus positive emotions now clear, but research has also shown that consumers who experience two emotional states that are negatively valenced process
information differently (Agrawal & Duhachek, 2010). This means that it is of importance to explore whether health messages might be more effective when appealing to specific
self-efficacy: a crucial component in the process of changing attitude and behaviors (Lewis et al., 2007).
It is critical to compare emotions in two ways: valence and type. First, comparing emotions in terms of type is of importance. Emotions are divided in literature in two types: basic emotions, such as sadness and joy, and self-conscious emotions, such as shame and pride. Self-conscious emotions differ from basic emotions in that the former require self-awareness and self-representation in order to be experienced (Lewis, 2011). Self-conscious emotions play a central role in motivating and regulating thoughts, feelings and behaviors. However, they have received relatively little attention, since they are more difficult to elicit than basic emotions (Tracy & Robins, 2004). Self-conscious emotions though, may be a particularly persuasive tool for reducing harmful behaviors, as they carry extremely strong personal implications, due to the presence of self-awareness and self-representation (Agrawal & Duhachek, 2010). Second, comparing emotions in terms of valence is of importance, as introducing other emotional appeals may help to draw new attention to an issue when a campaign has matured and the effect of one type of appeal has diminished (Nabi, 2002). In the current case, this means introducing positive emotions in a campaign that has been predominantly negative: even though the name suggests otherwise, in the most recent campaign “I love my ears’ the focus again was on the risks of hearing damage due to loud music and the negative consequences of that hearing loss (“‘I love my ears’ campagne gestart,” 2014). However, introducing positive emotions in such a campaign could gain attention especially of those who consider themselves as overly familiar with it.
Further, introducing positive emotions into a hearing protection campaign could also help bypass the so-called third person effect. According to Lewis et al. (2007) the third person effect occurs when an individual perceives a persuasive message as being of greater relevance and influence to others than to the self. They show that the third person effect is a main
shortcoming of negative appeals, as people see negative appeals as more relevant and influential for others, but not for the self. The third person effect is of concern for health messages, because an individual’s behavior is less likely to change or become consistent with the recommendations as stated in the message when the message is not perceived to be relevant to the self. This may imply that positive emotions are more influential, as they may be more easily perceived as relevant to the self.
In sum, it is believed that there is sufficient need and justification to explore the persuasive utility of emotional appeals that goes beyond valence and beyond just fear. In particular it is argued that it is necessary to gain insights into the effect of self-conscious versus basic emotions, and as such into the potential use of specific emotions. It is of great importance to learn what effect specific emotional appeals will have, because learning what exact information processes and persuasive outcomes specific emotions have provides the opportunity for more precise manipulation of actions or intentions. This results in the
increased ability to create applicable interventions based on the responses of those emotional experiences (Nabi, 2010). Therefore, the current research will focus on the following research question: To what extent, if at all, do positive versus negative and basic versus self-conscious emotions differ in their effect on persuasion outcomes and the third person effect?
Theoretical background
Discrete emotions are generally believed to be “internal mental states representing evaluative reactions to events, agents or objects that vary in intensity” (Nabi, 2010, p. 153). They are short-lived, complex, intense and directed at some external stimuli (Griskevicius et al., 2010; Nabi, 2002). The current body of literature is mainly based on two models:
dimensional and discrete. The focus of the dimensional model is on emotions as a more generalized state, looking at the degree of positive or negative feelings aroused by a stimulus and the effect this has on different kinds of cognitive and behavioral outcomes. Within this
model, emotions such as pride, happiness and hope might be assessed similarly though, as these all have a positive valence (Lang, Greenwald, Bradley & Hamm, 1993; Nabi, 2010). The discrete model, however, defines emotions as a more categorical state, meaning that each of the emotions has a unique set of personal interpretations or thought patterns, and thus all have distinctive associated outcomes, such as behaviors and subjective experiences (Angie et al., 2011; Izard, 2007; Nabi, 2010). As a consequence, this means that we both can identify specific emotions, as well as predict the outcomes of those emotions (Nabi, 2010). For example, research has shown that shame and guilt both have different psychological and behavioral consequences, in such a way that anger and perceived manipulative intent are more likely to be associated with shame than guilt (Boudewyns, Turner, & Paquin, 2013).
Because of these unique divergent outcomes, discrete emotions could certainly have implications for the process and the direction of persuasive influence (Nabi, 2002). This persuasive influence is becoming more and more important in advertising, as both functional and price advantages are becoming increasingly more eroded (O’Shaughnessy &
O’Shaughnessy, 2004). Also, research has shown that traditional persuasive information campaigns are not as effective as they used to be (Asbeek Brusse et al., 2010). Even though emotional appeals have been used in persuasion for decades, for example by marketers raising disgust in order for potential customers to buy certain health-related products, it is realized that there is a need to use more emotionally appealing methods in the relatively boring fields of health promotion and education (Asbeek Brusse et al., 2010; DeSteno, Wegener, Petty, Rucker & Braverman, 2004). A more in depth understanding of discrete emotions in
persuasive messages is thus essential for the field of health communication, as now it is more important than ever to learn what the most effective way to target an audience is.
Valence
There has been an increased amount of calls for better hearing conservation education. However, the medical community has so far failed to create a message that is effective in reducing the amount of youngsters at risk for hearing loss (Chung et al., 2005). There may be several reasons for this. First, it may be possible that the ones who need to be reached have become tired of the predominantly negative appeals in hearing loss interventions. Especially for youngsters who visit music events a lot and therefore feel as if they have seen the
messages too often, but also have the feeling that they are quite knowledgeable on the subject of noise induced hearing loss due to their experiences, the effect of one type of appeal, i.e. a negative one, might have diminished. Further, in order to develop interventions that bring about the desired behavioral change, it is necessary to learn about what the important determinants of that behavior are (Vogel et al., 2007). So far, however, the knowledge we have on health messages are based on cognitive appeals and fear appeals. From the latter we know that they can indeed evoke health behavior change (Agrawal et al., 2007; Stephenson et al., 2005). Obviously, this is not enough in the current case, looking at the rising number of youngsters who suffer noise-induced hearing loss. Because of these reasons, it could be assumed that a new approach, using a positive appeal, may have the desired effect. Even though it feels natural to focus on scary, gory or sad messages to change behavior, research has shown that positive appeals do have the promise of having beneficial effects on outcomes such as reasoning, judgment and decision making (Lewis et al., 2007; Pham, 2007). In line with the issue of hearing protection, this is also confirmed by Stephenson and his colleagues (2005) in their research on hearing protection among coal miners. They conclude that positive appeals indeed have potential, and may be more effective than negative appeals. Based on the previous discussion, the following hypothesis is proposed:
H1: exposure to a persuasive message including a positive emotion will result in significantly more favorable attitude, intention and behavior toward hearing protection, than exposure to a persuasive message including a negative emotion
Type
Until recently, self-conscious emotions have been poorly studied (Lewis, 2011). Self-conscious emotions are different from basic emotions in that they require self-evaluative processes (the cognitive ability to reflect on the self), i.e. awareness and
self-representations. People tend to feel self-conscious emotions when they become aware that they have (not) lived up to some self-representation (Tracy & Robins, 2014). For example, someone might feel great sadness after losing either a lottery or a quiz. However, the former event would (usually) not involve any self-evaluation (as it is simple luck), while the latter one would elicit a self-evaluative process (e.g. “What does this say about my intelligence?”). Thus, only losing a quiz would also generate a self-conscious emotion, such as shame or embarrassment. In contrast to self-conscious emotions, basic emotions can be elicited in the absence of self-evaluation. These emotions are biologically based, shared with other animals and identifiable via discrete, universally recognized facial expressions. This group of
emotions includes fear, happiness, sadness, disgust, anger and surprise (Tracy & Robins, 2014).
Another distinctive feature of self-conscious emotions is that they develop later in life: while basic emotions emerge already in the first year, it takes until the second half of the second year of life until the earliest of self-conscious emotions appear. Two main
explanations for the later development of self-conscious emotions are provided: one, because this is the age where a child reaches the point where self-awareness is developed; and second, because this is the age where a child comes to understand rules and standards, which
determine appropriate social behavior, and learn to create self-evaluations in line with that behavior (Lewis, 2011). So, in order to experience self-conscious emotions, specific cognitive ability is necessary, while this is not needed for basic emotions.
Self-conscious emotions can be particularly helpful in persuasive messages trying to reduce harmful behaviors, as they often carry strong personal implications (Agrawal & Duhachek, 2010). Due to the fact that it is necessary to make evaluations about the self in order to experience self-conscious emotions, it seems probable that a message is therefore more elaborately processed, as all parts of the message need to be evaluated in relation to the self. This means that if a message has attitude change as a consequence, the message
including a self-conscious emotion will evoke a relatively stronger, more enduring and resistant attitude change than the message including a basic emotion.
H2: exposure to a persuasive message including a self-conscious emotion will result in significantly more favorable persuasion outcomes compared to a persuasive message including a basic emotion
Negative Positive
Basic Fear Happiness
Self-conscious Shame Pride
Figure 1. Overview of the included emotions divided by type and valence
In the current research four emotions were chosen to be compared, as such to cross-compare both type as well as valence, Figure 1 gives an overview of these emotions. First, fear, as this basic negative emotion is used in current campaigning on hearing loss. Second, shame, as this emotion shares a negative appraisal with fear, however, in contrast to fear, is a
self-conscious emotion. Shame could be a useful emotion in a public health message, as it arises from seeing oneself through the eyes of another and realizing that behaviors producing the emotion are not congruous with his or her ideal self (Boudewyns et al., 2013). However, people want to seem consistent, as this is associated with intellectual strength. The desire to appear consistent exerts considerable influence over behavior (Cialdini & Sagarin, 2005), which means that arousing shame within a message, could have preferred behavioral change as a consequence. Third, happiness is included, as a basic positive emotion. Research has shown that happiness generates feelings of confidence, which in turn is essential in changing behavior, as an individual gets the sense that they can perform the requested behavior
effectively (Basil, Ridgway & Basil, 2008; Nabi, 2002). The final emotion to be included is pride, as it shares the positive valence appraisals with happiness, and even operates very similarly to happiness, as it also enhances esteem. The difference is, that pride is a self-conscious emotion, also resulting in expressive behaviors, such as public announcements of an achievement (Nabi, 2002). Research has shown that pride, dissociated from related constructs such as self-efficacy, self-esteem and generalized positive affect, seems to be a main motivator of adaptive behavior and social perception (Williams & DeSteno, 2009). Based on this discussion, and the ones presented earlier on positive and self-conscious emotions, the following hypothesis is proposed.
H3: valence and type will interact such that a positive emotion will result in
significantly the most favorable persuasion outcomes if this positive emotion is also a self-conscious emotion (i.e. pride)
That Is Not Meant for Me – Introducing the Third-Person Effect
Over the past decades, many empirical investigations have supported and developed the third-person effect hypothesis. The third-person effect hypothesis maintains that people systematically overestimate the influence that mass communications have on the attitudes and behavior of others. More specifically, this hypothesis suggests that communication’s greatest impact will not be on “me” or “you”, but on “them” – the third persons, which is an often erroneous viewpoint (Davison, 1983). For example, Davison talks about film censorship. The people who make the judgment on a violent movie would never admit to any unfavorable influence on their values or attitudes by that movie, however, they do decide that these same movies would affect the public, in such a way that they deem it necessary to restrict access or demand changes to the movies. Moreover, support for the third-person effect has been found when looking at negative political advertising, when researching neutral advertising, in the context of news media coverage as well as media messages (e.g. rap lyrics), and in more general media activities, such as watching television (White & Dillon, 2000).
In their study on drunk driving public service announcements, Lewis and colleagues (2007) found an association between negative emotional appeals and the third-person effect. They claim that health advertisers should be wary of that association, as previous research has shown that when people perceive greater influence of a message on others, rather than on themselves, their attitudes and behaviors are less likely to change or become consistent with the message’s recommendations (Lewis et al., 2007). Additionally, research has shown that for positive message attributes, such as desirability and congruency, more first-person effects are found. As people tend to place themselves in a positive light, the third-person effect decreases and/or a first-person effect emerges once media messages display more positive message attributes (White & Dillon, 2000). Based on this discussion, the following is expected.
H4: exposure to a message including a negative emotion will result in a significantly greater third-person effect than exposure to a persuasive message including a positive emotion
As stated earlier, self-conscious emotions require self-evaluative processes. During these processes, the focus is on the “I”, as an individual needs to reflect on the self. Therefore, it could be assumed that when a self-conscious emotion is successfully induced by a
persuasive message, automatically a process is triggered where the main focus is on “me”, rather than on “them”. Therefore, it could be assumed that the third-person effect decreases and/or a first-person effect emerges, due to the shifted focus by the aroused emotion. Based on this reasoning, the following is expected.
H5: exposure to a message including a basic emotion will result in a significantly greater third-person effect compared to a persuasive message including a self-conscious emotion
Finally, as previously stated, research has shown that negative appeals are associated with the third-person effect (Lewis et al., 2007) and fear is categorized as a basic emotion (Tracy & Robins, 2014). Therefore, it would be expected that the third-person effect is most present when a message incorporates a fear appeal. Not only should the negative appeal increase the third-person effect, but additionally, individuals are not triggered by this basic emotion to reflect on themselves. Therefore, the final expectation is as follows.
H6: valence and type will interact such that a negative emotion will result in significantly the greatest third-person effect, if this negative emotion is also a basic emotion (i.e. fear)
Figure 2. Conceptual Model with independent variable, moderator and four dependent
variables
Method
The current research is set up as a quantitative research as such to provide an empirical test of the influence of discrete emotions in a public health message on wearing hearing protection on persuasion outcomes, i.e. attitude, purchase intention, purchase behavior, and the third person effect. First, a pretest was done as such to determine the appropriate manipulation material.
Pretest
For the pretest, a true experimental 2 (valence: positive versus negative) x 2 (type: basic versus self-conscious) within-subjects research design was chosen to analyze which of the stimulus material was appropriate to use for the main study. This resulted in four
experimental conditions: Fear, Shame, Happiness and Pride.
Sampling procedure. In total, 120 Dutch participants between 18 and 25 years old filled out the online survey provided via Qualtrics. This specific target group was chosen, because interventions aimed at changing behaviors regarding loud music at events should be aimed at older adolescents and young adults, as these groups attend music events more frequently than younger ones (Vogel et al., 2007). They were selected by a convenient
snowball sample, through social media and e-mail. Participants who received the e-mail were part of the social circle of the researcher, while social media distribution happened both within and outside of the social circle of the researcher by the use of various Facebookpages (e.g. the Facebookpages of universities in the Netherlands, the Facebookpage ‘Respondenten gezocht’, and more). Given that there were four conditions; participants were randomly assigned to one of these four conditions (30 to each). 39.2% (SD = 0.49) of the participants was male, and the average age was 23 (SD = 1.96).
Procedure. The advertisements for each emotion condition were divided across three themes: listening to music, dancing to music, and style/being cool. Both listening and dancing to music were chosen as they relate to the content of the advertisement: wearing hearing protection when visiting music events. Style/being cool was chosen, as previous research has shown that wearing hearing protection may be outweighed by cultural pressures to conform to stylistic norms (Crandell, Mills, & Gauthier, 2004). Participants got to see all three themes for one condition only. Advertisements consisted of a picture with a sentence underneath it, both related to one theme and emotion. Also, in order to keep self-efficacy equal, all
advertisements were accompanied by a text on the side, providing some information on hearing loss, how to prevent hearing loss and where to buy hearing protection (see Appendix A – C for all advertisements and self-efficacy text). This information was retrieved from the website of the ‘I love my ears’ campaign, as such to provide participants with reliable information. After being shown one advertisement, participants answered a forced choice question indicating which emotion they felt most when looking at the advertisement; fear, shame, happiness or pride. Then, on a rating scale from 1 to 7, participants indicated to what extent they felt the following emotions when looking at the advertisement; fear, shame, happiness and pride, and other control emotions; disgust, anger, sadness and guilt. Further, participants were asked to rate their believe that wearing ear plugs protects their hearing, as well as the extent to which they felt able to buy ear plugs, as such to measure their
self-efficacy (both measures adapted from Stephenson et al., 2005). Also, they were tested on their critical view of the advertisement, by asking which term described the advertisement best according to them (on a 7 point scale): professional – unprofessional, credible – non-credible, trustworthy – untrustworthy. These measures were used to assess to what extent respondents looked at the advertisements as if they could be real in the sense that they were created by professionals, coming from a credible, trustworthy source. Procedure was the same for the second and third advertisement. The complete questionnaire can be found in Appendix D.
In order to analyze the manipulation material, one way ANOVA’s were done, as such to be able to compare each condition to each other, to see whether the advertisement had the desired effect (namely, significantly different answers across the dependent variables for each condition).
In Table 1 the results are summarized for the ANOVA’s done on set 1, which had “listening to music” as a theme. Overall, it shows that there were significant differences on the dependent variables, except for guilt. When looking at the four most important DVs: fear,
shame, happiness and pride, it can be stated that the manipulations in this set worked quite as intended. People in the fear condition rated fear significantly higher than people in the happiness and pride condition, though not higher than people in the shame condition. People in the shame condition rated fear significantly the highest. Second highest was shame, however, they only rated this significantly higher than people in the happiness condition, not than people in the fear or pride condition. People in the happiness condition rated happiness significantly higher than people in the shame and fear condition, but there was no difference with people in the pride condition. And finally, people in the pride condition rated pride significantly higher than people in the happiness and fear condition, but not in the shame condition. As the shame manipulation did not work as intended (fear was rated highest in this condition), it was decided that this set did not work well enough for the main study.
In Table 2 the results are summarized for the ANOVA’s done on set 2: “dancing to music”. Overall, it shows significant differences on the dependent variables, however, there is one exception: pride. As pride is one of the four emotions studied in the current research, it was decided that the manipulations in this set did not work well enough for the main study. Table 3 summarizes the results of the ANOVA’s done on set 3, themed “style/being cool”. Overall, this theme is the only one to present only significant values, meaning that for each of the conditions someone was in, the corresponding emotion was chosen significantly more than any of the other (control) emotions. People who were in the fear condition assigned a significant higher value to fear than people in the happiness and pride condition, although not than people in the shame condition. People who were in the shame condition rated shame significantly higher than people in the happiness and pride condition, though not than people in the fear condition. People in the happiness condition chose happiness significantly more often than people in the fear and shame condition, but not than people in the pride condition. Finally, people in the pride condition rated pride higher than people in the fear condition, but
not significantly higher than people in the shame and happiness condition. Moreover, a Chi square test was done on the forced choice question “While looking at the advertisement, I feel – fear, shame, happiness, pride”. Pearson Chi-square showed that there was a significant difference between the condition one was in and the answer one gave to this question, χ² (9, N = 120) = 70.89, p < .001, in such a way that for each of the conditions one was in, the highest percentage gave the correct answer (e.g. of people in the fear condition, the highest
percentage answered “fear” on the forced choice question, rather than another emotion). The fact that shame and fear, and happiness and pride, are not rated significantly different throughout the entire set may be because the emotions are assessed quite similarly, as the first two are both negative and the second two are both positive. This problem might be solved by asking participants more specifically about their responses. For each of the
emotions, several questions on the participants’ responses are asked that are in accordance to that emotion. Research has shown that response types are proven to be different across emotions, and by asking about those responses there is an opportunity for more detail as participants may not always know which responses belong to which emotion. For example, they may attribute crying to guilt, while this is actually a response of shame (Roseman, Wiest & Swartz, 1994).
Table 1
Results of ANOVA for set “listening to music”
DV Condition F(3, 119) p η² Fear M (SD) Shame M (SD) Happiness M (SD) Pride M (SD) Fear 4.57 (1.43) a 3.38 (2.00) ab 2.00 (1.47) c 2.34 (1.76) bc 13,480 .000 .258 Shame 3.37 (1.75) a 3.18 (2.05) ab 1.81 (1.47) c 2.10 (1.26) bc 6,155 .001 .137 Happiness 1.77 (1.17) a 2.44 (1.44) a 4.33 (1.49) b 4.52 (1.68) b 26,233 .000 .404 Pride 1.60 (0.93) a 2.38 (1.60) ab 2.19 (1.36) a 3.34 (1.84) b 7,104 .000 .155 Anger 2.83 (1.84) 2.35 (1.72) 1.74 (1.29) 1.79 (1.01) 3,367 .021 .080 Disgust 3.13 (1.72) ab 3.24 (1.76) a 1.44 (0.85) c 2.14 (1.36) bc 9,617 .000 .199 Guilt 3.07 (1.57) 2.85 (1.96) 2.07 (1.39) 2.45 (1.53) 2,063 .109 .051 Sadness 4.67 (1.45)c 3.15 (1.84) a 1.85 (1.13) b 2.07 (1.33) ab 21,842 .000 .361 Note. Means (M) per row having a different letter differ significantly from each other with p<.05 in Scheffe’s Post Hoc test
Table 2
Results of ANOVA for set “dancing to music”
DV Condition F(3, 119) p η² Fear M (SD) Shame M (SD) Happiness M (SD) Pride M (SD) Fear 4.10 (1.47) b 2.85 (1.71) a 2.07 (1.59) a 1.97 (1.45) a 11,601 .000 .231 Shame 3.13 (1.68) a 3.21 (1.86) a 1.89 (1.45) b 1.90 (.98) b 6,847 .000 .150 Happiness 1.77 (1.17) b 3.56 (1.88) a 4.70 (1.49) a 4.72 (1.93) a 20,740 .000 .349 Pride 1.90 (1.42) 2.62 (1.56) 2.70 (1.59) 3.03 (1.86) 2,601 .055 .063 Anger 2.73 (1.53) a 2.21 (1.63) ab 1.52 (.75) b 1.79 (1.29) ab 4,321 .006 .101 Disgust 2.53 (1.53) a 2.79 (1.87) a 1.44 (.89) b 1.83 (1.31) ab 5,337 .002 .121 Guilt 3.03 (1.59) a 2.56 (1.71) ab 1.78 (1.05) b 2.07 (1.36) ab 4,099 .008 .096 Sadness 4.13 (1.48) c 2.65 (1.81) a 1.48 (.89) b 2.00 (1.41) ab 17,905 .000 .317 Note. Means (M) per row having a different letter differ significantly from each other with p<.05 in Scheffe’s Post Hoc test
Table 3
Results of ANOVA for set “style/being cool”
DV Condition F(3, 119) p η² Fear M (SD) Shame M (SD) Happiness M (SD) Pride M (SD) Fear 3.53 (1.74) a 3.03 (1.96) ab 1.56 (.93) c 1.90 (1.47) bc 9,856 .000 .203 Shame 3.17 (1.70) a 3.32 (2.00) a 1.67 (1.21) b 2.14 (1.46) ab 7,049 .000 .154 Happiness 1.70 (1.26) a 2.32 (1.63) a 4.41 (1.42) b 4.45 (1.45) b 28,022 .000 .420 Pride 1.90 (1.35) a 2.18 (1.66) ab 2.56 (1.55) ab 3.24 (1.96) b 3,709 .014 .088 Anger 2.43 (1.46) 2.44 (1.69) 1.56 (1.25) 1.69 (1.29) 3,182 .027 .076 Disgust 2.50 (1.36) a 2.65 (1.70) a 1.41 (.80) b 1.72 (1.22) ab 5,964 .001 .134 Guilt 2.70 (1.49) a 2.26 (1.62) ab 1.59 (1.08) b 2.10 (1.59) ab 2,736 .047 .066 Sadness 2.97 (1.56) a 2.88 (1.90) a 1.41 (.75) b 1.79 (1.24) b 8,399 .000 .178
Note. Means (M) per row having a different letter differ significantly from each other with p<.05 in Scheffe’s Post Hoc test
Moreover, scale variables were formed for credibility and self-efficacy. Table 4-6 show the results for the Principal Component Analysis, reliability analysis of the scales and the ANOVA’s done on credibility and self-efficacy for each of the sets. In Table 4 it can be seen that for set 1 “listening to music” there was a significant difference in credibility between the fear and shame condition, in such a way that people in the shame condition assessed the advertisement as less credible. No significant differences were found for self-efficacy. Table 5 summarizes set “dancing to music”, and again shows a significant difference in credibility, though not in self-efficacy. Table 6 shows that for theme “style/being cool” there is no sign of difference in credibility between the four conditions, and credibility overall is assessed quite well (somewhere along 3-4 out of 7). This is good, as it is desirable that the participants see the advertisement as something they could possibly encounter in real life. For that, it is necessary that they find the advertisement credible. Self-efficacy did show a
difference, namely happiness was assessed a little lower, however, as the others scored fairly high (and happiness as well with a mean of 4.65 out of 7), there is no necessity to change the self-efficacy message.
Table 4
Results for Principal Component Analysis, reliability analysis and ANOVA for credibility and self-efficacy for set “listening to music”
DV Condition EV R² Cronbach’s α F(3,119) p η² Fear M (SD) Shame M (SD) Happiness M (SD) Pride M (SD) Credibility 4.56 (1.35) a 3.25 (1.40) b 4.20 (1.67) ab 3.87 (1.60) ab 2.48 82.50 .892 4,323 .006 .101 Self-efficacy 5.65 (1.18) 5.65 (1.33) 5.15 (1.13) 5.78 (1.00) 1.46 72.75 .716 1,558 .203 .039 Note. Means (M) per row having a different letter differ significantly from each other with
p<.05 in Scheffe’s Post Hoc test
Table 5
Results for Principal Component Analysis, reliability analysis and ANOVA for credibility and self-efficacy for set “dancing to music”
DV Condition EV R² Cronbach’s α F(3,119) p η² Fear M (SD) Shame M (SD) Happiness M (SD) Pride M (SD) Credibility 4.52 (1.15) a 3.43 (1.31) b 4.36 (1.48) ab 4.00 (1.76) ab 2.54 84.66 .907 3,637 .015 .086 Self-efficacy 5.63 (1.17) 5.60 (1.40) 4.96 (1.22) 5.64 (.96) 1.51 75.57 .676 2,098 .104 .051 Note. Means (M) per row having a different letter differ significantly from each other with
Table 6
Results for Principal Component Analysis, reliability analysis and ANOVA for credibility and self-efficacy for set “style/being cool”
DV Condition EV R² Cronbach’s α F(3,119) p η² Fear M (SD) Shame M (SD) Happiness M (SD) Pride M (SD) Credibility 4.14 (1.46) 3.57 (1.49) 3.40 (1.63) 4.14 (1.85) 2.62 87,40 .927 ,904 .442 .023 Self-efficacy 5.55 (1.20) ab 5.59 (1.33) b 4.65 (1.57) a 5.67 (.94) b 1.57 78.24 .716 3,955 .010 .093
Note. Means (M) per row having a different letter differ significantly from each other with p<.05 in Scheffe’s Post Hoc test
Main Study
For the main study, a true experimental 2 (valence: positive versus negative) x 2 (type: basic versus self-conscious) between-subjects research design was employed in order to answer the hypotheses. The independent variable was formed by the valence of the emotion, meaning either a positive or a negative emotion. There were four dependent variables: attitude,
intention, behavior and the third person effect. Type of emotion, basic versus self-conscious, was included in the research as moderator.
Participants. In total, 160 Dutch participants between 18 and 25 years old filled out the online survey via Qualtrics. The sampling procedure was similar to the sampling
procedure of the pretest, although for the main study all people who participated in the pretest (120 participants) were excluded from taking the survey. It was stated in both the e-mail as well as the social media message that if they did participate in the pretest, they could not participate in this experiment. In order to make sure that they were indeed not included in the data analysis, participants were asked in the survey whether they took part in the pretest. Again, participants were randomly assigned to one of the four manipulations (40 to each). 38.8% (SD = 0.49) of the participants was male, and the average age was 23 (SD = 2.03). Further, almost half of the participants (48.8%, SD = 1.43) completed either a bachelor or
master at a university. Finally, overall participants indicated their English proficiency as being quite fluent: with a mean of 5.78 (SD = 1.10) out of 7 it is safe to assume that language
formed a minimal barrier for participants in the experiment.
Procedure. Each of the participants got to see only one advertisement from the theme “style/being cool”. Again, in order to keep self-efficacy equal, all advertisements were
accompanied by the same text as in the pretest. The advertisements used can be found in Appendix C.
Measures for the main study. When indicating emotions, participants were asked the same forced choice and rating scales questions as in the pretest. Also, for each of the
conditions a specific scale could be created with the following questions, all indicated on a scale from 1 (not at all) to 7 (very much). Fear: (a) I felt my heart pounding; (b) I thought of how bad things could get; (c) I felt like looking away; (d) I wanted to create more safety for myself. These items were based on research by Roseman and colleagues (1994). The PCA showed that the items loaded on two factors which had and Eigenvalue of 1.98 (R² = .49) and 1.09 (R² = .27). Cronbach’s alpha was .65, however after deleting item 3 “I felt like avoiding the picture on the screen”, a reliable scale for fear was formed (Cronbach’s α = .71, M = 2.65, SD = 1.39). In order to form a reliable scale for shame, items from Roseman, et al. (1994)
were used as well: (a) I felt self-conscious; (b) I felt small; (c) It made me blush. The PCA showed that the items all loaded on one factor (EV = 2.08, R² = .52), and Cronbach’s alpha was .61. Deleting item 1 “I felt self-conscious” formed a reliable scale for shame (Cronbach’s α = .70, M = 1.70, SD = 0.91). The items used for the happiness scale (EV = 4.12, R² = .69, Cronbach’s α = .91, M = 3.41, SD = 1.51) were derived from the shortened version of the Oxford Happiness Questionnaire (Hills & Argyle, 2001): (a) I felt pleased with the way I am; (b) I felt that life is rewarding; (c) I felt satisfied about everything in my life; (d) I felt
pride, a scale developed by Tracy and Robins (2007) was used (Cronbach’s α = .96, M = 3,18, SD = 1.62): (a) I felt accomplished; (b) I felt like I was achieving; (c) I felt confident; (d) I felt
fulfilled; (e) I felt like I have self-worth; (f) I felt successful. The PCA showed that all items loaded on one factor, with an Eigenvalue of 4.92 (R² = .82).
Hearing protection behaviors. Participants indicated their current behavior by
responding to the following item on a seven-point Likert scale, ranging from 1 (not at all) to 7 (always): I wear hearing protection when visiting music events.
Behavioral intentions. Participants indicated their behavioral intent to wear hearing
protection with the following items on a seven-point Likert scale, ranging from 1 (strongly disagree) to 7 (strongly agree): (a) I intend to wear hearing protection regularly when visiting music events; (b) I will try to wear hearing protection regularly when visiting music events; and (c) I plan to wear hearing protection regularly when visiting music events (Ajzen, 2002). The PCA showed all three items loaded on one factor (EV = 2.77, R² = .92) and a reliable scale was formed (α = .96, M = 4.11, SD = 2.00).
Attitudes toward hearing protection. Participants indicated their attitude toward
hearing protection with the following items on a seven-point Likert scale, ranging from 1 (strongly disagree) to 7 (strongly agree): (a) wearing hearing protection is good; (b) wearing hearing protection is desirable; (c) wearing hearing protection is bad (reverse coded); and (d) wearing hearing protection is positive. Together these items formed a reliable scale (EV = 2.55, R² = .64, α = .80, M = 5.72, SD = 1.18).
Behavior. Behavior was measured by asking the participants whether they would like
to receive the links to the web stores where they could buy their own set of earplugs at the end of the survey.
Third-person effect. Participants indicated the impact of the information they were
all) to 7 (very much) retrieved from Price and Tewksbury (1996): (a) How much do you think the information you were presented with has influenced your own attitudes toward wearing hearing protection?; (b) How much do you think the information you were presented with has influenced most other people’s attitudes toward wearing hearing protection? A new variable TPE was computed in which the effect on others was subtracted by the effect on the self, so to see to what extent a third-person effect was in place (M = 0.29, SD = 1.33).
Control variables. To control for different explanations for the effects, standard
demographic data was collected, including gender, age and education level. Also, people were asked about their English proficiency. The complete questionnaire can be seen in Appendix E.
Results Manipulation Check
First, the manipulation checks were analyzed to see whether participants in a condition indeed felt that specific emotion more so than participants in other conditions. Three items were used to see whether the manipulation worked as intended. First, a forced choice emotion question was asked. Pearson Chi-square showed that there was a significant difference
between the condition one was in and the answer one gave to this question, χ² (9, N = 160) = 0.90, p < .001. Indeed output shows that people in the fear (62,5%), shame (45%) and happiness (80%) condition chose the corresponding emotions more often, only for the pride condition this was not the case (62,5% chose happiness). This may be attributable to the fact that pride and happiness both share a positive valence. The second manipulation check was a question where participants indicated on a scale to which extent they felt specific emotions. The results of this second manipulation check are summarized in Table 7. As Table 7 shows, when looking at this manipulation check, the shame manipulation seemed to have not worked as intended, as people in the shame condition did not rate shame significantly higher than people in the other conditions. The three other emotions do show a significant difference
(people in the happiness condition chose happiness significantly more than people in the other conditions), although not completely as intended: people in the fear condition did rate fear higher than people in the shame and pride condition, but not in the happiness condition, people in the pride condition did show a significant difference with people in the fear condition, but not with those in the shame and happiness condition.
Table 7
Results of the ANOVA for manipulation check 2
DV Condition F(3, 159) p η² Fear M (SD) Shame M (SD) Happiness M (SD) Pride M (SD) Fear 3,85 (1.98) a 2.55 (1.84) b 2.98 (1.67) ab 2.07 (1.54) b 7,289 .000 .123 Shame 2.68 (1.94) 2.78 (1.92) 2.22 (1.70) 2.20 (1.68) 1,086 .357 .020 Happiness 1.73 (0.99) a 2.33 (1.33) a 5.07 (1.44) b 4.03 (1.83) c 46,412 .000 .472 Pride 1.57 (0.81) a 2.53 (1.72) ab 3.22 (1.66) b 2.98 (1.80) b 8,775 .000 .144 Anger 2.12 (1.65) 2.15 (1.69) 1.55 (0.99) 1.45 (0.90) 2,981 .033 .054 Disgust 2.30 (1.71) 2.52 (1.74) 1.73 (1.38) 1.68 (1.12) 3,129 .027 .057 Guilt 2.90 (2.05) 2.38 (1.61) 2.60 (1.95) 2.20 (1.51) 1,139 .335 .021 Sadness 3.17 (1.89) a 2.50 (1.85) ab 1.95 (1.22) b 1.80 (1.09) b 6,412 .000 .110
Table 7. Manipulation check 2: emotion scale. Means (M) per row having a different letter differ significantly from each other with p<.05 in Scheffe’s Post Hoc test
Finally, a third manipulation check was done, to go even more in depth into what emotions participants were experiencing. They were asked about their specific responses, as response types are proven to be different across emotions: for example, fear is associated with avoidance or heightened readiness for action that can reduce the possibility of harm, while sadness has crying and cessation of active pursuit of an unattainable desired end as a
consequence (Roseman, et al., 1994). So, this question gave the opportunity to more detail. In Table 8 the results of this manipulation check are summarized. Table 8 shows that, again, the
shame manipulation seemed to have not worked as intended, as people in the shame condition did not rate shame responses significantly higher than participants in the other conditions. People in the fear condition did rate fear significantly higher than people in the shame condition, but not than those in the happiness and pride condition. People in the happiness condition did rate happiness higher than people in the other conditions, but there was no significant difference. People in the pride condition did rate pride higher than people in fear and shame condition, but not than people in the happiness condition; however these
differences were not significant either. Overall, it can be concluded that the manipulations did work quite as intended, with here and there an exception, especially in the shame condition.
Table 8
Results of the ANOVA for manipulation check 3
DV Condition F(3, 159) p η² Fear M (SD) Shame M (SD) Happiness M (SD) Pride M (SD) Fear 3.21 (1.58) a 2.29 (1.47) b 2.49 (1.29) ab 2.60 (1.04) ab 3.367 .020 .061 Shame 1.85 (1.11) 1.64 (0.71) 1.61 (0.72) 1.71 (1.02) .554 .645 .011 Happiness 3.27 (1.77) 3.11 (1.52) 4.03 (1.31) 3.23 (1.25) 3.153 .027 .057 Pride 2.80 (1.51) a 2.82 (1.55) a 3.96 (1.73) b 3.15 (1.47) ab 4.769 .003 .084
Table 8. Manipulation check 3: emotional components. Means (M) per row having a different letter differ significantly from each other with p<.05 in Scheffe’s Post Hoc test
Control for Other Explanations
In order to check for differences between people, respondents were asked some demographic questions. Before testing the hypothesis, it was checked whether there was need to control for one of these variables. First, bivariate correlations were done. Pearson’s r
showed that purchase intention and age were significantly negatively correlated, however, this relationship was negligible (r = -.18, p = .021 < α = .05). Further, Pearson’s r showed that attitude and English proficiency where significantly negatively correlated, however, again this
relationship was negligible (r = -.18, p = .022 < α = .05). Gender showed no correlation with the dependent variables, and education and third-person effect did show a relationship, χ² (24, N = 160) = .38, p = .034.
Main Analysis
In order to conduct the main analyses, two new variables were created:
emotion_valence with 1 = fear and shame, 2 = happiness and pride, and emotion_type with 1 = fear and happiness, 2 = shame and pride. To test the attitudes toward wearing hearing protection a univariate analysis of variance (ANOVA) was conducted with emotion_valence and emotion_type as independent variables (IVs), and the attitude scale variable as dependent variable (DV). The results showed no significantly higher attitude toward wearing hearing protection in the positive condition, F(1,156) = .51, p = .477, η² = .00, or in the self-conscious condition, F(1,156) = .30, p = .585, η² = .00. Also, no interaction effect between valence and type on attitude was found, F(1,156) = .00, p = .960, η² = .00.
Second, to test the intention toward wearing hearing protection, another ANOVA was conducted with emotion_valence and emotion_type as IVs and the intention scale variable as DV. Again, the results showed no significantly higher intention toward wearing hearing protection in the positive condition, F(1,156) = 2.49, p = .116, η² = .02, or in the self-conscious condition, F(1,156) = 1.45, p = .230, η² = .01. However, a marginally significant result was found for the interaction effect between valence and type on intention, F(1,156) = 3.41, p = .067, η² = .02, in such a way that people in the negative basic emotion condition (i.e. fear) scored the highest on the intention scale (M = 4.83, SD = 1.99), while people in the positive basic emotion condition (i.e. happiness) scored the lowest on the intention scale (M = 3.77, SD = 1.82). Figure 3 visualizes the results of the interaction between valence and type using the means of intention.
3,6 3,8 4 4,2 4,4 4,6 4,8 5 Negative Positive Intenti on, Me an Basic Self-conscious
Figure 3. Interaction effect of valence and type on intention. This figure shows that the intention is higher for participants in the negative basic emotion condition than for participants in the positive self-conscious condition.
Third, to test the behavior of wearing hearing protection, a binary logistic regression was done. The results of this test showed that there were no significant main effects of type or valence on behavior, and no interaction effect (all p’s > .260).
Finally, to test the third-person effect, a new variable TPE was created, where the answers people gave to how much the information had influenced others was subtracted from how much the information had influenced themselves. Then, a univariate analysis of
covariance (ANCOVA) was conducted, with emotion_valence and emotion_type as IVs, TPE as DV and education as control variable, as this variable has proven to have a significant relationship with third-person effect. The results showed no significantly greater third-person effect for people in the negative condition, F(1,155) = .2.44, p = .120, η² = .02. However, a marginally significant effect was found on the third-person effect for type, F(1,156) = 3.21, p = .075, η² = .02. Participants who were in the basic emotion condition showed a greater third person effect (M = .48, SD = 1.28) than people who were in the self-conscious emotion
condition (M = .10, SD = 1.36). Further, no interaction effect between valence and type on third-person effect was found, F(1,155) = 3.02, p = .188, η² = .01.
Conclusion and Discussion
One of the primary goals of the current experiment was to compare the effects of four emotions in persuasive messages on hearing protection attitudes, intentions and behavior. Neither for valence (H1) or for type (H2) statistically significant main effects on these persuasion outcomes could be found. However, when looking at the interaction of valence and type (H3) a marginally significant interaction effect could be found on intention, in such a way that people in the negative basic emotion had the highest intention compared to people in the other conditions. This was contradictory to what was expected. The other goal of the current experiment was to learn more about the third-person effect. For valence (H4) no significant main effects on the third-person effect were found. However, for type (H5) a marginally significant main effect was found, in such a way that participants who received a persuasive message including a basic emotion indeed showed a greater third-person effect than did participants in the self-conscious condition. So, H5 could be confirmed. Further, no significant interaction effect of valence and type (H6) on the third-person effect could be found.
Different from expected, no significant main effect of type or valence was found on the persuasion outcomes, and only a marginally significant interaction effect was found on intention. However, this interaction effect showed that a negative emotion, if it was a basic one (i.e. fear) had the best approach. These unexpected outcomes are quite different from what research has been showing us in the last few years: evidence suggests that positive appeals may have the potential to be more effective than negative appeals (Stephenson et al., 2005; Lewis et al., 2007), that people can become tired of one appeal being overly used in a campaign (Vogel et al., 2007), and that self-conscious emotions can be particularly helpful in
trying to reduce harmful behaviors, as they often carry strong personal implications (Agrawal & Duhachek, 2010). The fact that the results of the current experiment were contradictory to what previous research has been suggesting actually show that perhaps the notion of adopting the strategy of using a single emotion (or type or appeal) within a persuasive message may be too simplistic. Each emotion is associated with different outcomes and shortcomings, and making them work together possibly creates a better combined effect on the desired results.
Also, the fact that no significant results of type and valence were found on behavior could be explained by the behavioral measure that was used. As options were limited, participants were asked if they wanted to see the links to web stores where they could buy earplugs. However, it could be argued that wanting to see such a link is different from actually wearing hearing protection. Therefore, future research should include a behavioral measure that measures if and how often participants wear hearing protection when they visit music events, for example by asking them this in a follow-up question.
Moreover, the lack of significant results of type and/or valence on the persuasion outcomes as stated may also be due to the fact that in the current experiment, participants saw the advertisement only once. However, research has shown that seeing an advertisement only once is not enough for people to change their attitudes, intentions or behavior. Indications are that it is necessary for people to see an advertisement at least three times, as such to give them the opportunity to learn the ad and be reminded by the ad of the consequences that need to be fulfilled (Krugman, 1972). This means that perhaps, when giving the participants the
opportunity to see the advertisements created for the current experiment more often, significant results can be found.
Research has already shown that the third-person effect is associated with negative appeals (Lewis et al., 2007), however; the current experiment could not find the same conclusion. There was no significant main effect of valence on the third-person effect, nor a
interaction effect of valence and type. This lack of evidence could possibly be due to the order effect. The order effect is the idea that different orders in which the questions are presented may influence participants’ answers (Strack, 1992). Research has shown that there is a slight difference in the estimates of media impact when participants are in the others-then-self question condition than in the self-then-others condition, and even though these differences were found not to be significant, there were indications that question order may influence people’s estimates of media impact in subtle ways, for example via knowledge on the topic (Price & Tewksbury, 1996). This means that preexisting knowledge on hearing protection participants may have had could have influenced the relationship between valence and the third-person effect, and also the interaction effect. Controlling for knowledge on the topic may have rendered significant results and should thus be taken into account when researching third-person effects in the future.
Another explanation for the lack of significant results in the current experiment could be the manipulations: three manipulation checks were done, which showed that the
manipulations did not work fully as intended, especially when looking at shame. People have trouble completely differing between emotions sharing the same valence (i.e. pride and happiness, fear and shame), as well as sharing type (i.e. fear and happiness, pride and shame) by looking at only an image and text. Even though the manipulations used in the current research were extensively pretested, future research should try to create even better
manipulations, for example by adding music to the advertisement. Research has shown that music affects the functional neuroanatomy of transient mood changes, meaning that music affects certain responses in the body, which in their turn are responsible for mechanisms such as targeting attention and the processing of emotions (Mitterschiffthaler, Fu, Dalton, Andrew & Williams, 2007). This means that when people for example hear sad music, their body responds with negative emotional responses. As such, the emotions that are trying to be
conveyed through the advertisement could be strengthened. Furthermore, more research should be done with regard to self-conscious emotions. As the current research showed, happiness and fear were easier to manipulate than pride and shame. As self-conscious emotions are relatively little researched, and hard to elicit, increased knowledge on these emotions may help in creating appropriate affect manipulations.
Moreover, it could have been possible that participants were not involved enough with the advertisement/topic, in such a way that the manipulation did not have a chance to be effective enough. Using a television commercial rather than a print commercial may solve this problem, as research has shown that television, due to the combination of visual and audio modes, evokes more attention and thus has more impact (Dijkstra, Buijtels & van Raaij, 2005). This means that even low-involved participants will process the advertisement better, consistent with “learning without involvement” by Krugman (1965).
The results of this study hold several theoretical and practical implications. Up to now, research on emotional appeals has been mainly focused on valence or on fear appeals. This study adds a first extension to the subject by combining research on the effects of emotion valence as well as emotion type. Currently, a lot of attention has been paid to negative emotions as a strategy to change health behaviors, however, broadening findings on positive emotions and identifying times when a positive appeal may be more effective than a negative appeal is of significant importance, as it increases the chance of desired effects happening. Furthermore, even though research has already shown that different types of emotions can have different effects on persuasive outcomes, still little attention has been paid to the effects of self-conscious emotions. By comparing the effects of self-conscious emotions versus basic emotions, insights are created into the way these different types of emotions work. Also, the current study expands knowledge on the third-person effect. So far, it was known that the third-person effect was associated with negative emotions, although no evidence for this
notion was found in the present study. However, this research shows that type of emotion has an effect on the third-person effect, in such a way that the third-person effect is greater when one receives a message that includes a basic emotion than a message that includes a self-conscious emotion. This effect could be ascribed to the self-evaluative processes involved with a self-conscious emotion; however, this cannot be said with certainty. Future research should therefore focus on the underlying processes of this effect.
This study holds some practical implications. Learning more about the effects of specific emotions in public health messages gives us the increased ability to create applicable interventions. As we learn more about the responses of emotional experiences, we are better able to use these in order to overcome the third-person effect and/or increase the first-person effect, and thereby steer people in the direction of desired health attitudes, intentions and behaviors.
This research also shows that it is too simplistic to think that only one specific emotion should be included in a public health campaign. Thus, in order to increase the likelihood that a persuasive message will reach the intended target audience with the desired effect,
practitioners must prepare multiple strategies and tactics, informed by extensive research. References
1.1 miljard jongeren lopen risico op gehoorschade (2015). Retrieved March 19, 2015, from
RTL, http://www.rtlnieuws.nl/nieuws/binnenland/11-miljard-jongeren-lopen-risico-op-gehoorschade.
Agrawal, N., & Duhachek, A., (2010). Emotional compatibility and the effectiveness of antidrinking messages: A defensive processing perspective on shame and guilt. Journal of Marketing Research, 47 (2) 263-273.
Agrawal, N., Menon, G. & Aaker, J. L., (2007). Getting emotional about health. Journal of Marketing Research, 44 (1) 100-113.
Ajzen, I. (2002). Constructing a TPB Questionnaire: Conceptual and methodological considerations. Retrieved March 19, 2015, from
http://www.people.umass.edu/ajzen/pdf/tpb.measurement.pdf.
Angie, A. D., Connelly, S., Waples, E. P. & Kligyte, V., (2011). The influence of discrete emotions on judgement and decision-making: A meta-analytic review. Cognition and Emotion, 25 (8) 1393-1422.
Arlinger, S. (2003). Negative consequences of uncorrected hearing loss – a review. International Journal of Audiology, 42, 2S17-2S20.
Asbeek Brusse, E.D., Neijens, P.C., & Smit, E.G., (2010). Talking to deaf ears? Two routes to narrative persuasion in entertainment education. In International Conference on
Future Imperatives of Communication and Information for Development and Social
Change, Bangkok.
Basil, D. Z., Ridgway, N. M., & Basil, M. D., (2008). Guilt and giving: A process model of empathy and efficacy. Psychology & Marketing, 25 (1) 1-23.
Boudewyns, V., Turner, M. M., & Paquin, R. S., (2013). Shame-free guilt appeals: Testing the emotional and cognitive effects of shame and guilt appeals. Psychology and Marketing, 30 (9) 811-825.
Chung, J. H., Des Roches, C. M., Meunier, J., & Eavey, R. D., (2005). Evaluation of noise- induced hearing loss in young people using a web-based survey technique.
Paediatrics, 115 (4) 861-867.
Cialdini, R. B., & Sagarin, B. J. (2005). Principles of interpersonal influence. In T. C. Brock & M. C. Green (Eds.), Persuasion: Psychological insights and perspectives (second edition., pp. 143-169). Thousand Oaks, CA: Sage.
hearing loss and hearing protection among racial/ethnically diverse young adults. Journal of the National Medical Association, 96 (2) 176-186.
Davison, W. P. (1983). The third-person in communication. The Public Opinion Quarterly, 47 (1) 1-5.
DeSteno, D., Wegener, D.T., Petty, R.E., Rucker, D.D., & Braverman, J., (2004). Discrete emotions and persuasion: The role of emotion-induced expectancies. Journal of Personality and Social Psychology, 86 (1) 43-56.
Dijkstra, M., Buijtels, H. E. J. J. M., & van Raaij, W. F. (2005). Separate and joint effects of medium type on consumer responses: A comparison of television, print and the Internet. Journal of Business Research, 58, 377-386.
Gehoorschade door te harde muziek (nd.). Retrieved October 16, 2014, from GGD,
http://ggd.groningen.nl/milieu-gezondheid/gehoorschade.
Griskevicius, V., Shiota, M.N., & Neufeld, S.L., (2010). Influence of different positive emotions on persuasion processing: A functional evolutionary approach. Emotion, 10 (2) 190-206.
Hills, P., & Argyle, M., (2002). The Oxford Happiness Questionnaire: A compact scale for the measurement of psychological well-being. Personality and Individual Differences, 33 (7) 1073-1082.
‘I love my ears’ campagne gestart om gehoorschade te voorkomen (2014). Retrieved March 19, 2015, from Rijksoverheid, http://www.rijksoverheid.nl/nieuws/2014/10/18/i-love-my-ears-campagne-gestart-om-gehoorschade-te-voorkomen.html.
Izard, C. E. (2007). Basic emotions, natural kinds, emotion schemas, and a new paradigm. Perspectives on Psychological Science, 2, 260-280.
Krugman, H. E. (1965). The impact of television advertising: Learning without involvement. Public Opinion Quarterly, 29, 349-356.
Krugman, H. E. (1972). Why three exposures may be enough. Journal of Advertising Research, 12 (6) 11-14.
Lang, P. J., Greenwald, M. K., Bradley, M. M., & Hamm, A. O. (1993). Looking at pictures: Affective, facial, visceral, and behavior reactions. Pscychophysiology, 30, 261-273. Lewis, M. (2011). Self-conscious emotions. Emotions, 742.
Lewis, I. M., Watson, B., White, K. M., & Tay, R., (2007). Promoting public health
messages: Should we move beyond fear-evoking appeals in road safety? Qualitative Health Research, 17 (1) 61-74.
Miceli, M., de Rosis, F., & Poggi, I., (2006). Emotional and non-emotional persuasion. Applied Artificial Intelligence: An International Journal, 20 (10) 849-879.
Mitterschiffthaler, M. T., Fu, C. H. Y. & Dalton, J. A., Andrew, C. M. & Williams, S. C. R. (2007). A functional MRI study of happy and sad affective states induced by classical music. Human Brain Mapping, 28, 1150-1162.
Nabi, R. L. (2002). Discrete emotions and persuasion. The persuasion handbook: Developments in theory and practice, 289-308.
Nabi, R.L. (2010). The case for emphasizing discrete emotions in communication research. Communication Monographs, 77 (2) 153-159.
Pham, M.T., (2007). Emotion and rationality: A critical review and interpretation of empirical evidence. Review of General Psychology, 11 (2) 155-178.
Price, V., & Tewksbury, D. (1996). Measuring the third-person effect of news: The impact of question order, contrast and knowledge. International Journal of Public Opinion Research, 8 (2) 120-141.
Roseman, I. J., Wiest, C., & Swartz, T. S., (1994). Phenomenology, behaviors, and goals differentiate discrete emotions. Journal of Personality and Social Psychology, 67 (2) 206-221.
O’Shaugnessy, J. & O’Shaughnessy, N.J. (2004). Persuasion in advertising. Routledge. Stephenson, M.T., Witte, K., Vaught, Ch., Quick, B.L., Booth-Butterfield, S., Patel, D., &
Zuckerman, C., (2005). Using persuasive messages to encourage voluntary hearing protection among coal miners. Journal of Safety Research, 36 (1) 9-17.
Strack, F. (1992). “Order effects” in survey research: Activation and information functions of preceding questions. Context Effects in Social and Psychological Research (pp. 22-34). Springer New York.
Tracy, J.L., & Robins, R.W., (2004). Show your pride: Evidence for a discrete emotion expression. Psychological Science, 15 (3) 194-197.
Tracy, J. L., & Robins, R. W., (2007). The psychological structure of pride: A tale of two facets. Journal of Personality and Social Psychology, 92 (3) 506-525.
Vogel, I., Brug, J., van der Ploeg, C.P.B., & Raat, H., (2007). Young people’s exposure to loud music: A summary of the literature. American Journal of Preventive Medicine, 33 (2) 124-133.
White, H. A., & Dillon, J. F., (2000). Knowledge about others’ reaction to a public service announcement: The impact on self persuasion and third-person perception. Journalism and Mass Communication Quarterly, 77 (4) 788-803.
Williams, L. A., & DeSteno, D., (2009). Pride. Adaptive social emotion or seventh sin? Psychological Science, 20 (3) 284-288.
Appendix A
Advertisements theme “listening to music”
Appendix B
Advertisements theme “dancing to music”
Appendix C
Advertisements theme “style/being cool”
