Listen to your heart! Vote following your feelings. A study towards the effectiveness of celebrity endorsed, negative emotional political advertisements in the Netherlands.
Master’s Thesis Political Communication, Anne Venema, 11200081
Master’s programme of Communication Science, Graduate School of Communication
Thesis supervisor: dr. Alessandro Nai
29th of June 2018
Abstract
Political parties apply various strategies to convince voters. Previous research established that the use of anxiety and anger in political advertisements can be an effective way to reach that goal. However, in times of political disinterest, it is a challenge for political parties to grab voters’ attention towards their messages. In this research it is hypothesised that celebrity endorsements grab the attention of voters, and as such, activate the emotions present in that political advertisement. This was tested by applying a multi-experimental research design that exposed participants to political advertisements of four Dutch political parties. These advertisements contained either anxiety or anger and celebrity endorsements. At first glance, there were no significant relationships discovered. Yet, after taking the political ideology of participants into consideration, this study did find some interesting results for the green-left party GroenLinks. For left-wing participants, anger combined with celebrity endorsements was found to be a good predictor of positive change in attitudes and propensity to vote for GroenLinks. Moreover, for right-wing participants the combination of anxiety with celebrity endorsements was found to be a good predictor of positive change in propensity to vote. Implications and directions for future research are discussed.
Keywords: Dutch politics, anxiety, anger, celebrity endorsements, political
Introduction
Political parties frequently build campaigns aimed at triggering negatively valenced emotions, they use these negative emotions as a strategical weapon to convince voters (Gross, 2008; Ridout & Searles, 2011). Already 54 years ago, former U.S. president Lyndon Johnson aired the classic ‘Daisy Girl’ advertisement that tapped into anxiety to convince American voters to tick his name on the ballot. In the 2016 presidential campaign, Democratic candidate Bernie Sanders also created a campaign directing citizens’ anger towards Wall Street to attract voters. Moreover, the political campaign of Trump made use of anger and anxiety targeted at immigrants (Lindley, 2016). A non-American and very recent example of negative emotions being used as a strategical weapon can be found in the Netherlands: the campaign video ‘Islam is terror’ (PVVpers, 2018) of the Dutch right wing populist party Partij voor de Vrijheid (PVV) aimed to evoke both anger and anxiety in the hope of convincing Dutch citizens to vote for the PVV. So, from practice it emerges that negative emotions such as anxiety and anger are frequently used by political parties to attract votes. As Searles and Ridout (2017) also state in their study: the decision to use negative emotional appeals in political campaigns appears to be a strategic one.
Brader (2006) mentions negative emotional appeals can indeed be found everywhere in contemporary political advertisements: they make the ad compelling by eliciting specific emotions and, in doing so, change the way viewers respond to them. There is an extensive body of research on the persuasive effects of anxiety and anger in political campaigns (Brader, 2005; Huddy, Feldman & Cassesse, 2007; Marcus, MacKuen & Neuman, 2011; Marcus, Neuman & MacKuen, 2000; Nai, Schemeil & Marie, 2017; Valentino, Hutchings, Banks & Davis 2008). All these studies found that both anxiety and anger have different persuasive effects. Anxiety is known to make individuals stop and reconsider the message, angry individuals are more likely to attack the message of a political advertisement and reinforce their own views.
However, for campaigns to have an effect, individuals need to process them. The processing of political advertisements only happens when they spark interest or if individuals are pushed to do so. As the numbers of Pew Research Center (2015) prove, this does not happen very often: less than 50% of U.S. citizens are interested in politics. Political
advertisements can, for some groups of people, still be very abstract despite the fact that some of them contain negative emotional appeals. When voters are not willing to pay attention to the advertisements, the messages in them will not be effective. Additionally, as news media are offering vast amounts of content, politically uninterested individuals can easily avoid
exposure to political news (Prior, 2010) and find other types of information in for example entertainment media. To overcome the problem of disinterested citizens, politics and entertainment have become increasingly intertwined in the past two decades (Jackson & Darrow, 2005).
One of the ways in which both fields are connected is through the use of celebrity endorsements in political advertisements. Celebrities can be utilised as a strategical weapon to persuade voters, they can be an inspiration for many young people and often act as reference group for decisions (Kamins, 1990). They have proven to be effective in product
advertisements (Berqvist & Zou, 2016; Elberse & Verleun, 2012; Rossiter & Smidts, 2012). This can partly be explained by the Elaboration Likelihood Model (ELM) of Cacioppo and Petty (1984). They state that simple source cues (e.g. celebrity endorsements) can be important determinants of persuasion. Moreover, Roskos-Ewoldson, Bichsel and Hoffman (2002) found that a likeable source is more persuasive than an unlikeable source.
As politics is often viewed as an unappealing practice, emotional or not, celebrity endorsements might be of help in this context. Coming from the entertainment industry, celebrities can use their fame to endorse a political party. With this endorsement, an
individual that normally would have ignored any type of political advertisement is triggered to give attention to the advertisement. Following, this attention makes room for the emotions anger and anxiety to come into play. And next, these emotions will affect the persuasiveness of a political advertisement. This study argues that using celebrity endorsements in political advertisements grabs the attention of voters, and as such activates the emotions present in that advertisement.
This multi-experimental study aims to improve our understanding of the
persuasiveness of political advertisements that elicit anxiety and anger in combination with celebrity endorsements. It intends to answer the following research question:
To what extent does the presence of celebrity endorsements in political advertisements moderate the effects of anxiety or anger appeals on evaluations of the sponsor?
Using a multi-experimental research design, this question will be answered by testing the persuasive effects of various political advertisements for several Dutch political parties. As suggested by the research question, this study will only look at the effects of positive campaigns: campaigns that are created to promote the sponsor.
Effective and persuasive political campaigns are those campaigns that attract the most votes, but when do citizens decide to vote for a political party? The effectiveness of political advertisements is defined in this study on the basis of three variables: the attitude of an individual towards the sponsor, their evaluation of that party’s competence to solve (CTS) the issues that exist in a country and their propensity to vote (PTV) for that party.
This experiment has been conducted in the Dutch context. This is done because this is relatively unexplored territory. Most research on negative emotions and celebrity
endorsements has namely been conducted in a U.S. context. It would be interesting to gain more insights into if existing conclusions hold in the Netherlands and if endorsements
moderate the effects of anxiety and anger in this context. Moreover, based on intuition, Dutch political campaigns are, contrasted to U.S. politics, less emotional. The argument of Van Kleef (2012) is interesting in this light: he argued that Dutch citizens are more tolerant towards emotional politics than 25 years ago. Subsequently, a political party or leader that uses emotions, has a potential electoral advantage. A similar situation emerges concerning celebrity endorsements: celebrities that endorse political parties are not unusual in the U.S (Garthwaite & Moore, 2012). However, they are hard to find in the Netherlands
(Niemantsverdriet, 2011).
So, the goal of this study is to increase understanding on the persuasive effectiveness of anxiety, anger, celebrity endorsements and their moderating role in the so far unexplored Dutch political environment. Next to expanding knowledge on existing literature, this study also has practical implications: its insights can be taken into account by Dutch political parties when designing their political advertisements.
Theoretical background
Emotions and effective political advertisements
When a threat emerges in our environment, we experience mental and physical changes associated with for example anxiety or anger. These changes are often experienced before we are aware of it (Brader, 2005; Damasio, 2000). Similarly to Brader (2005), this article will refer to emotions as underlying responses to the perceived significance of external stimuli. Emotional appeals are communications aimed to elicit an emotional response from all who receive them.
But when do emotions become political? Marcus states (2002) that emotions are key in making political decisions, even for those who intentionally attempt to be rational. In an earlier study Marcus, Neuman and MacKuen (2000) stated that anger, anxiety and enthusiasm
are the three key emotions that determine the strategies that citizens use to construct their political views. An often-utilised theory that explains the link between emotions and political decision making is the Affective Intelligence Theory (AIT) of Marcus, Neuman and
MacKuen (2000). This theory proposes that individuals’ emotions help govern deliberation and attention to new political information and help managing their attention to politics (Marcus, MacKuen & Neuman 2011).
Following a neuroscientific approach, this theory states that people have two distinct emotional systems: a dispositional and a surveillance system. Both systems are responsible for how we react to routine and new situations. The dispositional system monitors habits and allows us to perform tasks without consciously considering them (Redlawsk, Civettini & Lau, 2007). So, when expectations of individuals are not challenged by novel information, they can rely on habitual responses to identified stimuli. The corresponding emotion to the
dispositional system is enthusiasm: when expectations are met, enthusiasm is generated. The surveillance system is operating when an individual encounters unexpected and novel information. As the habitual response is no longer relevant, conscious information processing takes over, resulting in longer and more careful consideration of the new information. This type of information processing is often triggered by anxiety (Redlawsk, Civettini & Lau, 2007). As Redlawsk and colleagues state: conscious awareness is reserved for when our expectations are violated, the surveillance systems functions in such a way that ‘needless waste of mental resources is prevented’. This means that we are often more
conscious of our emotions when they are negative. Thus, because novel information generates conscious emotional information processing through the activation of the surveillance system, negative emotions can be related to active learning. Huddy, Feldman & Cassesse (2007) found that anger has a similar influence as anxiety: this emotion leads to higher levels of involvement and attention to political messages. So, when something is detected as wrong in their surroundings through anger or anxiety, an individual will stop and reconsider their current assessment by taking a closer look at the situation. This state often opens the gate for persuasion, meaning that this is the state in which political strategists should be interested in.
Thus, according to the AIT, anger and anxiety are both emotions that have an influence on the effectiveness of a political message. However, as will be discussed in the following section, both emotions have distinctly different effects.
As mentioned earlier, the effectiveness of political advertisements will be defined with the aid of three concepts: the attitude towards the sponsor of the message, the evaluation of the party’s competence to solve a specific issue and the propensity to vote for that same
sponsor. More on these concepts and their definitions can be found in the Method section of this paper. Before moving to that section, it is first important to know what the different persuasive effects of anger and anxiety are.
Anxiety, anger and effective political advertisements
Thus, in situations of uncertainty, individuals experience anxiety (Lee & Choi, 2018). The results of the study of Brader, Valentino and Suhay (2008) proved that anxiety is critical in changing individuals’ attitudes and behaviours. Rigterink and Schomerus (2017) state that anxiety helps individuals forming political attitudes closer to their own beliefs. Moreover, MacKuen, Wolak, Keele and Marcus (2010) found that anxiety can be significantly related to an individual’s willingness to seek new perspectives and become more open to compromise. The study of Gadarian and Albertson (2014) indicated that anxious individuals exhibit biased information processing: they read, remember and agree with threatening information.
Additionally, the results of the study of Nai, Schemeil & Marie (2017) indicated that anxiety has the potential to decrease resistance to persuasion.
There are also multiple studies that researched the effects of anger. When individuals are confronted with stimuli in their environment that cause discomfort or pain they often feel aversion including feelings of anger. MacKuen et al. (2010) state that anger signals the need to confront the adversary by relying on previously learned routines. Often, angry individuals ignore uncomfortable, counter-attitudinal information (Valentino et al., 2008) and reinforce their own views seeking for conforming information (MacKuen et al., 2010). The findings of Redlawsk, Civettini & Lau (2007) add to that by suggesting that angry individuals have a decreased motivation to be open to and seek out new information. Similarly, MacKuen et al. (2010) find that anger reduces willingness to consider new information and increases
commitment to previous held attitudes. Huddy, Feldman and Cassesse (2007) concluded that anger is associated with rapid decision making.
The findings of these studies suggest that political advertisements that elicit anger have a different influence on its effectiveness when compared to those that evoke anxiety. Petersen (2010) directly contrasts the two emotions by suggesting that they are very dissimilar and that they ‘operate in distinct domains of political life’. Anxiety is triggered in uncertain circumstances, whereas anger is triggered in certain circumstances. Additionally, Lerner and Keltner (2001) found that both emotions have different effects on risk perception.
After reading the literature it becomes apparent that anxious individuals are more likely to stop and reconsider the message of a political advertisement, angry individuals are more likely to reinforce their own views and being closed to the new information in an
advertisement. Both emotions lead to different political outcomes (Parker & Isbell, 2010). A clear example of that is the study conducted by Garry (2014) on the EU referendum held in Ireland: he found that anxious voters are more likely to support the treaty than angry voters. This leads to the first hypothesis:
H1: Political advertisements that elicit anxiety lead to more positive attitudes towards the sponsor, a more favourable evaluation of the party’s competence to solve an issue and a more favourable voting propensity when compared to political advertisements that elicit anger.
Yet, as already briefly mentioned in the introduction, politics can be for some groups of people an uninteresting practice. Debating about and creating policies or laws is something that is not always appealing to every individual. This is proved by the numbers of Pew
Research Center (2015), the gender gap in political interest (Verba, Burns & Schlozman, 1997) and differences in political interest between the socially and economically advantaged and disadvantaged in society (Oskarson, 2007). Even when political parties use anxiety and anger in their advertisements, they will not be effective when an individual is not paying attention to them in the first place. In order to make the politically uninterested interested again, many political parties have tried to combine politics with a field that is appealing to more citizens: the entertainment industry. Jackson & Darrow (2005) confirmed this with their statement that politics and entertainment have become increasingly intertwined in the past two decades. Endorsements from celebrities from the entertainment industry might be of help in sparking political interest in politically uninterested individuals.
Celebrity endorsements and effective political advertisements
The strategical use of famous individuals from the entertainment industry, celebrity endorsements, by political parties is one of the ways in which entertainment and politics are connected. As Bennet (2012) also signals, politics is becoming more personalised. Celebrities are used by political parties because they can be an inspiration for many young people and act as a reference group for decisions (Kamins, 1990). They also convey the image values
associated with them to the political party they endorse (Wood & Herbst, 2007). So, for example, the positive image values of a celebrity that is inspiring and motivational to many people, will be conveyed to the political party.
The use of celebrity endorsements is common practice in U.S. politics (Garthwaite & Moore, 2012), one would therefore expect that much scholars have been dedicating their time
to this subject. However, scholars mostly have been focussing on the impact of endorsements on products or services (Berqvist & Zou, 2016; Elberse and Verleun, 2012; Rossiter and Smidts, 2012), not on its impact on politics. Erdogan (1999) states in his literature review on celebrity endorsements for commercial products that ‘the celebrity endorsement strategy can be an effective competitive weapon in mature and saturated markets in order to differentiate products from the products of competitors, as long as the “right” celebrity is found’. Butler, Cowan and Nilsson (2005) showed that celebrity endorsements can indeed be effective in this sector: Oprah Winfrey’s endorsement of a book improved its position on the best-sellers list.
Turning to politics, Wood and Herbst (2007) found that peers were more influential than celebrities in gathering support for a political candidate or party under young individuals. Nonetheless, Jackson and Darrow (2005) established that celebrities make unpopular
statements more agreeable, while increasing the level with already popular opinions. These effects were the strongest for young individuals. Garthwaite & Moore (2013) discovered that Oprah Winfrey’s endorsement had statistically and politically significant effects on Barack Obama’s political outcomes: it increased his share of votes and received campaign
contributions. Also, celebrity endorsements are used by political parties because they draw attention (Wood & Herbst, 2007). Based on these previous findings, the second hypothesis of this study is:
H2: Political advertisements with celebrity endorsements lead to more positive attitudes towards the advertisement, more favourable evaluation of the party’s competence to solve an issue and a more favourable voting propensity than political advertisements without celebrity endorsements.
Anxiety, anger and the moderating role of celebrity endorsements
As previously mentioned, due to politics being an unappealing practice to many citizens, they might choose to ignore all political information. However, as indicated by Wood & Herbst (2007), celebrity endorsements draw attention to the messages of a political party. And next, attention leads to more effortful information processing (Borcherding, Thompson, Kruesi, Bartko, Rapoport, Weingartner (1988). The reasons for why celebrity endorsements might be effective build on the dual-route information processing theory of the Elaboration Likelihood Model (ELM) of Cacioppo and Petty (1984). This model states that individuals can process new information either in an effortful and elaborative way, via the
central route, or in a less cognitive demanding and low elaborative way, via the peripheral route. Both routes can result in attitude change.
The heuristic-systematic model (HSM) of Chaiken (1980) adds to the ELM theory by distinguishing between a systematic and heuristic view on persuasion. According to the systematic view, individuals employ considerable cognitive effort in processing the
information. In contrast, the heuristic view of persuasion entails comparatively little effort in processing the message. About the heuristic view Chaiken (1980) argues that ‘rather than processing argumentation, recipients may rely on (typically) more accessible information such as the source’s identity or other non-content cues in deciding to accept a message’s conclusion’. In this statement, the endorsement of a celebrity can be shared under the label of ‘non-content cues’, thus, they might be effective when individuals process the political advertisements via the peripheral (ELM) or heuristic (HSM) route. Furthermore, Roskos-Ewoldsen, Bichsel and Hoffman (2002) found that when an individual is centrally processing a message with a source that he or she likes, that individual may also perceive the argument as more persuasive when compared to a message with an unlikeable source. This indicates that celebrity endorsements might be useful for both routes of persuasion.
The findings of Veer, Becirovic and Martin (2010) are especially interesting in the light of politically uninterested citizens: they found that celebrity endorsements have a significant impact on those individuals that do not actively engage with politics. So, when political parties make use of celebrity endorsements, an individual that normally would have ignored any type of political advertisement of any political party is triggered to give attention to the advertisement. Next, the emotions that are present in the advertisement can do their persuasive work. This research argues that the use of celebrity endorsements in political advertisements grabs the attention of voters, and as such activates the emotions present in that advertisement. They are expected to moderate the role of negative emotions on the
effectiveness of political advertisements. This leads to the third hypothesis:
H3: The effects of anxiety and anger will be especially strong in political advertisements that contain celebrity endorsements.
Political ideology
The effectiveness of political advertisements also depends on the individual that is processing the information. Taber & Lodge (2006) concluded their study that citizens are biased information processors making use of motivated reasoning. This means that
individuals, due to their previous beliefs, are evaluating attitudinally congruent arguments stronger than attitudinally incongruent arguments. The results of Kahan (2012) add to that by suggesting that motivated reasoning is symmetric across groups of opposing political
ideologies. Thus, political ideology might influence the effectiveness of anger, anxiety and celebrity endorsements in political advertisements. Adding an extra layer to the previous hypotheses, the final hypothesis of this study is:
H4: The effectiveness of anxiety, anger and celebrity endorsements in political advertisements depends on individuals’ political ideology.
Methods
Protocol
Before discussing the operationalisation of the variables and treatments, it is helpful to shortly explain the logic of this study. This research has tested all hypotheses by
administering four different experiments, one for each of the following Dutch political parties: the conservative-liberal Volkspartij voor Vrijheid en Democratie (VVD), the left-green GroenLinks party, the social-democratic Partij van de Arbeid (PvdA) and the
nationalistic right-wing Partij voor de Vrijheid (PVV). These parties were selected since they all have a clear stance on issues that are present in Dutch society. The VVD has been chosen due to this party having a clear opinion about security, the PvdA has been selected because this party has a clear viewpoint on healthcare, GroenLinks has been chosen as this party has a clear standpoint on the environment and the PVV has been selected as this party has a clear opinion about immigration. These issues are based on the set of policy issues the parties own: they have, according to Van der Brug (2004), a relatively good reputation on these issues. Moreover, all these issues and parties were selected having in mind that they needed to result in realistic stimuli, which increases the external and ecological validity of this study.
All participants took, mainly for practical reasons, part in all four experiments in a fixed order. The survey started with the PvdA experiment, followed by the VVD experiment, then the GroenLinks experiment and ending with the PVV experiment. In each of the
experiments the participants were randomly exposed to a political poster dealing with the issue owned by each party. So, all participants were randomly exposed to four political posters, that contained one of the six combinations of emotions (anxiety, anger, absent) and celebrity endorsements (present, absent). More information on how the posters and conditions
were created can be found in the following sections. The logic and order of all experiments is schematically presented in figure 1.
Figure 1. Schematic representation of the fixed order of all four experiments
Participants
All 218 participants (96 men, 122 women) have been recruited through convenience and snowball sampling starting from the network of friends, colleagues and family members of the author of this study. The survey has been distributed in the first week of May 2018 via e-mail, the authors Facebook profile and the messenger service WhatsApp. After the first distribution, three reminders, one per week, have been sent via the same communication channels. This strategy was applied to obtain a sample as similar as possible to the actual population.
Participants’ year of birth varied from 1938 to 1999, the mean age was 39.56 (SD = 18.46) years. The highest level of education they completed varied from high school to postgraduate doctoral degree, participants mean in highest education completed was
university bachelor. The sample was slightly leaning to the left side of the political spectrum, as the mean (M = 4.29, SD = 1.66) of all participants’ self-placement on the political left-right scale indicated. This scale varied from 1 (‘Extremely left’) to 10 (‘Extremely right’).
Participation in this study was on a voluntary basis. In order to participate all
participants needed to give consent and be eligible to vote in the Netherlands. Moreover, they also had to know Dutch fluently in order to understand and participate as this was the
language in which the survey has been written. Participants who failed to finish the survey (n = 70) were excluded from further analysis, making the final sample consist out of 218
participants.
Research design
This study has made use of a sequential series of between-subjects designed experiments. Experiments permit strong causal inference and, as Brader (2005) mentions, they are particularly appropriate for the study of emotion in political communication (Glaser
Experiment 1: PvdA Experiment 2: VVD Experiment 3: GroenLinks Experiment 4: PVV
and Salovey, 1998; Isbell and Ottati, 2002). Using an experiment as research method ensures high internal validity (Baxter & Babbie, 2003), as changes in all dependent variables are most likely the result of exposure to the stimuli.
So, the specific design of this study has been a multi-experimental 3 x 2 factorial design. The first factor was the presence of a specific negative emotion with three levels (anxiety, anger or absent), the second factor was the presence of a celebrity endorsement with two levels (present or absent). This design results in two experimental conditions and six experimental stimuli, which can all be found in table 1. Stimulus six functioned as the control group for both emotion and endorsement conditions simultaneously.
Table 1
The experimental conditions and stimuli
Celebrity endorsement
Negative emotion Present Absent
Anxiety Stimulus 1 Stimulus 4
Anger Stimulus 2 Stimulus 5
Absent Stimulus 3 Stimulus 6
All participants took part in all four experiments, within each of the four experiments all participants were randomly exposed to one of stimuli described above.
All experiments were conducted online via one internet survey created with the survey software tool Qualtrics. This method has been selected as it is both time and cost effective. The survey could be filled in from any location with internet access, this improved the ecological validity as participants filled in the survey at their own chosen time and place (Baxter & Babbie, 2003).
Treatments
All six stimuli were operationalised in fictional campaign posters, created for four different political parties. All posters contained the logo of the sponsor, a call to vote for the sponsor on the 12th of June (again a fictional date) and a sentence that made clear what the goal of the poster was. A benefit from the decision to create fictional posters is that all elements present in the posters were completely controlled. In the process of designing each campaign poster, care has been taken to make them as similar as possible to diminish chances
of other factors than the treatment itself being responsible for changes in the dependent variables.
Anxiety and anger
The emotional part of the treatment has been set up as a series of issue-related images, in line with the issue owned by the political parties on which the experiment is run. One image was presented per poster. To make sure all images used in the posters evoked either anxiety or anger, all of them were pre-tested. A small survey created in Qualtrics was
distributed to the peers of the author in which they were exposed to multiple images. After the exposure they needed to indicate which emotions they experienced while observing the image. This was done on a 7-point Likert scale ranging from 1 (‘Did not experience at all’) to 7 (‘Experienced a lot’) After the successful pre-test (N = 24), the images that evoked the strongest emotional reactions were selected for the actual treatments. Moreover, following the advice of Albertson and Gadarian (2016), to make sure that all participants were in an anxious or angry cognitive state all of them were asked to think about their worries concerning the issues of each of the posters before they were exposed to it. For example, before being
exposed to one of the six stimuli in the GroenLinks experiment, the participants were asked to think about their worries concerning the environment.
After the exposure to each of the campaign posters in the experiment, participants were asked to indicate on a 7-point Likert scale which emotion they experienced while observing the poster. To determine if the manipulation of all the treatments was successful, eight one-way ANOVAs were performed, two per political party. The mean emotion scores on anxiety and anger experienced in the anxiety, anger and control condition were compared with each other. The manipulation checks were successful for all four political parties, all compared means differed significantly from each other. Details on these ANOVAs can be found in appendix A.
Celebrity endorsements
All conditions that contained celebrity endorsements were created by providing the names of six politically neutral Dutch celebrities on the campaign posters. The decision to provide multiple names is guided by the reasoning that every participant will recognise and identify with at least one of those celebrities. The names of these celebrities were pre-tested in the same survey that was created for pre-testing the emotional images. In this pre-test each of the participants was presented a list of names of Dutch celebrities and was asked which of the celebrities they knew. Only the celebrities that were known by all participants of the pre-test were used in the actual treatments. After developing a list consisting out of 24 celebrities,
each celebrity was, based on intuition, assigned to a political party that had the most appropriate fit to their personality. For example, the news host Eva Jinek was placed at the PvdA as her views are expected to be less extreme than those of comedian Ruben Nicolai, who was placed on the PVV poster. This was done to create treatments that were as realistic as possible. All 24 posters can be found in appendix B, the list of selected celebrities and their details can be found in appendix C.
Variables
The dependent variable, attitude towards the sponsor, has been operationalised based on the 2016 measure of the American National Election Survey (ANES) to measure attitudes: a feeling thermometer. This measure has been slightly adapted: all participants were asked to indicate their feelings towards all four political parties on a thermometer ranging from 0 to 10. The question was asked before and after being exposed to the treatments.
The second dependent variable, the evaluation of a party’s competence to solve a specific issue has been measured with the aid of the scale that has previously been used by Bélanger & Meguid (2008). All participants were asked to indicate on a 7-point Likert scale ranging from ‘Extremely incompetent’ (1) to ‘Extremely competent’ (7) how competent they evaluated each of the four political parties in solving the issues that were assigned to them. This measure was designed as four batteries of questions that were asked before participants were exposed to the treatments, one per issue. After the exposure in each experiment, the measure was asked again, but only based on the issue that they just have seen on the poster.
The third dependent variable, propensity to vote for the sponsor was operationalised using a slightly adapted version of the PTV scale of the ANES (2016). Participants were asked to indicate on a 7-point Likert scale ranging from ‘Not at all likely’ (1) to ‘Very likely’ (7) how high their propensity to vote for a specific party was. This PTV-measure has been provided in a similar way as the CTS-measure. Firstly, before participants were exposed to the treatment it was provided as a battery of questions, one item per political party. After the exposure the PTV-measure was provided again, but only for one party: the party in which experiment they were taking part. The full list of measures can be found appendix D.
Next, based on the previously mentioned political left-right scale, participants were divided into three ideology groups. All participants that indicated 1 up to 3 were defined as left-wing, 4 ranging up to 6 was defined as moderate and participants that indicated 7 or higher were defined as right-wing.
Before starting the analysis, the variation in answers of all the participants was
captured in order to see if they changed their views due to the exposure to the treatments. This variation was determined by subtracting the pre-test scores from the post-test scores. After inspecting the frequencies of the newly calculated variables it became apparent that the variation in the answers of the participants between the pre and post-exposure measurements of every dependent variable was indeed not very big, stressing the importance of capturing this little variation.
Results
So, the effectiveness of a political advertisement is defined on the basis of three dependent variables: attitudes towards the sponsor, the evaluation of that sponsors’ CTS issues and the PTV for the sponsor of the message. The results will be structured per hypothesis, per political party and these three dependent variables.
Anxiety versus Anger
First it was tested if anxiety and anger have significantly different effects on all the dependent variables. Based on previous research it was hypothesised that anxiety would result in stronger positive changes in all dependent variables than anger. Independent samples t-tests were run with the exposure to an anxiety or anger evoking poster as categorical independent variable and either attitudes, evaluation of CTS or PTV as continuous dependent variable. This was done for all four political parties, resulting in 12 t-tests (3 dependent variables × 4 political parties). None of those tests showed significant differences. Differently put: those who were exposed to an anxiety eliciting poster did not show significantly different change in attitudes towards the sponsor, evaluation of that sponsors’ competence to solve issues and the propensity to vote for the sponsor than those who were exposed to anger. Consequently, H1 was rejected.
Celebrity endorsements
The second hypothesis stated that political advertisements with celebrity endorsements have stronger positive effects on all dependent variables than political advertisements that do not contain celebrity endorsements. Again, 12 independent sample t-tests (3 dependent variables × 4 political parties) were run with exposure to a poster with or without a celebrity endorsement as categorical independent variable and either attitudes, CTS or PTV as
VVD’s competence to solve issues with safety was marginally significant: t (216) = 1.98, p = .049. The mean for posters that contained an endorsement (M = -.74, SD = 1.01) was lower than the one for the poster that did not contain an endorsement (M = -.46, SD = 1.07). In other words, those who were exposed to the poster with endorsements lowered their evaluation of the VVD’s competence to solve issues with safety more than those who were exposed to the poster without endorsements. Regardless of political party, all other t-tests were insignificant for each of the dependent variables. So, with the exception of the VVD, the exposure to a political advertisement with endorsements did not cause significant changes in attitudes, CTS or PTV when compared to exposure to a political advertisement without endorsements. Hence, H2 was also rejected.
Moderation
H3 stated that celebrity endorsements moderate the effectiveness of anger and anxiety in political advertisements. The categorical independent variable in this hypothesis is the exposure to either an anxiety or anger eliciting poster, the categorical moderating variable is the exposure to a poster that either does or does not contain endorsements, and the continuous dependent variables are again attitude towards the sponsor, CTS and PTV. For all political parties and on all dependent variables multiple regressions with interactions were run,
resulting again in a total of 12 tests (3 dependent variables × 4 political parties). For example, a multiple regression was run to predict if change in attitudes towards the PvdA could be due to exposure to a poster with either anxiety or anger and with or without endorsement. The same was done with the other two dependent variables: CTS and PTV. Before starting the analysis, both the independent and moderating variables were recoded into 0-1 discrete variables, so they could be used in the analysis.
All the regressions that were performed, on every dependent variable for every political party, were statistically insignificant. This indicated that there was no interaction between anxiety, anger and the presence or absence of celebrity endorsements in predicting change in the participants’ attitudes towards the sponsor of the poster, their evaluation of that sponsors competence to solve issues and their voting propensity. Therefore, H3 was also rejected.
Political ideology
The fourth hypothesis stated that political ideology affects the effectiveness of anxiety, anger and celebrity endorsements. A new model was built, comprised of both continuous
emotion and endorsement variables, together with one of the three political ideology variables. For all ideology groups, all political parties and all dependent variables, multiple regressions were run. This resulted in a total of 36 regressions (3 ideology groups × 4 political parties × 3 dependent variables). For GroenLinks, four of these regressions resulted in
significant outcomes. The model that included all left-wing participants was marginally significant in predicting their change in propensity to vote for GroenLinks, F (7, 136) = 1.79, p = .095. Figure 2 displays the interaction effect of this model. From the graph it can be read that if GroenLinks wants to attract votes from left-wing participants they should use political advertisements that elicit anxiety but without celebrity endorsements. If GroenLinks uses anxiety in combination with an endorsement there is a higher probability that the change in propensity to vote will be negative. The effects for anger are reversed but also smaller: GroenLinks will have a higher chance on attracting votes using anger in combination with celebrity endorsements. However, the chances will decrease when this emotion is used without endorsement. A table with all the coefficients, their significance and the effect size of this regression model can be found in appendix E.
Figure 2. Direction of interaction effect on PTV for left-wing participants
Next, the model that predicted attitude change for all left-wing participants was not significant, F(7, 136) = 1.39, p = .22. However, the coefficient of the moderated interaction in the model was significant, p = .042. The interaction effect is visualised in figure 3. From there it can be read that if GroenLinks wants to cause positive change in attitudes for left-wing participants they should use anger in combination with celebrity endorsements. The opposite happens when they use anger without endorsements, then the effect will be negative. The difference in effects for anxiety are both negative and differ hardly from each other.
No endorsement Endorsement -1 -0,8 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1 Anxiety Anger Change in propensi ty to vot e
A table with all the coefficients, their significance and the effect size of this regression model can also be found in appendix E.
Figure 3. Direction of interaction effect on change in attitude for left-wing participants
The next model predicted if right-wing participants changed their evaluation of GroenLinks’ competence to solve environmental issues. This model was also not significant, F(7, 136) = .95, p = .47. However, the coefficient of the moderated interaction in the model was significant, p = .033. This interaction is plotted in figure 4. There it can be read that both emotions and endorsements cause the evaluation of Groenlinks’ competence to solve
environmental issues to drop. This drop is especially high after seeing the political advertisements that only used anxiety. Again, a table with all the coefficients, their significance and the effect size of this regression model can be found in appendix E.
Figure 4. Direction of interaction effect on change in CTS for right-wing participants
No endorsement Endorsement -1 -0,8 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1 Anxiety Anger Change in at tit udes No endorsement Endorsement -2 -1,6 -1,2 -0,8 -0,4 0 0,4 0,8 1,2 1,6 2 Anxiety Anger Change in com pet ence to sol ve
Lastly, the model that predicted the PTV of right-wing participants for GroenLinks was also not significant, F(7, 136) = 1.36, p = .23. However, the coefficient of the moderated interaction in the model was marginally significant, p = .054. This interaction has been plotted in figure 5. From this figure it can be read that if GroenLinks aims to attract votes from right-wing participants they may benefit from using political advertisements that elicit anxiety and include celebrity endorsements. However, when they use anger in combination with
endorsements the effects are negative. The effects for political advertisements without endorsements are much smaller: GroenLinks will have a slightly higher change in attracting votes when using this in combination with anger. A table with all the coefficients, their significance and the effect size of this regression model can be found in appendix E.
Figure 5. Direction of interaction effect on PTV for right-wing participants
For all other political parties, the PvdA, VVD and PVV, all of the tests ran for every ideology group and each of the dependent variables were insignificant. H4 is therefore partly confirmed, as some parts of the models of GroenLinks were significant.
Discussion
This study examined the effectiveness of the use of anxiety, anger, celebrity
endorsements and their combination in Dutch political advertisements for four Dutch political parties. Most of the hypothesised relationships could not be confirmed by the statistical analyses. For all political parties, no difference was found in the effectiveness of either anxiety or anger in political advertisements. There was also no difference found in the effectiveness for political advertisements that do or do not contain celebrity endorsements. However, there was one marginally significant finding concerning the VVD, their
No endorsement Endorsement -1 -0,8 -0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8 1 Anxiety Anger Change in propensi ty to vot e
competence to solve issues with safety and the use of celebrity endorsements. When the liberals use celebrity endorsements it is likely that individuals’ perception of this party to be competent in solving safety issues is negatively affected. The VVD should thus be cautious using celebrity endorsements. Moreover, the moderation hypothesis could also not be confirmed.
So, to iterate the research question: did celebrity endorsements moderate the effects of anxiety and anger in political advertisements? No. Yet, after taking the political ideology of participants into account this study did discover some interesting evidence of the effectiveness of negative emotions in combination with celebrity endorsements for one party: GroenLinks. Although the overall models of GroenLinks were not always significant, significance was found for the parameters in those models. Based on these sometimes marginally significant parameters, the following can be concluded: depending on the goal of the political campaign of this party, negative emotions can be effective in combination with celebrity endorsements. When GroenLinks aims to convince left-wing voters, they can best use anger in combination with endorsements. When doing this they have the highest chance that voters’ attitudes about the party will be positively affected and it also increases the chances that these voters will vote for GroenLinks. These findings make sense in the light of the theory of motivated
reasoning from Taber & Lodge (2006): all left-participants reinforced their own views as they agreed with the attitude congruent information of GroenLinks. Moreover, it also is in line with the results of MacKuen et al. (2010): angry participants reinforced their own views.
Another interesting finding is that if GroenLinks wants to attract right-wing votes they should use anxiety in combination with celebrity endorsements. These findings make sense in the light of the studies of Gadarian and Albertson (2014): anxious individuals agree more with threating information. Also, they are in line with the results of the study of Nai, Schemeil & Marie (2017): anxiety decreased resistance to persuasion. However, they are contrary to what one would expect according to the motivated reasoning theory (Taber & Lodge, 2006). Following this theory, one would expect that right-wing voters would evaluate GroenLinks more negatively instead of more positively after being exposed to the political advertisements. This might be an indication on how anxiety can overthrow the effects of motivated reasoning theory, stressing the persuasive strength of this emotion.
All of the other combinations in the significant models negatively affected the effectiveness of the political advertisements of GroenLinks. Furthermore, taking political ideology into consideration did not yield any significant models that could predict the
effectiveness of negative emotions and celebrity endorsements in political advertisements for the PvdA, VVD and the PVV.
The implications of this study are highly practical for two Dutch political parties: GroenLinks and the VVD. Concerning the use of emotions and celebrity endorsements both parties now have an indication of what steps to take (and which not) to increase the
effectiveness of their political advertisements. Furthermore, scientific understanding of the persuasive effectiveness of both negative emotions and celebrity endorsements in political advertisements has increased. Apparently, as indicated by many insignificant findings, the mechanisms behind the constructs used in this study worked differently than in previous studies.
A reason for many of the insignificant findings might be that most of the previous results were based on the U.S. political system and culture. This two-party system differs from the multiparty system in the Netherlands. Abramowitz and Saunders (2005) established that U.S. politics is highly polarised, which is confirmed by the numbers of PEW (2014). Polarised views are extreme views, and extreme views often come with extreme emotions (Smith, Seger & Mackie, 2007). So, because politics in the U.S. is more emotional, emotions in political advertisements are expected to be more effective in that context when compared to the Dutch context. Moreover, U.S. culture is different from Dutch culture. Both cultures are western, but they score differently on the dimensions of Hofstede’s (2003) model of culture. This might have caused the differences in results of this study when compared to previous studies.
Concerning the insignificance in results for endorsements, Jackson and Darrow (2005) established that celebrity endorsements had the strongest effects on young individuals.
Despite that there were some young participants in this study, the mean age was 39 years, indicating that the sample of this study was generally not very young. Moreover, Veer, Becirovic and Martin (2010) established that when an individual is actively engaged with politics the endorsement of a celebrity will be negated. As the engagement of this sample with politics was unknown, this could also potentially have influenced the results.
All participants took part in a fixed order of experiments. This might have caused order-effects bias in the results. Moreover, as all participants needed to go through all experiments to complete the survey they might have experienced it as repetitive. In times of survey fatigue (Bryman, 2016), this also might have affected the results. Furthermore, because participants could choose freely where and when they would fill in the survey, the circumstances were realistic but also not completely controlled. Affective Intelligence theory
is about emotional responses to individual pieces of information and the broader context in which this information is situated is crucially important (Pantoja & Segura, 2003). So, completing this survey after viewing an emotional movie, on the train home or after a couple of beers could all have influenced the emotions experienced while filling in the survey and thus the results.
Additionally, despite that this study has pre-tested the stimuli and incorporated the advice of Albertson and Gadarian (2016) it becomes clear from the manipulation checks that the levels of anxiety and anger experienced by participants was relatively low. When these levels were higher the results of this study were probably different.
Regarding the use of celebrity endorsements, the decision to provide the names of multiple Dutch celebrities on the political advertisements increased the chance that an
individual liked and identified with at least one of those celebrities. However, it also increased the chance that an individual disliked and got irritated by one of the celebrities.
Nonetheless, this was the first time that negative emotions and celebrity endorsements were combined in a study towards the effectiveness of political advertisements. Due to the multi-experimental design of this study it was possible to provide detailed information about the effectiveness of emotions and endorsements for various political parties, all relatively cost and time effective and with high internal validity (Baxter & Babbie, 2003). The design is very suitable to use in other contexts: in other countries, with other political parties, other emotions and other dependent variables. Hence, it allows other researchers to explore many
relationships in the field of emotions, celebrity endorsements and political advertising. Future research should make use of this opportunity, starting with investigating if the findings of this study could be replicated with a higher sample size. Also, the focus of future research should be at taking contextual factors such as the mood of country, shifts in public opinion or timing after elections into account. Moreover, focus should also be on exploring the different effects of anxiety, anger and celebrity endorsements for specific groups in society.
The main conclusion of this study is that for some political parties, and for some groups of people, anxiety, anger combined with celebrity endorsements increase the persuasiveness of political advertisements. Your political ideology, the education that you have received, your religion or your gender: the effectiveness of negative emotions and celebrity endorsements in political advertisements depends on who you are.
References
Abramowitz, A., & Saunders, K. (2005). Why can't we all just get along? The reality of a polarized America. In The Forum (Vol. 3, No. 2). De Gruyter. doi:10.1.1.557.5928 Albertson, B., & Gadarian, S. K. (2016). Did that scare you? Tips on creating emotion in
experimental subjects. Political Analysis, 24(4), 485-491. doi:10.1093/pan/mpw022 American National Election Survey, ANES (2016). 2016 Time Series Study. Retrieved from:
http://www.electionstudies.org/studypages/anes_timeseries_2016/anes_timeseries_201 6.htm
Baxter, L. A., & Babbie, E. R. (2003). The basics of communication research. Cengage Learning.
Bennett, W. L. (2012). The personalization of politics: Political identity, social media, and changing patterns of participation. The ANNALS of the American Academy of Political and Social Science, 644(1), 20-39. doi:10.1177/0002716212451428
Bergkvist, L., & Zhou, K. Q. (2016). Celebrity endorsements: a literature review and research agenda. International Journal of Advertising, 35(4), 642-663. doi:
10.1080/02650487.2015.1137537
Bélanger, É., & Meguid, B. M. (2008). Issue salience, issue ownership, and issue-based vote choice. Electoral Studies, 27(3), 477-491. doi:10.1016/j.electstud.2008.01.001 Borcherding, B., Thompson, K., Kruesi, M., Bartko, J., Rapoport, J. L., & Weingartner, H.
(1988). Automatic and effortful processing in attention deficit/hyperactivity disorder. Journal of Abnormal Child Psychology, 16(3), 333-345. Retrieved from: https://link.springer.com/content/pdf/10.1007%2FBF00913804.pdf
Brader, T. (2005). Striking a responsive chord: How political ads motivate and persuade voters by appealing to emotions. American Journal of Political Science, 49(2), 388-405. doi: 10.2307/3647684
Brader, T. (2006). Campaigning for hearts and minds: How emotional appeals in political ads work. University of Chicago Press.
Brader, T., Valentino, N. A., & Suhay, E. (2008). What triggers public opposition to immigration? Anxiety, group cues, and immigration threat. American Journal of Political Science, 52(4), 959-978. Retrieved from:
http://www.jstor.org/stable/25193860
Bryman, A. (2016). Social research methods. Oxford university press.
Butler, R. J., Cowan, B. W., & Nilsson, S. (2005). From obscurity to bestseller: Examining the impact of Oprah’s book club selections. Publishing Research Quarterly, 20(4),
23-34. Retrieved from: https://link.springer.com/content/pdf/10.1007/s12109-005-0045-2.pdf
Cacioppo, J. T., & Petty, R. E. (1984). The elaboration likelihood model of persuasion. ACR North American Advances. Retrieved from:
https://web.b.ebscohost.com/ehost/pdfviewer/pdfviewer?vid=2&sid=c6dae1f4-45c5-4ebb-af04-96aaa0b2ec5a%40sessionmgr104
Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source versus message cues in persuasion. Journal of personality and social
psychology, 39(5), 752. Retrieved from:
http://www.unc.edu/~fbaum/teaching/articles/jpsp-1980-Chaiken.pdf
Damasio, A. R. (2000). A second chance for emotion. Cognitive neuroscience of emotion, 12-23.
Elberse, A., & Verleun, J. (2012). The economic value of celebrity endorsements. Journal of Advertising Research, 52(2), 149-165. doi:10.2501/JAR-52-2-149-165
Erdogan, B. Z. (1999). Celebrity endorsement: A literature review. Journal of marketing management, 15(4), 291-314. doi: 10.1362/026725799784870379
Gadarian, S. K., & Albertson, B. (2014). Anxiety, immigration, and the search for information. Political Psychology, 35(2), 133-164. doi:10.1 11 l/pops. 12034 Garry, J. (2014). Emotions and voting in EU referendums. European Union Politics, 15(2),
235-254. doi:10.1177/1465116513514780
Garthwaite, C., & Moore, T. J. (2012). Can celebrity endorsements affect political outcomes? Evidence from the 2008 US democratic presidential primary. The journal of law, economics, & organization, 29(2), 355-384. doi:10.1093/jleo/ewr031
Glaser, J., & Salovey, P. (1998). Affect in electoral politics. Personality and social psychology review, 2(3), 156-172. Retrieved from:
https://web.a.ebscohost.com/ehost/pdfviewer/pdfviewer?vid=1&sid=f87f588d-aac0-4576-80e9-683cb6f910c7%40sessionmgr4009
Gross, K. (2008). Framing persuasive appeals: Episodic and thematic framing, emotional response, and policy opinion. Political Psychology, 29(2), 169-192. Retrieved from: http://www.jstor.org/stable/20447111
Huddy, L., Feldman, S., & Cassese, E. (2007). On the distinct political effects of anxiety and anger. The affect effect: Dynamics of emotion in political thinking and behavior, 202-230.
Hofstede, G. (2003). Culture's consequences: Comparing values, behaviors, institutions and organizations across nations. Sage publications.
Isbell, L. M., & Ottati, V. C. (2002). The emotional voter. In The social psychology of politics (pp. 55-74). Springer, Boston, MA.
Jackson, D. J., & Darrow, T. I. (2005). The influence of celebrity endorsements on young adults’ political opinions. Harvard international journal of press/politics, 10(3), 80-98. doi:10.1177/1081180X05279278
Kahan, D. M. (2012). Ideology, motivated reasoning, and cognitive reflection: An experimental study. doi: 10.2139/ssrn.2182588
Kamins, M. A. (1990) An investigation into the ‘Match-up’ hypothesis in celebrity advertising: When beauty may only be skin deep. Journal of Advertising, 19(1), 4. Retrieved from: https://www.jstor.org/stable/4188750?seq=1#page_scan_tab_contents Lee, J., & Choi, Y. (2018). Expanding affective intelligence theory through social viewing:
Focusing on the South Korea's 2017 presidential election. Computers in Human Behavior, 83, 119-128. doi: 10.1016/j.chb.2018.01.026
Lerner, J. S., & Keltner, D. (2001). Fear, anger, and risk. Journal of personality and social psychology, 81(1), 146. doi: 10.1037//O022-3514.81.1.146
Lindley, R. (2016). ‘A sadistic dentist eager to make us feel pain: Why low-information voters fall for Donald Trump.’ Salon. Retrieved from:
http://www.salon.com/2016/05/23/a_sadistic_dentist_eager_to_make_us_feel_pain_w hy_low _information_voters_fall_for_donald_trump/
MacKuen, M., Wolak, J., Keele, L., & Marcus, G. E. (2010). Civic engagements: Resolute partisanship or reflective deliberation. American Journal of Political Science, 54(2), 440-458. Retrieved from: http://www.jstor.org/stable/25652216
Marcus, G. E. (2002). Sentimental citizen: Emotion in democratic politics. Penn State Press. Marcus, G. E., MacKuen, M., & Neuman, W. R. (2011). Parsimony and complexity:
Developing and testing theories of affective intelligence. Political Psychology, 32(2), 323- 336. doi: 10. 1 1 1 1/j. 1467-9221 .2010.00806.x
Marcus, G. E., Neuman, W. R., & MacKuen, M. B. (2000). Affective intelligence and political judgment. Chicago: University of Chicago Press.
Nai, A., Schemeil, Y., & Marie, J. L. (2017). Anxiety, Sophistication, and Resistance to Persuasion: Evidence from a Quasi-Experimental Survey on Global Climate Change. Political Psychology, 38(1), 137-156. doi:10.1111/pops.12331
Niemantsverdriet, T. (2011) BN’ers in de politiek. Vrij Nederland. Retrieved from: https://www.vn.nl/bners-in-de-politiek/
Oskarson, M. (2007). Social risk, policy dissatisfaction, and political alienation. The political sociology of the welfare state, 117-148.
Pantoja, A. D., & Segura, G. M. (2003). Fear and loathing in California: Contextual threat and political sophistication among Latino voters. Political Behavior, 25(3), 265-286. Retrieved from:
https://link.springer.com/content/pdf/10.1023%2FA%3A1025119724207.pdf Parker, M. T., & Isbell, L. M. (2010). ow I vote depends on how I feel: The differential
impact of anger and fear on political information processing. Psychological Science, 21(4), 548-550. doi: 10.1177/0956797610364006
Petersen, M. B. (2010). Distinct emotions, distinct domains: Anger, anxiety and perceptions of intentionality. The Journal of Politics, 72(2), 357-365.
doi:10.1017/s002238160999079x
Pew Research Center. (2014). Political polarization in the American public. Pew Research Center: U.S. Politics & Policy. Retrieved from
http://www.peoplepress.org/2014/06/12/political-polarization-in-the-american-public/ Pew Research Center (2015). Political Interest and Awareness Lower Among Millennials.
Millenials & Political News. Retrieved from: http://assets.pewresearch.org/wp-content/uploads/sites/13/2015/06/Millennials-and-News-FINAL-7-27-15.pdf
Prior, M. (2010). You’ve either got it or you don’t? The stability of political interest over the life cycle. The Journal of Politics, 72(3), 747-766. doi: 10.1017/s0022381610000149 PVVpers (2018) Nieuwe video PVV Zendtijd Politieke Partijen. Retrieved from:
https://www.youtube.com/watch?v=NoBeCQhWFfY
Redlawsk, D. P., Civettini, A. J., & Lau, R. R. (2007). Affective intelligence and voting: Information processing and learning in a campaign. The affect effect: Dynamics of emotion in political thinking and behavior, 152-179. Retrieved from:
https://www.researchgate.net/profile/Andrew_Civettini
Ridout, T. N., & Searles, K. (2011). It's my campaign I'll cry if I want to: How and when use emotional appeals. Political Psychology, 32(3), 439-458. doi: 10.1 1 1 1/j. 1467-922 1.201 0.008 19.x
Rigterink, A. S., & Schomerus, M. (2017). The fear factor is a main thing: How radio
influences anxiety and political attitudes. The Journal of Development Studies, 53(8), 1123-1146. doi: 10.1080/00220388.2016.1219348
Roskos-Ewoldsen, D. R., Bichsel, J., & Hoffman, K. (2002). The influence of accessibility of source likability on persuasion. Journal of Experimental Social Psychology, 38(2), 137-143. doi:10.1006/jesp.2001.1492
Rossiter, J. R., & Smidts, A. (2012). Print advertising: Celebrity presenters. Journal of Business Research, 65(6), 874-879. doi:10.1016/j.jbusres.2011.01.010
Searles, K., Ridout, T.N. (2017). Emotion Researcher, ISRE’s Sourcebook for Research on Emotion and Affect, Andrea Scarantino (Ed.) Retrieved from:
http://emotionresearcher.com/the-use-and-consequences-of-emotions-in-politics/# Smith, E. R., Seger, C. R., & Mackie, D. M. (2007). Can emotions be truly group level?
Evidence regarding four conceptual criteria. Journal of personality and social psychology, 93(3), 431. doi:10.1037/0022-3514.93.3.431
So, J. (2013). A further extension of the extended parallel process model (E-EPPM):
Implications of cognitive appraisal theory of emotion and dispositional coping style. Health Communication, 28(1), 72-83. doi: 10.1080/10410236.2012.708633
Taber, C. S., & Lodge, M. (2006). Motivated skepticism in the evaluation of political beliefs. American Journal of Political Science, 50(3), 755-769. Retrieved from: https://www.jstor.org/stable/3694247
Valentino, N. A., Hutchings, V. L., Banks, A. J., & Davis, A. K. (2008). Is a worried citizen a good citizen? Emotions, political information seeking, and learning via the internet. Political Psychology, 29(2), 247-273. Retrieved from:
http://www.jstor.org/stable/20447114
Van der Brug, W. (2004). Issue ownership and party choice. Electoral Studies, 23(2), 209-233. doi: 10.1016/S0261-3794(02)00061-6
Van Kleef, G. (2012). Emoties lijsttrekkers geven de doorslag in 'emocratie'. Trouw.
Retrieved from: https://www.trouw.nl/home/emoties-lijsttrekkers-geven-de-doorslag-in-emocratie-~aae94936/
Veer, E., Becirovic, I., & Martin, B. A. (2010). If Kate voted conservative, would you? The role of celebrity endorsements in political party advertising. European journal of marketing, 44(3/4), 436-450. doi: 10.1108/03090561011020516
Verba, S., Burns, N., & Schlozman, K. L. (1997). Knowing and caring about politics: Gender and political engagement. The Journal of Politics, 59(4), 1051-1072. Retrieved from: http://www.jstor.org/stable/2998592
Wood, N. T., & Herbst, K. C. (2007). Political star power and political parties: Does celebrity endorsement win first-time votes? Journal of Political Marketing, 6(2-3), 141-158. doi:10.1300/J199v06n02_08
Appendices
Appendix A: manipulation checks posters of all political parties
The analysis of variance in the difference in means in anxiety experienced between the anxiety, anger and control condition for the PvdA was statistically significant F(5, 212) = 6.24, p < .001. Moreover, another analysis of variance in the difference in means on anger experienced between the anxiety, anger and control condition for the PvdA was also
statistically significant F(5, 212) = 31.87, p < .001. Both tests showed that the manipulations for the PvdA were successful. The means and standard deviations of all PvdA conditions can be found in table A1.
Table A1
Means and standard deviations of anxiety and anger experienced per PvdA poster
Experienced emotion Anxiety Anger
Stimulus M SD M SD n
Anxiety poster 2.53 1.37 2.29 1.06 38
Anger poster 3.05 1.79 4.49 1.71 37
Control poster 1.78 1.23 1.92 1.27 36
The difference in means in anxiety experienced between the anxiety, anger and control condition for the VVD was also statistically significant F(5, 212) = 8.60, p < .001.
Furthermore, another analysis of variance in the difference in means in anger experienced between the anxiety, anger and control condition for the VVD was also statistically significant F(5, 212) = 10.56, p < .001. Both tests showed that the manipulations for the VVD were successful. The means and standard deviations of all VVD conditions are given in table A2.
Table A2
Means and standard deviations of anxiety and anger experienced per VVD poster
Experienced emotion Anxiety Anger
Stimulus M SD M SD n
Anxiety poster 3.03 2.02 2.92 2.01 36
Anger poster 2.68 1.55 3.41 1.94 37
Next, the difference in means in anxiety experienced between the anxiety, anger and control condition for GroenLinks was again statistically significant F(5, 212) = 11.76 , p < .001. Moreover, another analysis of variance in the difference in means on anger experienced between the anxiety, anger and control condition for the GroenLinks was also statistically significant F(5, 212) = 29.2, p < .001.Both tests showed that the manipulations for
GroenLinks were also successful. The means and standard deviations of all GroenLinks conditions can be found in table A3.
Table A3
Means and standard deviations of anxiety and anger experienced per GroenLinks poster Experienced emotion Anxiety Anger
Stimulus M SD M SD n
Anxiety poster 2.82 1.73 2.09 1.53 34
Anger poster 3.24 2.11 4.22 1.93 37
Control poster 1.22 0.58 1.24 0.60 37
The difference in means in anxiety experienced between the anxiety, anger and control condition for the PVV was significant F(5, 212) = 2.30, p < .05. Moreover, another analysis of variance in the difference in means on anger experienced between the anxiety, anger and control condition for the PVV was also statistically significant F(5, 212) = 13.48, p < .001. The means and standard deviations of all PVV conditions can be found in table A4.
Table A4
Means and standard deviations of anxiety and anger experienced per PVV poster
Experienced emotion Anxiety Anger
Stimulus M SD M SD n
Anxiety poster 2.51 2.01 4.43 1.95 37
Anger poster 2.69 1.91 4.97 1.77 36
Appendix B: stimuli
PvdA: Anxiety, endorsement PvdA: Anxiety, no endorsement
PvdA: No emotion, endorsement PvdA: No emotion, no endorsement
VVD: Anger, endorsement VVD: Anger, no endorsement
GroenLinks: Anxiety, endorsement GroenLinks: Anxiety, no endorsement
GroenLinks: No emotion, endorsement GroenLinks: No emotion, no endorsement
PVV: Anger, endorsement PVV: Anger, no endorsement
s
