Negative emotions in political campaigns
The effect of fear and anger on voter turnout
Annika Schippers; 1057946 Master’s thesis
Master’s program Communication Science Graduate School of Communication
University of Amsterdam
Supervisor: Dr. Joost van Spanje June 27th 2014
1
Abstract
Little to no research has been done evaluating the effects of negative emotions in political campaigns on voter turnout. The research that has been done only uses data from two-party systems while it is possible that different results can be found in a country with a multi-party system. Therefore, this research evaluates the effects of anger and fear in political campaign material on peoples’ voting intention. Instead of using U.S. data, this research uses data that is collected through an online experiment in the Netherlands. The results show that anger has a rather strong negative effect on peoples’ voting intention and that fear has a rather strong positive effect on peoples’ voting intention. This is the opposite of what was expected based on the previous literature. This research also evaluated if the need for affect positively
moderates the relationship between anger or fear and peoples’ voting intention. However, the results show no significant moderation.
Keywords: emotions, anger, fear, political campaigns, negative campaigns, need for affect, voter turnout, voting intention
2
Introduction
Negative emotions have been used in political campaigns for decades. Whether it was used to make people afraid of war or terrorism, or to make them angry about tax raises or, more currently in the U.S., a new health insurance system, negative emotions are strongly represented in political campaigns. However, the most dominant school of thought when assessing voting decisions is primarily based on rational choice models (Simon, 1955). This is a process where people rationally determine how to reach their goals (ibid.). It is not until more recently that scholars researched the role of emotions in decision-making processes and in politics in general (Miller, 2006). Emotions in campaigns can trigger a cognitive effect that can lead to action (Nabi R. L., 1999). But emotions that people feel while watching, hearing or reading a political campaign can also, without a cognitive effect, stimulate them to become active. This however works differently for every specific emotion (ibid.).
Emotions can be categorized as either positive or negative. However, each discrete emotion1 can affect people’s behavior in a different way. This research will focus on negative campaigns because it will evaluate the effect of negative emotions on voting intention. Of all negative emotions, fear proves to have one of the strongest effects on people’s reasoning (Valentino, Hutchings, Banks, & Davis, 2008) while anger proves to have a strong effect on political participation (Valentino, Brader, Groenendyk, Gregorowicz, & Hutchings, 2011). Anger, according to Brader (2005), is a dispositional emotion. This means that anger leads to reliance on what people already know and will make them stick to his or her prior choices and beliefs. Fear on the other hand, is an emotion that triggers surveillancing. It leads to
uncertainty, after which there is a higher chance of information seeking. Hence, the first research question of this paper is: How and to what extent do anger and fear in campaign
material influence voters’ turnout intention?
1
3
Some people are more amenable to emotions than others. This is called a person’s need for affect which is “(…) a construct that describes individual differences in the tendency to approach or avoid emotion-inducing situations and activities” (Appel, Gnambs, & Maio, 2012, p. 2). Maio and Esses (2001) indicate that this includes the desire to experience and understand the emotions of oneself and others, and that citizens with a high need for affect believe that emotions are useful for shaping judgments and behavior (Haddock, Maio, Arnold, & Huskinson, 2008). The need for affect is also associated with the degree to which
someone’s attitude is guided by affective information (Haddock & Huskinson, 2004). Haddock et al (2008) suggest that individuals with a higher need for affect show a greater receptivity to a message that emphasizes affective information. Affect-based messages have a stronger persuasive power over individuals with a higher need for affect (ibid.). By
incorporating the need for affect in this research, it is possible to evaluate if being more amenable to emotions strengthens the effect of anger and fear in political ads on voter turnout. Therefore, the second research question of this paper is: Is the effect of emotions in campaign
material on voter turnout stronger for voters who are more amenable to emotions?
In the U.S., political campaigns often use affect-laden symbols, language and music to make people feel a certain way (Chang, 2001). These emotions can either be negative or positive. Negative emotions are often used in negative campaigns (ibid.). Negative campaigns are campaigns that attack another politician or their policy and are used frequently in the U.S. (Lau, Sigelman, & Rovner, 2007). As a result, almost all of the research on this topic uses data from the U.S. (Walter, 2012). This is problematic because the U.S. is in many ways a particular case. It has a two-party system in which vote-seeking is equivalent to
office-seeking, leading to a more aggressive campaign style than in countries that have a multi-party system (ibid.) (Djupe & Peterson, 2005). That is why negative campaigns are used more
4
frequently in this type of political system (ibid.). In a multi-party system, multiple parties will need to form a coalition and work together after the elections (Walter, 2012). A benefit of having more than two parties to choose from is that citizens have a greater chance of finding a party that they identify with and that they would like to support (Norris, 1997). It is thus possible that campaign effects are different in countries with a multi-party system than in countries with a two-party system.
There are almost no studies about the effects of negative campaigns in countries that have a multi-party system. As a result, the generalizability of the existing theories is open to question. Recently, more research is done on negative campaigning in multi-party systems (Hansen & Pedersen, 2008). The effect of negative emotions in political campaigns however, has not yet been researched in a country with a multi-party system. Therefore, this study will try to fill this gap in the current literature. The results of this research will also be useful for political parties in multi-party systems. It will show them how citizens respond to negative emotions (fear and anger) in political campaigns. By using these results, parties can anticipate voters’ reactions to negative campaigns in the future. If fear or anger (or both) increase voter turnout, using these emotions in political campaigns could also be beneficial to the democracy in a country (Franklin, 2004). If more people vote, the government will be a better
representation of its citizens and minorities will also be represented better.
This paper will start by evaluating the existing literature on the main topics after which the main variables will be discussed intensively. This will be done in the theoretical
framework chapter of this paper. After assessing the literature each variable will be defined and operationalized, which will be done in the methodology section of this paper. This section will also discuss the experimental research design and the statistical analyses that will be used to answer the research question. In the next chapter, the results of the performed analysis will be described and the hypotheses will either be accepted or rejected. Finally, in the last chapter
5
conclusions will be drawn from the results and possible improvements, recommendations and implications for further research will be discussed.
Theory
During election times, politicians and political parties usually try to convince voters to vote for them and their party. To gain votes, politicians can use positive campaigns, negative campaigns or a mixture of both of these types. Negative campaigns focus on criticizing the opponent while positive ads focus on good characteristics or accomplishments by the sponsoring party (Kaid & Johnston, 1991). Negative campaigns are often imbued with negative emotions to make people feel more negative about another candidate or policy (Chang, 2001). This research will focus on negative campaigns that trigger anger or fear.
Types of negative campaigning
There are multiple definitions of negative campaigning. A division can be made based on who or what a campaign is criticizing. A common definition of negative campaigning is that it criticizes political opponents on content as well as personal characteristics (Geer, 2006) (Lau & Pomper, 2004). Others define negative campaigning as “(…) a reference by one candidate or the candidate's campaign challenging a same-party candidate's fitness, issue positions, experience, temperament, etc.” (Djupe & Peterson, 2002). This definition only focuses on personal characteristics and arose from the idea that negative campaigning is primarily a ‘dirty’ type of campaigning. Attacking an opponent’s character or looks is viewed as less legitimate than criticizing an opponent’s policies or political opinions (ibid.). There are therefore two main definitions in literature. The first one is the directional definition, which feels that any attack towards an opponent is seen as negative campaigning (Walter, 2012). The second definition is the evaluative definition, which feels that not all criticism directed
6
towards an opponent is seen as a negative campaign, but only those critiques that can be considered unfair, illegitimate, dishonest, and dealing with trivial issues (ibid.).
The effects of negative campaigning on participation
There is a lot of discussion between academics on the effects of negative campaigning on voter participation. According to some, it demobilizes the electorate (Ansolabehere, Iyengar, & Simon, 1999). They argue that negative campaigns demobilize the electorate because it increases voter disgust towards the competing candidates and voter alienation toward the political process in general. (Kahn & Kenney, 1999). After witnessing several negative campaigns of both opposing politicians, voters become convinced that both candidates are not worth their vote (ibid.). According to others, negative campaigns can mobilize the electorate (Freedman & Goldstein, 1999) (Lau & Pomper, 2001) (Lau & Pomper, 2002) (Namkoong, Fung, & Scheufele, 2012) (Wattenberg & Brians, 1999). Lau & Pomper (2001) have conducted a survey with 8300 respondents across six different election years. They also evaluated 189 election campaigns and came to the conclusion that, given the level of campaign negativism and spending that had occurred in these elections, turnout was about 3.4% higher than when there would have been a completely positive campaign (Lau & Pomper, 2001). Kahn & Kenney (1999) find that people make a distinction between
legitimate and tempered criticism. Criticism that is substantive and well reasoned is perceived by voters as useful information to make a good decision on who to vote for, while excessive mudslinging by candidates that is covered frequently by the news media, actually turns voters away (ibid.). Taking these findings into account, this research will only evaluate the effects of substantive criticism that focuses on policy.
Explaining voter turnout
Voter turnout can be explained on two different levels; the individual level and the aggregate level. Scholars who research what affects voter turnout at the aggregate level, come
7
to several conclusions, although there are several factors that they often agree upon (Geys, 2006). These factors can be categorized into three different categories: socio-economic (for example population size and electoral closeness), political (for example campaign
expenditures) and institutional (for example compulsory voting and electoral system) (ibid.). Research that focuses on individual explanations for voter turnout also comes to different conclusions sometimes, though there are some factors that are known to influence voter turnout. These factors are for instance: gender, age, income, education, political interest, etc. (Powell Jr., 1986) (Krupnikov, 2011). This research will focus on explaining voter turnout on the individual level.
The first scholars who evaluated what affects voter turnout came to the conclusion that people were more likely to vote when the outcome hung in the balance or when vital issues were at hand (Boechel, 1928). Low voter turnout would have therefore been blamed on the character of the elections (Franklin, 2004). But it was the rational choice approach that
changed the way scholars thought about voting behavior (ibid.). The chance that people really believe that their vote can change society for the better, so that they would benefit, is very low. Therefore, the only rational reason for people to vote, would be to gain nonmaterial benefits from it such as being able to truthfully tell your friends and colleagues that you voted, or because it gives you the feeling that you contributed something to society, or other aspects of civic virtue (ibid.).
In the following years, those who studied electoral participation focused on “(…) voting as a habit that people learned during their formative years; a learning experience dominated by education and social status” (ibid., p 2-3). This resulted in what was called the baseline, or resource model (Franklin, 2004). This model explained electoral participation by income, occupation and education (Verba & Nie, 1972). It was thus assumed that a person’s likeliness to vote depended on personal characteristics. More recently, this baseline model has
8
been joined by a ‘mobilization model’. This model takes into account external reasons for voting such as being mobilized by a party, interest groups or a candidate (Brady, Verba, & Schlozman, 1995) (Rosenstone & Hansen, 1993). Taking external reasons for voting into account places more emphasis on political campaigns and their strategy. As a result,
campaigning becomes more professionalized. This research will focus on external reasons for voting because it will evaluate the effects of emotions in political campaign material on voters’ turnout intention.
Emotions in political campaigns
In the U.S., political campaigns often use affect-laden symbols, language and music to make people feel a certain way (Chang, 2001). It is therefore important to take emotions into account as well when evaluating the effect of political campaigns on voter turnout. Political behavior and decisions are not merely based on cognitive processes, emotions also play a significant role in determining people’s behavior and decisions (Miller, 2006). Emotion can trigger a cognitive effect that could then lead to action (Nabi R. L., 1999). This effect
however works differently for every specific emotion (ibid.). For a long time, emotions have been categorized as either negative or positive. Discrete emotions had been rather overlooked until more recent studies (Breckler, 1993). Positive political campaigns use positive emotions such as hope, pride, enthusiasm and reassurance while negative political campaigns use negative emotions such as anger, uncertainty, guilt and fear (Chang, 2001). There is growing evidence that discrete emotions have different effects on voter turnout. It is thus important that the effects of these individual emotions are researched (Brader, 2005).
According to Damasio (2008), both positive and negative emotions can improve decision-making processes. “Our brains use emotions to register and signal the positive and negative implications of all alternative courses of action” (Valentino, Brader, Groenendyk, Gregorowicz, & Hutchings, 2011, p. 158). After rapidly going over these alternative courses,
9
our brain gives us a ‘gut feeling’ about the best way to proceed (ibid.). Emotions can
encourage reward-seeking or danger-averting behavior (Gray, 1990) . Both of these can lead to action according to Gray (1990).
The effects of fear and anger
Anxiety occurs when danger or negative outcomes can be expected. Fear triggers a distinct cognitive system (the surveillance system) which has the effect that an individual pays extra attention to its environment, is more attentive to new information and is more likely to break out of habitual patterns of behavior (Valentino, Hutchings, Banks, & Davis, 2008). Brader (2005) finds this same effect in political advertising. Ads that are likely to trigger anxiety make citizens rely less on predispositions and make them want to search for more and new information. But Valentino et al. argue that individuals who experience fear, would rather avoid than attack a problem that is at hand. Fear prepares the body to act, but an anxious individual might abstain from action (Lerner & Keltner, 2001) (Valentino, Brader, Groenendyk, Gregorowicz, & Hutchings, 2011) because they lack a sense of control and are unable to attribute blame to someone2 (Smith & Ellsworth, 1985). Individuals are thus less likely to act when they are anxious, but this is only the case for costly political actions3 (Valentino, Brader, Groenendyk, Gregorowicz, & Hutchings, 2011). Thus, the hypothesis is:
H1: Political campaign material that evokes fear will decrease people’s intention to vote.
Anger derives from a different relationship between the individual and their
surroundings. Anger rises when a person is certain that something bad has happened or will happen (Tiedens & Linton, 2001) and the individual feels that they have control over the situation (Valentino, Brader, Groenendyk, Gregorowicz, & Hutchings, 2011). It leads to risk-accepting behavior, because people rely more on heuristics than on systematic information
2
This could be a person, a political party, an international organization, etc.
3
Actions that take more effort, like for instance participating in a campaign, voting, demonstrating etc. Less costly types of political participation include talking about politics with friends or family, signing an e-petition, etc.
10
processing (Tiedens & Linton, 2001). Anger, according to Brader (2005), falls in the category of dispositional emotions. This means that anger leads to reliance on what you know already, top-down processing from your prior beliefs, and will thus make voters stick with what was already working for them. But anger also encourages us to approach the problem at hand, the thing that makes us angry (Valentino, Brader, Groenendyk, Gregorowicz, & Hutchings, 2011). When a person is thus exposed to political campaign material that makes them angry, they would want to act on that emotion and will therefore be more likely to vote. Hence the hypothesis is:
H2: Political campaign material that evokes anger will increase people’s intention to vote.
Explaining peoples’ need for affect
The effects of emotions on voter turnout might be different for every individual. Some people are more amenable to emotions than others and like to expose themselves to emotions more often than others (Maio & Esses, 2001). People who often expose themselves to
emotions do so because they have had positive experiences with emotions or emotional situations in the past (ibid.). People who do not like to expose themselves to emotions have usually had negative experiences with emotions or emotional situations (ibid.). Past
experiences with emotions thus shape to what extent people want to expose themselves to emotions, leading to some people avoiding emotions and emotional situations while others approach them. This concept is described as the need for affect (ibid.). When a person has a higher need for affect, it is likely that the emotional triggers in a political ad will have a stronger effect on this person than on a person with a low need for affect. This is expected because a person with a high need for affect would want to hold on to emotions while a person with a low need for affect would want to escape the emotional situation as quickly as possible. People thus respond differently to emotional triggers. Therefore the need for affect is
11
expected to positively moderate the main effect between emotion in campaigns and voter turnout. The hypotheses are:
H3: The higher the need for affect, the stronger the negative effect of fear in political campaigns on voting intention will be.
H4: The higher the need for affect, the stronger the positive effect of anger in political campaigns on voting intention will be.
Method
Selection of the research unit
When selecting a case to measure the effect of negative emotions in political campaign material on peoples’ voting intention, this research requires data from another country than the U.S.. The U.S. case is a very specific one because they have a two-party system. A lot of the current research is based on U.S. data which might lead to less generalisable theories and findings (Walter, 2012). Because most research uses U.S. data, the findings of those articles might not be the same in a country with another type of political system. A perfect example of a case that has the complete opposite political system, is the Netherlands.
According to Duverger (1954), the electoral system strongly influences the amount of political parties in a country. A country with a majority system, such as the U.S., usually only has two parties (ibid.). A country with a system of proportional representation usually has multiple parties, such as the Netherlands, which currently has 11 different parties in
parliament (ibid). This means a party can never rule the country on its own, because no party holds a majority of seats in parliament (Walter, 2010). The benefit of having more parties in parliament is that it implies a wider range of options for people. It is thus more likely that people find a party with opinions that are close to their own beliefs, than would be the case if there were less parties to choose from. This is also found when comparing the turnout
12
percentages of the Netherlands and for instance the U.S. While turnout scores for the Dutch national elections in 2012 where 74,20%, the U.S. turnout scores for the presidential elections in 2012 were 54.62% (IDEA International Institute for Democracy and Electoral Assistance) (Zwakhals, Giesbers, & Deuning, 2012). The fragmented political system of the Netherlands might thus positively affect voter turnout, since there is a greater range of political expression available (Crepaz, 1990). The fact that the Netherlands has a multi-party system that is extremely fragmented makes it almost the complete opposite of the U.S. where there are only two parties. The Dutch case is thus ideal for testing whether the previously found theories hold up in a country with a different political system, or if different effects will be found.
Characteristics of the respondents
The respondents are not a representative sample of society, because of the expected difficulties in finding a representative sample with the means that are available. But because all respondents will be randomly assigned to a campaign poster, it is assumed that each group will consist of roughly the same type of people (age, education, etc.) Therefore, the groups can be compared to one another. Because it is known that almost all people process
information in the same way, it is expected that, even though the sample is not representative, the results will be generalisable. Not having a representative sample will slightly decrease the representativeness of the results, but it will not be problematic.
Data will be collected through an online experiment among Dutch citizens that are legally allowed to vote. 200 respondents have filled out the entire questionnaire (N = 200), hence there are no missings. 40.5% of the respondents in male while 58.5% of the
respondents in female4. It is expected that a large share of the respondents are students because 67% of the respondents is between the age of 18 and 25 and 80.5% is highly educated.
4
13
Research design
A survey experiment will be conducted to examine the effects of anger and fear in political campaign material on peoples’ voting intention. Only one campaign medium is chosen, namely a campaign poster. To trigger anger or fear among all the respondents and to avoid the influence of partisanship, the campaign material was specifically made for this experiment and the posters did not contain any political party logo or name. The campaign posters that will be used focus on European issues that could trigger fear or anger. To allow for variation in voter turnout, the European elections are chosen, because voter turnout is much lower in these elections than in the national elections. To trigger fear, the respondents will be exposed to an ad that shows a realistic threat for everyone in the Netherlands. To trigger anger, the respondents will be exposed to an ad that shows something outrageous that the Dutch government has done due to being a part of the European Union. Phrases as well as visuals will be used to trigger anger and fear. The ads that are used can be found in Appendix I. Each respondent is randomly assigned to one of the three campaign ads. All respondents will receive either a neutral ad that does not trigger any emotion, an ad that triggers anger or an ad that triggers fear.
Before using these ads in the experiment, a manipulation check will be done, asking the respondents what kind of emotions they feel when being exposed to the campaign material. Before being shown a political campaign poster, the respondents have to fill out a questionnaire, determining their level of need for affect. After being exposed to the ad, participants will answer questions about their intention to vote. The questionnaire will also contain some buffer questions asking about respondents’ political knowledge, political trust and likelihood to vote for a party. By doing so, it will not be evident to the respondents what exactly the experiment tries to measure, making it more likely for respondents to answer the questions honestly.
14
Measurements
When measuring the main concepts of this research, it is important that these concepts are defined properly so that it is clear what exactly will be measured and how it will be measured. Therefore, the operationalization section of this paper will focus on clarifying the concepts that are used in this research and will explain how they will be measured in the experiment.
Voting intention
Because there are currently no elections, this research will use the intention to vote as a measure for voter turnout. The stimuli that will be presented will focus on the European elections, because voter turnout is not usually high for these elections, while the voter turnout for national elections is very high which will therefore allow for more variation on the
dependent variable.
A problem that can occur when measuring voter turnout this way, is over-reporting. Over-reporting refers to the fact that people are more likely to say that they have voted or will vote, than that they actually have voted or will vote. The most common reason for over-reporting is that respondents give socially desirable answers (Duff, Hammer, Park, & White, 2007). Over-reporting can range from 8 to 14 percent, depending on how one measures actual turnout and misreporting (ibid.). These high levels of over-reporting can have significant consequences for the research attempting to explain voter turnout. It is especially problematic when over-reporting is systematic, meaning some respondents are more likely to over-report than others (ibid.). However, the chances for socially desirable answers are much lower for European elections than for national elections leading to more valid conclusions (Karp & Brockington, 2005).
Research by Duff et al. has proven that providing respondents with socially acceptable excuses for not voting will decrease over-reporting with 8 percent. Especially for respondents
15
that are the least likely to vote, the voting estimates will be considerably lower than when using traditional questions about voter turnout (ibid.). However, Karp et al. (2005) conclude that respondents who feel that they should have voted (respondents with political interest who are higher educated) usually over-report and that questions that provide the respondent with socially acceptable excuses can decrease over-reporting for these groups as well. The questions that can be used to decrease over-reporting are:
- In talking to people about elections, we often find that a lot of people were
not able to vote because they weren’t registered, they were sick, or they didn’t have time. How about you – did you vote in the elections yesterday? 1. Yes, voted. 5. No, didn’t vote.
- Which of the following statements best describes you: 1. I did not vote; 2. I thought about voting this time but didn’t; 3. I usually vote but didn’t this time; 4. I am sure I voted.
However, to use these questions, some alterations have to be made. First, these questions focus on whether or not respondents have voted (past), while this research will evaluate if respondents will go and vote (future). Second, in the Netherlands voters do not have to be registered, people can just vote when they are 18 years old or older. Therefore this research will use the following question to determine if respondents are likely to vote:
- When talking to people about elections, we often find that a lot of people are not able to vote because they are sick, or they don’t have time. How about you, if the European elections were held tomorrow, how likely is it that you would vote? (10-point scale; I would definitely not vote – I would definitely vote) (M = 6.99, SD = 3.64).
16
Because this is a rather broad scale, this research will also use a question with a 3-point scale. This question will be used later on to form a dichotomous scale, making a binary logistic regression analysis possible5.
- If the European elections were held tomorrow, would you go and vote? (no - maybe - yes) (M = .63, SD = .485).
Emotions in campaigns
Political campaigns are often filled with emotional triggers, to make people feel a certain way (Chang, 2001). This research will evaluate the two main negative emotions, fear and anger. Each respondent will thus be exposed to a campaign ad with a different emotional trigger. Therefore, the sample will be randomly divided into three groups, each being exposed to a different ad. These groups will be as follows:
- a group that will receive a political ad without emphasis on emotion (N = 66) - a group that will receive a negative ad with a focus on fear (N = 69)
- a group that will receive a negative ad with a focus on anger (N = 65)
Anger and fear are triggered differently. While people feel angry when confronted with something that they dislike and that has already happened, they feel scared when confronted with something bad that might happen. So to trigger anger, a campaign poster will be made that confronts the respondents with something that has already happened that can be perceived by everyone as bad. For this experiment, the poster focuses on the Greeks early retirement while the Dutch people are working until they are 67. To trigger fear, a campaign poster will be made that confronts the respondents with something that might happen that is perceived by everyone as bad. For this experiment, the poster focuses on a possible war with Russia6. Each poster will be an independent variable where respondents who have been exposed to the
5 Therefore the mean and standard deviation are given for the dichotomous variable. 6
17
poster will be coded as 1 and respondents who saw one of the other two posters will be coded as 0.
Need for affect
The need for affect is a moderator variable in this research, which in this case means that it will be evaluated whether a higher need for affect will strengthen the effect between the independent and the dependent variable. The need for affect will be measured in the
questionnaire based on 26 questions that are specifically designed to measure this variable (Maio & Esses, 2001). All of the 26 questions are tested by Maio & Esser (2001) and together have proven to be a valid measure of a respondent’s need for affect. 13 questions will measure the level of avoiding emotions (M = 40.14, SD = 15.05) and 13 questions will measure the level of approaching emotions (M = 58.51, SD = 11.74) (ibid.). The questionnaire contains statements like “I like to dwell on my emotions” and “I would prefer not to experience either the lows or highs of emotions” (ibid.).7
All of these 26 questions will have a 7-point answering scale. This means that the range of each individual variable will be -3 to +3 (strongly disagree – strongly agree). There is a strong internal consistency in the scales for measuring emotion approach (α = .88), emotion avoidance(α = .88) and the total scale for the need for affect (α = .78). As was expected according to Maio and Esser (2001), the approach and avoidance scales were significantly negatively correlated, r = -.25, p < .001. To convert all 26 questions into 1 variable, the scores of all questions about avoiding emotions will be added together, and the scores of those about approaching emotions will be also added
together. After adding these scores, the emotional avoidance score will be subtracted from the emotional approach score. The outcome of this will be a respondent’s score on the variable need for affect (M = 18.37, SD = 19.64).The need for affect scores range from -48 to 61.
7
18
Analyses
To answer the two research questions, two different kinds of statistical analyses will be performed. To measure if anger and fear have an effect on peoples’ voting intention, a bivariate linear regression will be performed. Because the second research question of this paper asks if the need for affect will moderate the effect between anger or fear and peoples’ voting intention, the analyses will also be performed as a multivariate regression analyses that will contain the main independent variable, the need for affect and the interaction variable.
Analyses and results
In this part of the paper the analyses of the data and the results will be discussed. First, the assumptions for regression analysis will be discussed shortly, after which the effectiveness of the stimuli will be tested by doing a manipulation check. After this, the variables will be tested through a bivariate regression analysis as well as through a binary logistic regression. After this, the analyses will be done including the moderator variable as an interaction variable to evaluate if the need for affect strengthens the effect of the emotions that are
induced by the campaign material. This will be done through a bivariate regression analysis as well as through a binary logistic regression.
Assumptions
When a regression analysis is used to analyze the data, one should be “(…) aware that the validity of the technique depends on whether certain assumptions are satisfied” (Allison, 1999, p. 119). This research will use the set of assumptions provided by Allison (1999). He provides 5 assumptions that have to be satisfied before starting a regression analysis:
Linearity, mean independence, homoscedasticity, uncorrelated disturbances and the normality assumption. But to be sure that the collected data is completely suited for the analyses that will be done, there will also be checked for multicollinearity and outliers. For almost all
19
variables all five assumptions have been verified and none of them poses a serious problem for the analysis. However, for one of the dependent variables, ‘when talking to people about elections, we often find that a lot of people are not able to vote because they are sick, or they don’t have time. How about you, if the European elections were held tomorrow, how likely is it that you would vote?’, one assumption is not satisfied. This is the normality assumption. However, according to Allison (1999, p. 130) “If the sample is moderately large8, we can dispense with the normality assumption entirely”. Therefore, it will be kept in mind when looking at the results but will not be taken to seriously because this experiment has 200 respondents. Because of the strange distribution on this question, the research model will also be evaluated using the 3-point scale dependent variable. None of the independent variables proved to be multicollinear9. To check for outliers, Cook’s distance was used. The
demarcation for Cook’s distance is calculated by 4/n, so in this case 4 / 200. This gives the following number: 0.02. None of the values showed a Cook’s distance higher than 0.02. There are therefore no outliers in the data.
Manipulation check
In measuring the effect of emotions in campaign material on voting intention is important to verify if the stimuli made the respondents feel the correct emotion. This means that respondents who were shown a campaign poster that should induce fear should feel more scared than respondents who were shown a neutral or anger-inducing poster. The same applies to the campaign poster that should induce anger: respondents who were shown this poster should feel more angry than respondents that were shown a neutral or fear-inducing poster. To measure if the respondents were more angry or scared after seeing the randomly assigned poster, each respondent had to answer how angry, scared or neutral they felt after being exposed to the poster, on a 7-point scale. To verify if the stimuli that were used in the
8
100 cases or more
20
experiment significantly influenced the respondents feelings, an independent sample t-test was used. Respondents who were exposed to a campaign poster inducing fear were more scared (M = 2.56, SD = .99) than respondents who were exposed to a neutral or anger-inducing campaign poster (M = 2.19, SD = .61), t(198) = 4.054, p <.001 (two-tailed). The mean difference between these two groups is - .461 with a 95% confidence interval ranging from -.686 to -.23710. Respondents who were exposed to a campaign poster inducing anger were more angry (M = 3.29, SD = 1.497) than respondent who were exposed to a neutral or fear-inducing campaign poster (M = 2.21, SD = 1.373), t(198) = -5.045, p <.001 (two tailed). The mean difference between these two groups is -1.077 with a 95% confidence interval ranging from -1.499 to .656.
Bivariate regression analysis
To evaluate if there is a direct effect between being exposed to one of the two emotion-inducing posters and voting intention, a bivariate regression analysis was run separately for the emotion-inducing posters. Anger and fear prove to not have a statistically significant effect on voting intention at the level p = < .0511. Hypotheses 1 and 2 are thus rejected.
Logistic regression analysis
In order to run a logistic regression, the dependent variable has to be dichotomous. Therefore the variable ‘if European elections were held tomorrow, would you vote?’ is
recoded into 0 = no/maybe (N = 75) and 1 = yes (N = 125). Tables 4 and 5 show that both fear (B = .675, SD = .322) and anger (B = -.539, SD = .309) have a significant effect on voting intention when p < .10. The coefficient for fear has a Wald statistic equal to 4.394 which is significant at the level p < .05. When respondents were exposed to a fear-inducing poster, the logged odds of voting increases by .675. The exp(b) column in table 4 shows that when
10 The tables of the T-test can be found in Appendix IV. 11
21
respondents were shown a fear-inducing campaign poster, their odds of voting increased with 96.5%. This effect is significant when p < .05. Fear thus has a rather strong significant
positive effect on peoples’ voting intention. However, the direction of this effect is the opposite of what was expected. Therefore Hypothesis 1 ‘political campaign material that
evokes fear will decrease people’s intention to vote’ is still rejected.
Table 4
Binary logistic regression analysis fear
Independent variable B S.E. Wald Sig. Exp(B)
Fear .675 .322 4.394 .036** 1.965
Constant .292 .177 2.736 .098* 1.339
Cox & Snell R2 .023
Nagelkerke R2 .031
Dependent variable: If European elections were held tomorrow, would you vote? 0 = no/maybe and 1 = yes
N = 200 ** P < .05 * P < .10
Table 5 shows that the coefficient for anger has a Wald statistic equal to 4.394 which is significant at the level p < .10. When respondents were exposed to an anger-inducing poster, the logged odds of voting significantly decrease by .539. The exp(b) column shows that when respondents were shown an anger inducing campaign poster, their odds of voting decrease with 41.7%. This effect is significant when p <.10. Anger thus has a rather strong negative effect on peoples’ voting intention. However, the direction of this effect is the opposite of what was expected. Therefore Hypothesis 2 ‘political campaign material that
22
Table 5
Binary logistic regression analysis anger
Independent variable B S.E. Wald Sig. Exp(B)
Anger -.539 .309 3.050 .081* .583
Constant .693 .183 14.414 .000** 2.000
Cox & Snell R2 ,015
Nagelkerke R2 ,021
Dependent variable: If European elections were held tomorrow, would you vote? 0 = no/maybe and 1 = yes
** P < .05 * P < .10 N = 200
Interaction effect
Since it is expected that the effect of the campaign would be stronger for respondents with a higher need for affect, the analyses will also be run with the need for affect as a moderator variable. This will be done separately for both anger and fear inducing campaigns and will also be tested for both dependent variables. Table 6 indicates that the three predicting variables only explain 1.2% of the variance (R2=.01, F (3, 196) = .775, p < .509) 12. The effects are very insignificant hence little can be concluded from the results, except that the need for affect does not moderate (β = -.02, SD = .03, p < . 597) the relationship between emotions in campaign material and voting intention. Therefore hypotheses 3 ‘the higher the
12
23
need for affect, the stronger the negative effect of fear in political campaigns on voting intention will be’ will be rejected.
Table 7 indicates that the three predicting variables only explain 1.1% of the variance (R2= .01, F (3, 196) =.717, p < .543) 13. However, very little can be concluded from the results in table 7 because both effects are not significant, except that the need for affect does not moderate (β = -.02, SD = .02, p < .365) the relationship between emotions in campaign material and voting intention. Therefore, hypothesis 4 ‘The higher the need for affect, the
stronger the positive effect of anger in political campaigns on voting intention will be’ will be
rejected.
Interaction effect with a binary logistic regression analysis
In the previous analysis one can see that using a logistic regression, and thus a dichotomous measurement for the dependent variable, resulted in significant effects for both anger and fear in campaign material. To be certain that the need for affect is not a significant moderator variable, the analysis is also performed using a logistic regression. Table 8 shows that the moderator variable unexpectedly shows a negative relationship (B = -.013, SD = .018,
p < .444) which would mean that a higher need for affect weakens the effect of fear on voting
intention14. However, this relationship is not significant which means that little can be concluded from the results, except that the need for affect does not moderate the relationship between emotions in campaign material and voting intention. Therefore hypotheses 3 ‘the
higher the need for affect, the stronger the negative effect of fear in political campaigns on voting intention will be’ is still rejected.
The effect of both anger and the moderator variable for need for affect prove to be insignificant15. The moderator variable shows no effect (B = .000, SD .016 p < .958). Since the effects are insignificant, little can be concluded from the results, except that the need for
13
Table 7 can also be found in Appendix VI.
14 Table 8 can be found in Appendix VII. 15
24
affect does not moderate the relationship between emotions in campaign material and voting intention. Therefore, hypothesis 4 ‘The higher the need for affect, the stronger the positive
effect of anger in political campaigns on voting intention will be’ is still rejected.
Discussion and Conclusion
The main goal of this research was to evaluate if negative emotions in campaign material affect if people intend to go and vote. Dutch data was used to evaluate this effect, because it has an opposite political system from the data used in previous research. The main research question of this paper is: How and to what extent do anger and fear in campaign material
influence voters’ voting intention? To find an answer to this question an experiment has been
done with a sample of 200 respondents, each being randomly exposed to a different political campaign poster. To be able to answer the second research question, ‘Is the effect of emotions
in campaign material on voter turnout stronger for voters who are more amenable to
emotions?’ respondents were asked 13 questions measuring their emotional approach and 13
questions measuring their emotional avoidance.
Voting intention has been measured on two different scales, a 10-point scale and a dichotomous scale. Therefore a bivariate linear regression analysis as well as a logistic regression analysis has been performed. The linear regression analysis found no significant results for both anger and fear in campaign material. This can be partially explained by the fact that the distribution of this variable was not normal but rather extreme. On a 10-point scale the values 1 (11.5%) and 10 (45%) were chosen most often, while all the values in between were chosen only a few times (3% to 7.5%).
However, the binary logistic regression did show a significant positive effect for fear and a significant negative effect for anger. After evaluating the outcomes of the analyses the
25
first research question ‘how and to what extent do anger and fear in campaign material
influence voter’s voting intention?’ can be answered. Fear has a significant positive effect on
voting intention when voting intention is measured on a dichotomous scale. When
respondents were shown a fear-inducing campaign poster, their odds of voting increased with 96.5 percent. This means that respondents who were exposed to the fear-inducing campaign poster almost have twice as much chance of voting, as respondents that were exposed to one of the other two posters. Fear thus stimulates people’s voting intention.
Anger on the other hand had a significant negative effect on voting intention when voting intention was measured on a dichotomous scale. When respondents were shown an anger-inducing campaign poster, their odds of voting decreased with 41.7 percent. This means that respondents who were exposed to the anger-inducing campaign poster are much less likely to vote than respondents who were exposed to one of the other two posters. Anger in political campaign material, thus demotivates people to vote.
Why both emotions had the opposite effect than was expected is not clear. Using a manipulation check for both posters showed that both positively affected the targeted emotions. This could therefore not be the reason for opposite results. Accessing the Dutch case instead of a country with a two-party system, could lead to different results. However, it would then be expected that the effects might be stronger, because people have a more parties to choose from and because people feel that their vote has more meaning in a system of proportional representation (Norris, 1997). But it still does not explain why the results that were found in this experiment, are the opposite of what was expected.
It is arguable that when being exposed to a poster that confronts the respondent with something bad that has already happened, even though the respondent voted in the previous elections, could make the respondent feel like it does not matter if you vote or not. The anger poster in this experiment was about the Dutch citizens having to work until they are 67 years
26
old, while the Greeks still retire at 55. Money has already been given to Greece and the retirement age in the Netherlands has already been raised, which could lead to respondents feeling like there is nothing they can do about it now, thus giving them no reason to vote. The anger-inducing poster however was about a very current issue, the rising danger of a war between Russia and Ukraine. This has not happened yet but citizens are scared that it might happen. Since it has not happened, citizens might feel like they still have something to say about this matter, leading to a higher intention of voting among the respondents. These arguments are only based on assumptions, but it could be a reason for the reversed findings. However, further research on this matter is necessary.
The second research question ‘is the effect of emotions in campaign material on voting
intention stronger for voters who are more amenable to emotions?’ can also be answered. The
answer to this question is that the effect of emotions in campaign material on voting intention is not stronger for voters that are more amenable to emotions. Contrary to what was expected, a respondent’s need for affect does not seem to positively moderate the effect of emotions in campaign material on voting intention. It was striking that for both anger and fear, the need for affect seemed to either have no effect or a negative effect, meaning that a higher need for affect would weaken the relationship between anger or fear and voting intention. This effect was not significant but it would still be interesting to research why the need for affect at least seems to have a negative effect on the effect of emotions in campaign material on peoples’ voting intention.
It is arguable that when respondents who usually avoid emotions are confronted with emotion inducing campaign material, they experience a stronger effect of these emotions, because they are not used to emotional situations. And that when respondents who approach emotions experience a smaller effect when confronted with emotion-inducing campaign
27
material, because they are used to these sorts of emotional situations. This could explain the reversed direction of the findings, but further research on this is necessary.
Limitations and recommendations for future research
A limitation of this research is that it only measured a change in voting intention, and not in actual voter turnout. This is a limitation because even though this research tried to minimize over-reporting, it is still likely that it has occurred. To really measure the effect of voter turnout is very difficult, but the results would be more accurate if it were possible. This research also measured peoples’ voting intention immediately after respondents were exposed to the emotion-inducing campaign. Therefore these conclusions can only be taken into
account as short-term effects. It would also be interesting to see if there is a long-term effect, or to evaluate when in the campaigning period emotion inducing campaign material would be useful.
Another limitation of this research is that it used a sample of 200 respondents that were not a representative sample of society. It was not possible to collect a more
representative sample due to a rather short time span and because there was no funding. But using a more representative sample would increase the generalizability of the results of this research. It would also be interesting to do a cross-country research on this topic to evaluate if the findings were typical for the Dutch case or that the same results can be found in countries with a similar political system. However, this research is a first step in measuring the effect of negative emotions in political campaigns on voting intention in a country with a multi-party system. A suggestion for future research would also be to focus on evaluating whether a respondent’s party choice changes after being exposed to an emotion-inducing political campaign, so if emotions in campaigns could lead to a vote-switching effect.
28
Bibliography
Allison, P. D. (1999). Multiple Regression: A primer. Thousand Oaks, California, United States of America: Pine Forge Press.
Andeweg, R., & Irwin, G. (2005). Governance and Politics of the Netherlands . Basingstoke: Palgrave Macmillan.
Ansolabehere, S. D., Iyengar, S., & Simon, A. (1999). Replicating Experiments Using Aggregate and Survey Data: The Case of Negative Advertising and Turnout. The American
Political Science Review , 4, 901-909.
Appel, M., Gnambs, T., & Maio, G. (2012). A short measure of the need for affect. Journal of
personality assessment , 94 (4), 418-426.
Arkes, H. (1993). Can emotion supply the place of reason? In G. Marcus, & R. Hanson,
Reconsidering the democratic public (pp. 287-305). University Park: Pennsylvania State
University Press.
Blais, A., & Carty, R. (1990). Does proportional representation foster voter turnout?
European Journal of Political Research , 2, 167-181.
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.
Brady, H., Verba, S., & Schlozman, K. (1995). Beyond SES: A resource model of political participation. American Political Science Review , 89 (2), 271-294.
Breckler, S. (1993). Emotion and attitude change. In M. Lewis, & J. Haviland, Handbook of
emotions (pp. 461-473). New York: Guilford Press.
Chang, C. (2001). The impacts of emotion elicited by print political advertising on candidate evaluation. Media psychology , 3 (2), 91-118.
29
Crepaz, M. (1990). The impact of party polarization and postmaterialism on voter turnout.
European Journal of Political Research , 18 (2), 183-205.
Damasio, A. (2008). Descartes' error: Emotion, reason and the human brain. Random House.
Djupe, P., & Peterson, D. (2002). The impact of negative campaigning: Evidence from the 1998 senatorial primaries. Political Research Quaterly , 845-860.
Djupe, P., & Peterson, D. (2005). When primary campaigns go negative: the determinants of campaign negativity. Political Rsearch Quarterly , 58 (1), 45-54.
Duff, B., Hammer, M., Park, W., & White, I. (2007). Good excuses: Understanding who votes with an improved turnout question. Public Opinion Quarterly , 67-90.
Duff, B., Hanmer, M., Park, W., & White, I. (2007). Good excuses: Understanding who votes with an improved turnout question. Public Opinion Quarterly , 71 (1), 67-90.
Duverger, M. (1954). L'Influence des Systemes Electoraux sur la Vie Politique. Paris: Armand Colin.
Franklin, M. (2004). Voter turnout and the dynamics of electoral competition in established
democracies since 1945. Cambridge University Press.
Freedman, P., & Goldstein, K. (1999). Measuring Media Exposure and the Effects of Negative Campaign Ads. American Journal of Political Science , 43 (4), 1189-1208.
Geer, J. (2006). In Defense of Negativity: Attack Ads in Presidential Campaigns. Chicago: Chicago University Press.
Geys, B. (2006). Explaining voter turnout: A review of aggregate-level research. Electoral
30
Gray, J. A. (1990). Brain Systems that Mediate both Emotions and Cognition. Cognition and
Emotion , 4 (3), 269-288.
Haddock, G., & Huskinson, T. (2004). Individual differences in attitude structure. In G. Haddock, & G. Maio, Contemporary perspectives on the psychology of attitudes (pp. 35-56). Hove, U.K.: Psychology Press.
Haddock, G., Maio, G., Arnold, K., & Huskinson, T. (2008). Should persuasion be affective or cognitive? The moderating effects of need for affect and need for cognition. Personality
and Social Psychology Bulletin , 34 (6), 769-778.
Hansen, K., & Pedersen, R. (2008). Negative campaigning in a multiparty system.
Scandinavian Political Studies , 31 (4), 408-427.
IDEA International Institute for Democracy and Electoral Assistance. (sd). Voter turnout data
for the United States. Opgeroepen op May 25, 2014, van http://www.idea.int:
http://www.idea.int/vt/countryview.cfm?CountryCode=US
Kahn, K., & Kenney, P. (1999). Do negative campaigns mobilize or suppress turnout? Clarifying the relationship between negativity and participation. American Political Science
Review , 877-889.
Kaid, L., & Johnston, A. (1991). Negative versus positive television advertising in US presidential campaigns, 1960-1988. Journal of communication , 41 (3), 53-64.
Karp, J., & Brockington, D. (2005). Social desirability and response validity: A comparative analysis of overreporting voter turnout in five countries. Journal of Politics , 67 (3), 825-840.
Krupnikov, Y. (2011). When does negativity demobilize? Tracing the conditional effect of negative campaigning on voter turnout. American Journal of Political Science , 55 (4), 797-813.
31
Lau, R. R., & Pomper, G. M. (2001). Effects of negative campaigning on turnout in U.S. Senate elections, 1988-1998. Journal of Politics , 63, 804-819.
Lau, R. R., & Pomper, G. R. (2002). Effectiveness of Negative Campaigning in U.S. Senate Elections. American Journal of Political Science , 46 (1), 47-66.
Lau, R., & Pomper, G. (2004). Negative Campaigning. An Analysis of the U.S. Senate
Elections. Lanham: Rowman & Littlefield.
Lau, R., Sigelman, L., & Rovner, I. (2007). The Effects of Negative Political Campaigns: A Meta-Analytic Reassesment. Journal of Politics , 69 (4), 1167-1209.
Lerner, J., & Keltner, D. (2001). Fear, anger and Risk. Journal of Personality and Social
Psychology , 81 (1), 146-159.
Maio, G., & Esses, V. (2001). The need for affect: Individual differences in the motivation to approach or avoid emotions. Journal of personality , 583-614.
Mair, P. (2008). Electoral Volatility and the Dutch Party system: A Comparative Perspective.
Acta Politica , 43, 235-253.
Marcus, G. E., & Mackuen, M. B. (1993). Anxiety, Enthusiasm, and the Vote: The Emotional Underpinnings of Learning and Involvement During Presidential Campaigns. The American
Political Science Review , 87 (3), 672-685.
Miller, A. (2006). Watching viewers watch TV: Processing live, breaing, and emotional news in a naturalistic setting. Journalism and Mass Communication Quarterly , 511-529.
Nabi, R. L. (1999). A Cognitive-Functional Model for the Effects of Discrete Negative Emotions on Information Processing Attitude Change, and Recal. Communication Theory , 9 (3), 292-320.
32
Nabi, R. L. (2002). Anger, fear, uncertainty, and attitudes: a test of the cognitive- functional model. Communication Monographs , 204-216.
Nabi, R. L., & Wirth, W. (2008). Exploring the Role of Emotion in Media Effects: An Introduction to the Special Issue. Media Psychology , 11 (1), 1-6.
Namkoong, K., Fung, T. K., & Scheufele, D. A. (2012). the Politics of Emotions: News Media Attention, Emotional Responses, and the Participation During the 2004 U.S. Presidential Election. Mass Communication and Scoiety , 15 (1), 25-45.
Powell Jr., G. (1986). American voter turnout in comparative perspective. The American
Political Science Review , 17-43.
Rosenstone, J., & Hansen, J. (1993). Mobilization, Participation and Democracy in America. New York: Macmillan.
Scherer, K. (1984). On the nature and function of emotion: A component process approach. In K. Scherer, & P. Ekman, Approaches to emotion (pp. 293-318). Hillsdale: Erlbaum.
Simon, H. (1955). A behavioral model of rational choice. The quarterly journal of economics
, 69 (1), 99-118.
Smith, A., & Ellsworth, P. C. (1985). Patterns of Cognitive Appraisal in Emotion. Journal of
Personality and Social Psychology , 48 (4), 813-838.
Swanson, D., & Mancini, P. (1996). Politics, media, and modern democracy: An international
study of innovations in electoral campaigning and their consequences. Greenwood Publishing
Group.
Tiedens, L., & Linton, s. (2001). Judgment under emotional certainty and uncertainty: the efects of specific emotions on information processing. Journal of personality and social
33
Valentino, N. A., Brader, T., Groenendyk, E. W., Gregorowicz, K., & Hutchings, V. L. (2011). Election Night's Alright for Fighting: the Role of Emotions in Political Participation.
The Journal of Politics , 73 (1), 156-170.
Valentino, N., Hutchings, V., Banks, A., & Davis, A. (2008). Is a worried citizen a good citizen? Emotions, political information seeking, and learning via the Internet. Political
Psychology , 247-273.
Verba, S., & Nie, N. (1972). Participation in America. New York: Harper & Row.
Walter, A. (2010). Negatieve campagnevoering in de Nederlandse consensusdemocratie: de ontwikkelingen sinds Fortuyn. Res Publica (3), 293-213.
Walter, A. (2012). Negative campaigning in Western Europe: Beyond the vote-seeking
perspective.
Wattenberg, M. P., & Brians, C. L. (1999). Negative Campaign Advertising: Demobilizer or Mobilizer? The American Political Science Review , 93 (4), 891-899.
Zwakhals, S., Giesbers, H., & Deuning, C. (2012, September 14). Opkomstpercentage Tweede
Kamer verkiezingen 2012. Opgeroepen op May 25, 2014, van http://www.zorgatlas.n:
http://www.zorgatlas.nl/beinvloedende-factoren/sociale-omgeving/stemgedrag/opkomstpercentage-2012/
34
Appendix I
Neutral
35
36
Appendix II
Questionnaire16
- What do you feel is currently the most important issue in national politics? - What do you feel is currently the most important issue in European politics?
Answer the following questions (completely disagree – completely agree) - I am interested in politics
- I feel like I know enough about politics to make an informed vote - I trust the government
Answer the following statements (completely disagree – completely agree) - It is important for me to be in touch with my feelings.
- I think that it is important to explore my feelings. - I am a very emotional person.
- It is important for me to know how others are feeling. - Emotions help people get along in life.
- Strong emotions are generally beneficial
- I feel that I need to experience strong emotions regularly.
- I approach situations in which I expect to experience strong emotions - I feel like I need a good cry every now and then.
- I like to dwell on my emotions. - We should indulge our emotions.
- I like decorating my bedroom with a lot of pictures and posters of things emotionally significant to me
16 The original questionnaire was written in Dutch but is translated into English, making it understandable for a
37
- The experience of emotions promotes human survival. - I do not know how to handle my emotions, so I avoid them.
- I find strong emotions overwhelming and therefore try to avoid them.
- Emotions are dangerous—they tend to get me into situations that I would rather avoid. - I would prefer not to experience either the lows or highs of emotion.
- If I reflect on my past, I see that I tend to be afraid of feeling emotions.
- I would love to be like “Mr. Spock,” who is totally logical and experiences little emotion.
- I have trouble telling the people close to me that I love them. - Displays of emotions are embarrassing.
- Acting on one’s emotions is always a mistake.
- I am sometimes afraid of how I might act if I become too emotional. - Avoiding emotional events helps me sleep better at night.
- I wish I could feel less emotion.
- People can function most effectively when they are not experiencing strong emotions.
Answer the following questions (completely disagree – completely agree) - When I vote, I base my choice solely on factual information
- When I vote, I make my choice by relying on my emotions and feelings
Each respondent will be randomly assigned to a political campaign poster (neutral, fear or anger).
When talking to people about elections, we often find that a lot of people are not able to vote because they are sick, or they don’t have time.
38
- How about you, if the European elections were held tomorrow, how likely is it that you would vote? (10-point scale; I would definitely not vote – I would definitely vote) - If the European elections were held tomorrow, which party would you vote for? - If the European elections were held tomorrow, would you go and vote? (no - maybe -
yes)
- If the European elections were held tomorrow, how likely is it that you would vote for (name of each of the 11 parties)
You just were exposed to a campaign poster. Thinking back to this moment, how did you feel?
- Angry (completely disagree – completely agree) - Afraid (completely disagree – completely agree) - Neutral (completely disagree – completely agree)
- What is your gender? (male/female) - What is your age? (6 categories)
- What is your highest completed education? (4 categories) - What is your annual income? (4 categories)
39
Appendix III
Multicollinearity tables
Model Tolerance VIF
Anger .506 1.974
Interaction .329 3.040
Need for affect .485 2.061
Dependent variable: Voting intention N= 129 (anger and neutral)
Model Tolerance VIF
Fear .532 1.878
Interaction .395 2.530
Need for affect .527 1.989
Dependent variable: Voting intention N= 135 (fear and neutral)
40
Appendix IV
Manipulation check
Table 1: T-test for mean difference independent variable fear
Campaign 95% CI for Mean
Difference
Fear Neutral
or Anger
Mean SD Mean SD Mean Difference Sig. Lower Upper df
How afraid were the respondents? 2,65 ,997 2,19 ,609 -,461* .000 -,686 -,237 19 8 ** P < .05. ** p < .01 Fear: N = 69
Neutral and Anger: N = 131
Table 2: T-test for mean difference independent variable anger
Campaign 95% CI for Mean
Difference
Anger Neutral
or Fear
Mean SD Mean SD Mean Difference Sig. Lower Upper df
How angry were the respondents? 3,29 1,497 2,21 1,373 -1,077* .000 -1,499 -,656 19 8 * p < .01 Anger: N = 65
41
Appendix V
Bivariate regression analyses
Table 3
Bivariate regression analysis for anger and fear
Variables Coefficient Std. Dev. Sig. R2
Fear campaign .723 .541 .183 .009
Anger campaign -.623 .550 .258 .006
Dependent variable: When we ask people about elections, we often see that people do not vote because they were sick or did not have any time. This could also happen to you of course. If the European elections were held tomorrow, how likely is it that you would vote on a scale from 0 to 10?
N= 135 (fear and neutral) N= 129 (anger and neutral) ** P < .05
42
Appendix VI
The moderation effect of the need for affect
Table 6
Interaction effect for fear
Variables Coefficient Std. Dev. Sig. R2
Fear Campaign .950 .029 .195 .012
Need for affect -.002 .016 .889
Interaction -.015 .029 .597
Dependent variable: When we ask people about elections, we often see that people do not vote because they were sick or did not have any time. This could also happen to you of course. If the European elections were held tomorrow, how likely is it that you would vote on a scale from 0 to 10?
N= 200 ** P < .05 * P < .10
Table 7
Interaction effect for anger
Variables Coefficient Std. Dev. Sig. R2
Anger -.985 .763 .198 .011
Interaction .020 .028 .472
Need for affect -.015 .016 .365
Dependent variable: When we ask people about elections, we often see that people do not vote because they were sick or did not have any time. This could also happen to you of course. If the European elections were held tomorrow, how likely is it that you would vote on a scale from 0 to 10?
N = 200 ** P < .05 * P < .10
43
Appendix VII
Logistic regression analyses for fear, anger and the moderator variable
Table 8
Logistic regression analysis fear and moderator
Independent variable B S.E. Wald Sig. Exp(B)
Fear .887 .464 3.660 .056* 2.428
Interaction -.013 .018 .585 .444 .987
Need for affect -.005 .009 .348 .555 .995
Constant .397 .252 2.482 .115 1.448
Cox & Snell R2 .032
Nagelkerke R2 .044
Dependent variable: If the European elections would be held tomorrow, would you vote? 0 = no/maybe and 1 = yes
N = 200 ** P < .05 * P < .10
Table 9
Logistic regression analysis anger and moderator
Independent variable B S.E. Wald Sig. Exp(B)
Anger -.504 .441 1.308 .253 .604
Interaction .000 .016 .003 .958 .999