Negative campaigning: time to face the consequences
Estimating the effects of different dimensions of negative campaigning on the likeability of the target, the likeability of the sender and political cynicism.
Hester van Soest 10553363 Master’s Thesis
Graduate School of Communication Political Communication
Dhr. Alessandro Nai 27-6-2019
Abstract
Negative campaigning has been thoroughly researched in the United States; however, the U.S. is not the only country where negative campaigning is present. Negative campaigning is also commonly used in multiparty systems such as the Netherlands. Research has lacked to investigate negative campaigning in a Dutch context. Furthermore, little attention has been given to the different dimensions of negative campaigning, such as a difference in aim or a difference in tone. Negative campaigning is presumed by politicians and campaign professionals to work in altering the view of the opponent, however, it has unintended effects as well. Thus, this study aims to investigate the effects of different dimensions of negative campaigning (aim: issue vs. personal/tone: civil vs. uncivil) on the likeability of the target (intended), the likeability of the sender and political cynicism (unintended). Furthermore, the research on the influence of personal attitudes on these expected effects is remarkably scarce. Therefore, this study also investigates the moderating effect of populist attitudes on the effects found on political cynicism. No significant difference was found between either dimension of negative campaigning on the likeability of the target/sender. It was, however, found that a personal attack results in more political cynicism than an issue attack. An uncivil attack also results in more political cynicism than a civil attack. There was no evidence found for the moderation of populist attitudes. The results show, that different dimensions of negative campaigning do not have the desired effects but do have the unintended effect of increasing political cynicism. This leads us to advise politicians to be aware of the consequences of negative campaigning. They should not just focus on gaining votes but should take into account their important role in society.
Keywords: Negative campaigning; aim; tone; likeability of the sender; likeability of target; political cynicism; populist attitudes.
Introduction
When one looks at trends in politics and political marketing, it is almost impossible not to discuss the United States. Most of the important changes and trends in politics first arise there (Baines, Scheucher & Plasser, 2001). A rise in negative campaigning was noted by scholars in the United States as early as the 1990s (Kaid & Johnston, 1991; Jamieson, 1992). Negative campaigning is defined as ‘criticism leveled by one candidate against another’ (Geer, 2006). The rise in negativity seems to continue with the 2012 U.S. presidential election being described as the most negative one yet (Hill, Capella & Cho, 2014).
A lot of the examples of negative campaigning are indeed from the United States, but this does not mean that negative campaigning is not used in the rest of the world. Elmelund-Præstekær (2010) states that the American two-party system seems to generate more negativity than the European multiparty systems, nonetheless negative campaigning still exists in Europe. One of the countries where negative campaigning is present is the Netherlands (Walter & van der Brug, 2013). Very few scholars have researched negative campaigning there (Walter, 2014a). The U.S. and the Netherlands are each other’s complete opposites when it comes to the political system; in the U.S. the democrats and republicans compete for the presidency, whereas in the Netherlands no less than 28 parties participated in the last national elections. Due to these immense differences in political systems it is interesting to research if the effects of negative campaigning, found in American research, can also be found in the Netherlands.
When looking at the presence of negative campaigning in Dutch politics, the right-wing party PVV is one of the more well-known examples, with party leader Geert Wilders attacking other parties on a daily basis (Vliegenthart & Walter, 2010). Newcomers in Dutch politics, like DENK and Forum voor Democratie, take this negativity to a whole new level with very personal attacks. DENK leader Tunahan Kuzu shamed the speaker of the house of representatives
(Khadija Arib), on her becoming too ‘verkaasd’1 and forgetting her Moroccan heritage
(Kouwenhoven & Logtenberg, 2017). More established Dutch parties are also taking a turn for the negative; Klaas Dijkhoff (VVD) called Rob Jetten (D66) a ‘Klimaat drammer’2, saying he
was whining too much about climate change and should focus on other issues (Jonker & de Winther, 2019). These examples show that negative campaigning is very present in Dutch politics (Walter, van der Brug & van Praag, 2014), and the effects previously found in predominantly U.S. research should therefore also be investigated in the Netherlands.
Negativity campaigning can have different dimensions and some of these dimensions are especially important when looking at negativity in the Netherlands. There can be differences in the aim of the negativity or the tone of the attack (Auter & Fine, 2016; Fridkin & Kenney, 2008). The differences in aim are often either an issue attack or a personal attack, and the differences in tone are often referred to as being either civil or uncivil (Kenski, Filer & Conway-Silva, 2018; Auter & Fine, 2016). These two dimensions are important in the Netherlands, since multiparty systems need to take coalition forming into account (Haselmayer & Jenny, 2018). When really uncivil attacks or really personal attacks are used during the campaign, politicians might burn bridges which makes it difficult to form coalitions. Coalition forming is becoming harder in the Netherlands, with the last formation of government taking a record time of 225 days. The effects that certain types of negative campaigning can have on coalition forming is precisely the reason why this is important to study in the Netherlands.
Negative campaigning can have intended and unintended effects (Baumgartner, 2013). One of the intended effects of negative campaigning could be to change the opinion about the target of the campaign. That is why this experiment will investigate the effect of the different dimensions of negative campaigning on the likeability of the target. While trying to change the likeability of their target, politicians can also influence their own likeability in a negative way,
1 Literally: becoming too much of a ‘cheese head’, meaning a person is becoming very Dutch. 2 Literally: climate whiner, ‘whining’ too much about climate change issues.
because the constituency can perceive attacks as being too harsh or unfair (Roese & Sande, 1993). This ‘backlash-effect’ is often not taken into account when politicians decide to go negative. Scholars have also given less attention to this effect since they predominantly focus on the intended effects of negative campaigning (Lau, Sigelman & Rovner, 2007). Another unintended effect can arise in a more ‘long-term’ way. Negativity has been shown to increase political cynicism, which is not an intended goal of politicians (Dardis, Shen & Edwards, 2008). Political cynicism is ‘the extent to which people hold politicians and politics in disrepute’ (Agger, Goldstein & Pearl, 1961). Politicians should set an example for citizens and should in no way reinforce cynicism. Research on the intended and especially the unintended effects of these different types of negative campaigning is scarce and needs to be extended to fill the gap in current literature. Moreover, research is needed to warn and educate politicians about the harm they might cause society when using certain types of negative campaigning.
Just like political cynicism, populism has seen a rise in recent years (Inglehart & Norris, 2016), both of these phenomena have a negative influence on the state of our democracy. This is why not only the direct effects of negativity on political cynicism need to be examined, but also the role populist attitudes play in reinforcing these effects. Research has focused on the way personal convictions can strengthen the effects of negative campaigns (Rooduijn, van der Brug, de Lange & Parlevliet, 2017). Studies have shown that populists become more cynical towards politics than non-populists after being exposed to negativity (Rooduijn et al., 2017). It is, therefore, interesting to see how populist attitudes moderate the direct effects found of negative campaigning on political cynicism. These results could again warn politicians of the unintended effects of negative campaigning. This leads to the following research question:
RQ: Do different forms of negative campaigning (aim: issue vs. personal; tone: civil vs. uncivil) have different effects when it comes to the likeability of the target (intended), the
likeability of the sender and political cynicism (unintended) and are the effects found on political cynicism stronger for people with populist attitudes?
Besides the theoretical relevance, this research will also be of importance to politicians and campaign professionals. This study can help them decide if negative campaigning could have a desirable effect and what specific form of negative campaigning would work best for them. Furthermore, it could warn them of the unintended effects that can also arise when using negative campaigning. And, if people with populist attitudes are more likely to be affected by certain dimensions of negative campaigning, this can help politicians and campaign professionals to gain more insight into their constituency.
To test the research question, an experiment will be executed through Qualtrics using mock tweets, to test the effects of the aim of the attack (issue vs. personal) and the tone of the attack (civil vs. uncivil) on both the likeability of the target (intended), the likeability of the sender and political cynicism (unintended). Tweets are shown by multiple scholars to be very useful in political campaigning (Lee & Xu, 2018; Gross & Johnson, 2016) and will therefore also be used as a negative campaigning platform in this research.
Theoretical Framework Negative campaigning
In the following section negative campaigning and its different dimensions (aim: issue vs. personal/tone: civil vs. uncivil) will be discussed. Furthermore, their effects on the likeability of the target (intended), the likeability of the sender and political cynicism (unintended) will be examined. Moreover, it will be discussed how the main effects found on political cynicism are expected to be stronger for people with populist attitudes.
The concept of negative campaigning is thoroughly researched. However, there is still no consensus on its effects and importance (Lau et al., 2007). Pattie, Devner, Johns and Mitchell (2011) look at the distinction between positive and negative campaigning. They conclude that campaigns combine both positive and negative elements and that both of these are functional in campaigns running up to elections. Without positive rhetoric, voters will not know what they vote for and without negative rhetoric, the claims of opponents are not thoroughly tested. Due to all the different articles about negative campaigning, it is difficult to give a precise definition of the concept. Geer (2006) defined negative campaigning as ‘criticism leveled by one candidate against another’. This definition is quite broad and not necessarily precise but does cover the whole concept of negative campaigning. Walter (2014b) also has difficulty defining negative campaigning. She ascribes this difficulty to the different dimensions of negativity. This article will work with the general definition by Geer (2006), since it covers the whole concept, but will also dive further into the different dimensions of negative campaigning.
Nuances in negative campaigning: aim
Previously, the general definition by Geer (2006) of negative campaigning has been discussed. When trying to specify this definition more, two dimensions are important: aim and tone. As explained in the introduction, these dimensions are especially important when looking at negativity in multiparty systems, because in those systems coalition potential should be taken into account and this can be compromised by very personal or very uncivil attacks. Auter and Fine (2016) look at aim and make a distinction between issue attacks and personal attacks, where personal attacks ‘focus their vitriol on the personal traits of an opponent, deriding them as being ‘untrustworthy’, ‘a bad leader’, or describing character flaws or past personal misdeeds.’, and issue attacks ‘focus on the substance of the target’s policy preferences, agenda or past political actions.’ What Auter and Fine (2016) see as issue attacks, Fridkin and Kenney
(2011) define as relevant attacks, personal attacks are defined as irrelevant attacks. This article will make the same distinction that Auter and Fine (2016) made, the attack can be either a personal attack or an issue attack.
Nuances in negative campaigning: tone
Apart from a distinction in the aim of the attack (personal or issue), a difference in tone can also be found in examples of negative campaigning. Often scholars refer to civility and incivility when discussing a difference in tone. Kenski et al. (2018) try to give a clear definition of civility and incivility but state that (in)civility is notoriously difficult to define, due to people perceiving (in)civility in different ways. Taking into account previous definitions, they define civility as ‘speech that shows respect for others and facilitates open, democratic discourse’ and incivility as ‘features of discussion that convey an unnecessarily disrespectful tone toward the discussion forum, its participants or its topics’. According to Brooks and Geer (2007), civility has to at least involve some notion of mutual respect and courteousness. Incivility is operationalized by them as ‘claims that are inflammatory and superfluous.’ Hence, this research will define civil campaigning as attacking the opponent, but it should still be respectful and courteous. Uncivil negative campaigning will be defined as attacking the opponent with unnecessary disrespectfulness and inflammatory comments.
The effect on likeability
The main reason that politicians and their campaign teams use negative campaigning is in the end to gain votes. However, this is not the only subject that they can influence. Multiple scholars have looked into the effect of negative campaigning on the likeability of both the sender and the target. For example, Pattie et al. (2011) state that negative campaigning makes the party that does the campaigning less likeable. This could be seen as a backlash effect; the
likeability of the sender decreases. Furthermore, Roese and Sande (1993) found that with negative campaigning the likeability of the sender was negatively affected because the constituency perceived the attack as being too harsh, whereas the likeability of the target stayed almost the same. This backlash effect is unintended and does not have the desired effect for the politicians using negative campaigning (Roddy & Garramone, 1988) since it is, in the end, their goal to decrease the likeability of the opponent and not their own likeability. Nevertheless, negative campaigning has intended effects as well. Roddy and Garramone (1988) found that negative political ads decrease the evaluation of the target. Furthermore, Baumgartner (2013) found that during the 2012 campaign the feelings towards Mitt Romney decreased when he sent out an attack and when he was attacked, showing both the backlash effects for the Romney-campaign and the intended effects of the Obama-Romney-campaign.
When we refocus on the different dimensions of negative campaigning, with aim (issue vs. personal) as the first dimension, scholars found that these differences in aim have different effects on likeability (Roddy & Garramone, 1988). Fridkin and Kenney (2011) find that issue-based attacks seem to have the intended effect of changing the voters’ view of the target. On the other hand, personal attacks seem to create a backlash effect, where the opinion towards the sender becomes more negative. Pfau and Burgoon (1989) show that issue-based attacks are better at changing attitudes than character attacks. Roddy and Garramone (1988) also show that issue-based attacks work best for decreasing the likeability of the target. Furthermore, they concluded that personal attacks decrease the likeability of the sender since they are perceived as being less justifiable than an issue attack. These previously found differences between issue and personal attacks, where issue attacks decrease the likeability of the target and personal attacks decrease the likeability of the sender, lead us to the following hypotheses:
H2: A personal attack will lead to less likeability of the sender than an issue attack
The second dimension of negative campaigning: tone (civil vs. uncivil), also has different effects on the likeability of both the sender and the target. Fridkin and Kenney (2008) find that uncivil messages result in an increase in negative evaluations of the target. In 2011 they again looked into the effect of (in)civility on likeability and found in corroboration with their earlier results that incivility helps senders decrease the perception of their target. However, this time around they also found a backlash effect. Incivility helps to alter the perceptions of the opponent, which is intended but also creates the unintended effect of negatively altering the perceptions of the sender (Fridkin & Kenney, 2011). The conclusions by Fridkin and Kenney (2008; 2011) lead us to expect the following:
H3: An uncivil attack will lead to less likeability of the target than a civil attack H4: An uncivil attack will lead to less likeability of the sender than a civil attack
The effect on political cynicism
Besides the intended effects on the likeability of the target, and the unintended effect on the likeability of the sender, negative campaigning can also have other effects. These are not focused on the evaluation of politicians but rather alter the general view of the political system. One of these factors is political cynicism. Political cynicism was first introduced by Agger et al. (1961) and defined by them as ‘the extent to which people hold politicians and politics in disrepute, the extent to which these words symbolize something negative rather than something positive.’ This concept has been thoroughly researched in the past years and researchers have noted a rise in political cynicism, especially in western societies (Pattyn, van Hiel, Dhont and Onraet, 2012).
Multiple scholars looked into how negative campaigning can influence political cynicism. Toros (2017) found that negative political messages negatively alter the trust levels of citizens. Trust in politics is one of the components that is often used to define political cynicism (Boukes & Boomgaarden, 2015). Dardis et al. (2008) found that high exposure to negative ads nurtures the perception that politics are complex, which leads individuals to believe that they cannot change much and therefore increases political cynicism. Furthermore, Lau et al. (2007) conclude from their meta-analysis that negative campaigning tends to lower feelings of political efficacy and trust in government. This can in turn, according to the studies that they analyzed, lead to an increase in political cynicism.
When, once again, focusing on the first dimension of negative campaigning: aim (issue vs. personal), it is interesting to mention Schenck-Hamlin, Procter and Rumsey (2000). They found that personal attacks lead to more cynical comments than issue attacks. However, they did not measure the full concept of political cynicism, but instead decided to measure ‘cynical comments’ made by participants (Dardis et al., 2008). To complement the research done by Schenck-Hamlin et al. (2000), and to fill the still existing gaps in the literature, Dardis et al. (2008) investigated the effect of a difference between issue attacks and personal attacks on political cynicism. They find that issue-based attacks lead to greater cynicism among the participants, which is contrary to the results found by Schenck-Hamlin et al. (2000). Since Dardis et al. (2008) use a scale that is more complete than just measuring cynical comments, this study will keep building on their conclusions. This leads us to the following hypothesis:
H5: An issue attack will lead to more political cynicism than a personal attack
When looking at the second dimension of negative campaigning (tone: civil vs. uncivil), it is interesting to note that multiple scholars have found evidence that negative campaigning
decreases trust in politics and in the end increases political cynicism (Dardis et al., 2008; Lau et al., 2007). Mutz and Reeves (2005) look specifically at the distinction between civil and uncivil negative campaigning and conclude that incivility decreases political trust, which leads to more political cynicism. This can be ascribed to negativity being perceived as even more negative when the attack is uncivil. Drawing on the conclusions by Mutz and Reeves (2005) the following is expected:
H6: An uncivil attack will lead to more political cynicism than a civil attack
Populist attitudes
Earlier studies found that personality traits, personal convictions and attitudes have an influence on different phenomena (Dawes et al., 2014; Brandstätter & Opp, 2014). Scholars have, for example, theorized that citizens with populist attitudes can respond differently to negative campaigning, then citizens without these attitudes (Rooduijn et al., 2017; Hameleers & Schmuck, 2017). Populist parties and their rhetoric are notoriously harsh and negative (Jagers & Walgrave, 2007). This harshness and negativity become especially clear when comparing populists to non-populists (Nai, 2018a). One could think that being negative could be harmful to politicians, but populists seem to gain a lot of votes by using these negative techniques (Rooduijn et al., 2017). An explanation for this could be that the constituency of populists respond extremely well to these types of messages (Wirz, 2018). Sniderman, Hagendoorn and Prior (2004) studied ‘galvanizing effects’: individuals are particularly affected by a message if they are already inclined to agree with this message. Since populists are notoriously harsh and negative in their rhetoric (Jagers & Walgrave, 2007), it is in line with ‘galvanizing effects’ to expect that populist voters will be particularly affected by harsh and negative messages
(Sniderman et al., 2004). In the context of this current study, it is therefore interesting to look at the moderating role populist attitudes can play in the main effects of negative campaigning.
Pattyn et al. (2012) found a correlation between political cynicism and authoritarianism, meaning that people who resent authority more, are also more cynical toward the political system. Since populists are per definition anti-authority and resent authority more than others (Jagers & Walgrave, 2007) it could be concluded that they will also be more cynical towards politics. Rooduijn et al. (2017) find that voters become more cynical towards politics after being exposed to populist messages, however, this result seems to only be present among populist voters. It could thus be expected that people with populist attitudes who are exposed to either type of negative campaigning, will become even more cynical towards politics than participants who score lower on populist attitudes. This expectation is in line with the research by Rooduijn et al. (2017), the following is therefore hypothesized:
H7: The direct effects found on political cynicism are even stronger for participants with populist attitudes.
Method Research design
This research aims to investigate the effect of different dimensions of negative campaigning (issue vs. personal; civil vs. uncivil) on the likeability of the target (intended), the likeability of the sender and political cynicism (unintended) and how the direct effects found on political cynicism are moderated by the populist attitudes of the participant. The ideal method to use is an experiment, since an experiment makes it possible to reveal causal relations between variables. Another important reason to use an experiment is that the researcher is able to manipulate the different conditions. To test the hypotheses, participants will be shown 1 out
of 3 tweets composed by the researcher (see appendix). This results in the following manipulations: 1) issue/civil attack, 2) issue/uncivil attack, 3) personal attack.
A between-subjects experimental design is used. The experiment contains two factors (aim and tone) which each have two levels (aim: issue vs. personal; tone: civil vs. uncivil).
Sample
This experiment uses a convenience sample. To gather respondents for this research, the Qualtrics link was distributed among friends, family, colleagues and acquaintances through WhatsApp, e-mail, Facebook, LinkedIn and Slack (platform to communicate with colleagues). Furthermore, close friends and family were asked to forward the survey to their friends, family, and colleagues, so the largest number of participants would be reached. The survey was distributed at the beginning of May 2019 and was kept open for approximately two weeks. A reminder was placed on Facebook after one week. Participants were also reminded through WhatsApp after a week and once again after a week and a half.
The number of responses reported was 194, however, not all participants finished the survey, these were therefore deleted from the dataset. This resulted in N = 156 participants. The average age of participants was 34, 58.30% was female and 41.70% was male. The participants varied in their highest finished level of education: elementary education (0.60%), secondary education (12.10%), vocational education (1.90%) and higher education (69.80%)3. All
participants were legally allowed to vote in the Netherlands. On average participants responded with M = 3.31 on the populist attitudes scale, from 1 (low on populist attitudes) too 7 (high on populist attitudes).
3 In the Dutch system elementary education responds to: primary school or no education; secondary education responds to: MAVO/VMBO, HAVO & VWO; vocational education responds to: MBO; higher education responds to: HBO, WO Bachelor & WO Master/doctorate.
Procedure
The participants were first introduced to the goal of the research, had to agree to the informed consent and answer questions, concerning their age, gender, highest finished level of education and legal ability to vote in the Netherlands. Afterward, they were asked to answer the ‘populist attitudes’-scale. Then, they were exposed to one of the three constructed tweets, which will be further explained in the next paragraph. Next, they answered the manipulation check and the three scales which measured (in order of appearance): the likeability of the target, the likeability of the sender and political cynicism. Finally, the participants were debriefed about the actual goal of the research and the fact that the tweets were fake and had the option to leave comments or remarks. The survey can be found in the appendix.
Experimental stimuli
To test the effects of different dimensions of negative campaigning, the participants were exposed to one of three experimental stimuli. These stimuli are all tweets by Jesse Klaver (GroenLinks) about Rob Jetten (D66). As shown by multiple scholars, it is often incumbents who are attacked and candidates far from the ‘ideological-center’ that are more likely to attack (Nai, 2018b). This is why in this experiment the attacker will be part of the opposition; Jesse Klaver and the target of the campaign will be current incumbent Rob Jetten. Both, Rob Jetten and Jesse Klaver are, compared to other Dutch politicians, not that ‘controversial’. The choice for these politicians was therefore made, to ensure existing opinions about the politicians would not influence the results too much. Multiple mock tweets were made and shown to approximately five Dutch citizens to decide on the best ones to use. This resulted in the following manipulations; Tweet 1 (issue/civil): the first tweet consists of an issue attack (climate change policy) in a civil manner. Tweet 2 (issue/uncivil): the second tweet also consists of an issue attack on climate change policy; however, this tweet is uncivil and harsh. Tweet 3
(personal): the final tweet is a personal attack against Rob Jetten, which does not focus on any policy issues, but just attacks Rob Jetten on being too young and inexperienced. The design was constructed in such a way that we are able to compare civil and uncivil tweets (tweet 1 vs. tweet 2) and issue vs. personal tweets (tweet 1 + tweet 2 vs. tweet 3). All the tweets were kept approximately the same length, and all have the same date and time stamp, in order to keep them as similar as possible. The tweets can be seen in the appendix of this research.
Manipulation check
To test if the manipulation with the tweets was successful, a manipulation check was executed. Participants were asked if the tweet they saw was negative, funny, mean, positive and believable on a scale from 0 to 100. The participants found the personal attack to be significantly more negative, meaner, less positive and less realistic than the issue attack. Furthermore, the participants found the uncivil attack to be significantly more negative, meaner, less positive and less realistic than the civil attack. No significant difference was found for ‘funny’. Results of the t-tests can be found in Table 1 (tweets aim) and Table 2 (tweets tone) in the appendix. The results show that the manipulation worked, the uncivil tweet and the personal tweet were both perceived to be significantly more negative and meaner, which is in line with the expectations.
Scales
The first scale that had to be measured was ‘populist attitudes’, which was measured using the same scale as Akkerman, Mudde and Zaslove (2014). They use eight different questions to measure populist attitudes in their research, these questions were all translated into Dutch and used in this research except for the following question: The political differences between the elite and the people are larger than the differences among the people. This question
was reviewed by multiple Dutch natives, including the researcher, as being too complicated and too difficult to translate to Dutch. To not confuse participants, it was decided not to use this question. The seven different items used can be seen in Table 3 in the appendix. The factor analysis shows one underlying dimension that has an Eigenvalue higher than 1 which explains 43.58% of the variance. Q1: ‘The politicians in the Dutch parliament need to follow the will of the people’ and Q7: ‘Interest groups have too much influence over political decisions’ both had a value below .60 and were not used to form a scale. The reliability test of the remaining items showed a Cronbach’s Alpha of a = .78 which could not be improved by deleting items.
The likeability of both the target (Rob Jetten) and the sender (Jesse Klaver) was measured using the scale as constructed by Reysen (2005). The items of the scale for likeability can be found in the appendix in Table 4. The questions were translated into Dutch and were slightly rephrased to contain the name of the target or the sender. This resulted in questions phrased as follows: Rob Jetten is friendly, Rob Jetten is likeable, Jesse Klaver is warm, Jesse Klaver is similar to me, etc. Both the factor analyses of the likeability of Rob Jetten and the likeability of Jesse Klaver showed two components with an Eigenvalue above 1. For the likeability of Rob Jetten, the first component explained 39.84% of the variance and the second component explained 22.27% of the variance. With the likeability of Jesse Klaver, the first component explained 49.08% of the variance and the second component explained 16.59% of the variance. Because of the large percentage of explained variance in both of the first components, a reliability test was done to see if the reliability of all these items together would still be high. The reliability test for the likeability of Rob Jetten (Q3 and Q5 had values below .60 and where therefor excluded from the reliability test and scale) showed a Cronbach’s Alpha of a = .80 and the reliability test for the likeability of Jesse Klaver (all items included) showed a Cronbach’s Alpha of a = .89, both scales could barely be improved by deleting items. This shows that even if a second component seems to exist at first, the reliability of the scale for the
likeability of the target and the likability of the sender is still very high, which makes us conclude that the concept of likeability is still unidimensional and can therefore be used as one scale in the calculations.
The final scale that was measured is political cynicism. The article by Boukes and Boomgaarden (2015) was used to construct this scale. Not all the questions used by Boukes and Boomgaarden (2015) were deemed to be useful. Also, it was decided not to distinguish between ministers, junior ministers, politicians and MP’s as done by Boukes and Boomgaarden (2015), but just to use the term politicians, as this was deemed easier to understand for participants. All the items used in the scale for political cynicism can be found in Table 5 in the appendix. These questions were also translated to Dutch. The factor analyses showed three components with an Eigenvalue above 1, with the first one explaining 40.32% of the variance, the second one explaining 10.20% of the variance and the third one explaining 9.29% of the variance. The values of items Q9 and Q10 were below .60 and could therefore not be used in the scale. Since the first component explained a relatively large part of the variance compared to the other components a reliability test was conducted to see if all the items together could still form a reliable scale. The reliability analyses of the items showed a Cronbach’s Alpha of a , which could barely be improved by deleting one of the items. This means that, even if three components seem to exist at first, this high Cronbach’s Alpha shows that political cynicism can still be measured as a unidimensional concept and can, therefore, be used in the calculations as one single scale.
Proposed data-analyses
To test the hypotheses for the main effects multiple independent t-tests will be executed. Besides the main effects, a moderation effect is also expected. To test this moderation, a linear regression analysis will be done.
Results Randomization checks
To test the randomization of gender among the different conditions a Chi Square test was executed for both tone attack and aim attack. It was shown that the variable gender was not significantly different in aim attack in level 1 (M = 1.58, SD = 0.50) compared to level 2 (M = 1.58, SD = 0.50), χ2 (1) > 0.01, p = .977. It was also shown that in tone attack gender was not significantly different in level 1 (M = 1.62, SD = 0.49) compared to level 2 (M = 1.54, SD = 0.50), χ2 (1) = 0.72, p = .395. This means that gender is equally distributed over the different conditions, the randomization of gender is successful.
To test the randomization of age among the different conditions another Chi Square test was executed. It was shown that the variable age was not significantly different in aim attack level 1 (M = 31.85, SD = 13.85) compared to level 2 (M = 35.40, SD = 15.86), χ2 (37) = 32.43, p = .683. Furthermore, it was shown that the variable age was not significantly different in tone attack in level 1 (M = 35.23, SD = 15.06) compared to level 2 (M = 35.58, SD = 16.82), χ2 (33) = 32.45, p = .494. This means that age is equally distributed over the different conditions, the randomization of age is also successful.
Analyses
To test the H1, H2 and H5 an independent sample t-test was done to see if a difference could be found between level 1 of aim attack (personal) and level 2 of aim attack (issue) on the dependent variables. The results of the independent sample t-test for H1 indicate that a personal attack (M = 3.80, SD = 0.84) does not result in significantly higher likeability of Rob Jetten than an issue attack (M = 3.83, SD = 0.73), t (154) = -0.23, p = .819. This means that the first hypothesis is not confirmed.
The results of the independent sample t-test for H2 indicate that a personal attack (M = 3.93, SD = 0.86) does not result in significantly lower likeability for Jesse Klaver than an issue attack (M = 4.06, SD = 0.96), t (154) = -0.80, p = .428. The second hypothesis is not confirmed.
The results of the independent sample t-test for H5 indicate that a personal attack (M = 3.94, SD = 1.02) results in significantly higher political cynicism than an issue attack (M = 3.48, SD = 0.82), t (154) = 3.05, p = .003. It can, therefore, be said that the fifth hypothesis is not confirmed, on the contrary, the opposite was found to be significant. A personal attack results in significantly higher political cynicism than an issue attack.
To test H3, H4 and H6, an independent sample t-test was done to see if a difference could be found between level 1 of tone attack (civil) and level 2 of tone attack (uncivil) on the dependent variables. The results of the independent sample t-test for H3 indicate that a civil attack (M = 3.90, SD = 0.76) does not result in a significantly higher likeability of Rob Jetten than an uncivil attack (M = 3.76, SD = 0.70), t (101) = 0.98, p = .328. The third hypothesis is not confirmed.
The results of the t-test for H4 indicate that a civil attack (M = 4.22, SD = 0.92) does not result in a significantly higher likeability of Jesse Klaver than an uncivil attack (M = 3.88, SD = 0.97), t (101) = 1.83, p = .070. The fourth hypothesis is not confirmed.
The results of the t-test for H6 indicate that a civil attack (M = 3.23, SD = 0.77) results in significantly lower political cynicism than an uncivil attack (M = 3.74, SD = 0.79), t (101) = -3.25, p = .002. It can therefore be said that the sixth hypothesis is confirmed, an uncivil attack results in significantly more political cynicism than a civil attack.
To test the final hypotheses, a linear regression was done with ‘populist attitudes’ as a moderator for the main effect of either aim attack or tone attack on political cynicism. No significant moderation can be found for the aim attack on political cynicism, b = -.12, p = .073. The same linear regression was done for ‘populist attitudes' as a moderator for the main effect
found on tone attack on political cynicism. No significant moderation was found for the tone attack on political cynicism, b = -.06, p = .499. The final hypothesis is not confirmed.
Conclusion & discussion
The aim of this study was to investigate the different dimensions of negative campaigning (aim: issue vs. personal/tone: civil vs. uncivil) and their effect on the likeability of the target (intended), the likeability of the sender and political cynicism (unintended) and how the direct effects on political cynicism are moderated by populist attitudes.
Based on the results of this study, the conclusion can be drawn that no significant difference can be found between the dimensions of negative campaigning (aim: issue vs. personal/tone: civil vs. uncivil) on the likeability of the target (Rob Jetten) or the likeability of the sender (Jesse Klaver). A significant difference between the dimensions of negative campaigning can, however, be found on political cynicism. A personal attack results in more political cynicism than an issue attack. Furthermore, an uncivil attack results in more political cynicism than a civil attack. There was no evidence found for the expectation that the effects on political cynicism are stronger for participants with populist attitudes.
The lack of evidence for H1, H2, H3 and H4 is not in line with the expectations as derived by the articles from multiple scholars about the effects of the different dimensions of negative campaigning on both the likeability of the sender and the likeability of the target (Roddy & Garramone, 1988; Fridkin & Kenney, 2008). Multiple explanations can be given for the lack of significant evidence. For example, the personal tweets and the uncivil tweets were perceived by participants to not be very realistic. Participants ascribed much higher values of realism, on a scale from 0 to 100, to the civil and the issue tweet compared to the uncivil and the personal tweet. Since these tweets were not perceived as realistic, participants are less likely to be influenced by them and this will, in turn, influence the results. Furthermore, it was noted
by multiple participants, both in the comments and mentioned directly to the researcher, that many of them did not know the target Rob Jetten. Rob Jetten has been the party leader of D66 for about a year but is still less known than his predecessor or his opponent Jesse Klaver. Multiple participants mentioned that this lack of knowledge about Rob Jetten led them to give average answers on the likeability scale since they had neither positive nor negative feelings towards him. Another explanation for the fact that there was no difference found between the different dimensions of negative campaigning on the likeability of Rob Jetten or Jesse Klaver could be that they are both known by participants (Jesse Klaver obviously more than Rob Jetten, as previously explained). Since they are both in the public eye, people often already have an opinion towards them, which makes it difficult to find a difference between participants after they have been exposed to either dimension of negative campaigning. These two politicians were, however, consciously picked since they are ideologically similar, and one is in the opposition and the other one is an incumbent, which are all determining factors to go negative (Nai, 2018b). The decision to use Rob Jetten and Jesse Klaver was consciously made and it would not be recommended (at least in a Dutch context) to use other politicians, since a lot of the other well-known politicians are quite controversial, which makes it even more difficult to measure a difference between groups.
This study also does not provide evidence for H5, on the contrary, a significant effect in the opposite ‘direction’ is found; personal attacks lead to higher political cynicism than issue attacks. This result is contrary to the results found by Dardis et al. (2008), on which the hypothesis was based. As discussed in the theoretical framework, only two studies investigated the effect of either an issue attack or a personal attack on political cynicism. Due to the inconsistency in the previous studies, it was decided to follow the study by Dardis et al. (2008), which measured political cynicism in approximately the same way as done here. This study found, contrary to the expectations, the same results as Schenck-Hamlin et al. (2000); personal
attacks lead to more political cynicism than issue attacks. This discrepancy could be explained when one thinks about how personal negativity is perceived by citizens. Personal attacks could be seen as politicians just bickering amongst each other instead of doing their jobs. This could make citizens lose their trust in government, which in turn leads to more political cynicism (Dardis et al., 2008; Lau et al., 2007). As expected, uncivil negative campaigning does lead to significantly more political cynicism than civil negative campaigning. The explanation previously given, could also explain the significant effect found for uncivil negative campaigning on political cynicism. The more uncivil the attack is the more it is perceived as bickering among politicians and this in turn leads to a decline in the trust in government and therefore, to an increase in political cynicism (Dardis et al., 2008; Lau et al., 2007).
Moreover, this study does not provide significant evidence for the moderating effect of populist attitudes on the main effects found on political cynicism, which is contrary to the expectations given by previous research (Rooduijn et al., 2017). One of the reasons for this lack of evidence could be that the sender (Jesse Klaver) is a left-wing politician. He is the leader of GroenLinks and has therefore arguably other opinions than populist parties and populist voters in the Netherlands. This could mean that people with populist attitudes do not resonate with him at all and therefore are not more influenced by him sending out negative campaigning, than non-populists.
Multiple recommendations for future research can be given, both based on the limitations of this study and based on still existent gaps in the literature. To ensure in the future that the knowledge about a politician does not interfere with the results, it would be advisable to also measure the knowledge that the participant has about the politician. This gives researchers the ability to distinguish between groups of participants which results in more internal validity, instead of the participant just filling out the questions ‘randomly’. Another recommendation for future research, in line with the previous one, would be to first introduce
the politicians with an introduction text before exposing participants to the manipulation. This could either spark already existing knowledge about the politicians or could introduce the politicians in such a way that participants do not just give average answers. Furthermore, experiments could be done using not existing politicians, this will ensure more internal validity but does make the results less transferable to real-life situations and makes it harder to give actual advice to politicians about the effects of negative campaigning. Additionally, in line with the explanation given for the lack of evidence for H7, it would be interesting to see if a moderation effect of populist attitudes can be found if the sender of the negative campaign is also populist, instead of a left-wing politician. It would also be interesting to dive even further into the concept of attitudes or character traits as a moderator for the effects of negative campaigning. Research on, for example, the big-five personality traits is ‘booming’ at the moment and could also be of great importance in studying the effects of negative campaigning. If this research is extended and personality traits are starting to play a bigger part in political marketing, this could result in ads targeted on, and specifically tailored to certain citizens, which will help professionalize political marketing even more. It would also be advisable for future research, to test even more dimensions of negative campaigning to conclude in the end which type of negative campaigning has which desired or undesired effects. All these recommendations will close the still existing gaps in literature even more and will result in the ability to give solid advice to campaign professionals and politicians.
The aim of this paper was not only to close the existing gaps in the literature, which has been done by both the significant and the non-significant results found in this study, but also to help politicians and campaign professionals in their use of negative campaigning. As mentioned before, negative campaigning is commonly used in politics, even though there is not that much convincing evidence that it actually achieves the desired goals (Lau et al., 2007). This study shows that different dimensions of negative campaigning do not result in the desired effects of
changing the likeability of the target. On a positive note for the sender; this study did not find a backlash effect either, even though multiple other studies did (Fridkin & Kenney, 2011; Roddy & Garramone, 1988; Pfau & Burgoon, 1989). The effects found on political cynicism are, however, very important for politicians and campaign professionals to take into account. Politics should be engaging for citizens and should welcome them to participate in society. However, as made clear by these results and by the results of multiple other studies, negative campaigning does not contribute in a positive way to the feelings of citizens towards politics. On the contrary; personal and uncivil negative campaigning results in more political cynicism among citizens. Politicians should be aware of the long-lasting effects that their campaigns can have on citizens. They have a role to educate and engage citizens and should keep this in mind when deciding to go negative. Politicians should not just focus on gaining votes in spite of everything else.
References
Agger, R. E., Goldstein, M. N., & Pearl, S. A. (1961). Political cynicism: Measurement and meaning. The Journal of Politics, 23, 477–506.
Akkerman, A., Mudde, C., & Zaslove, A. (2014). How Populist Are the People? Measuring Populist Attitudes in Voters. Comparative Political Studies, 47(9), 1324-1353.
Auter, Z., & Fine, J. (2016). Negative Campaigning in the Social Media Age: Attack Advertising on Facebook. Political Behavior,38(4), 999-1020.
Baines, P. R., Scheucher, C., & Plasser, F. (2001). The “Americanisation” myth in European political markets-A focus on the United Kingdom. European Journal of Marketing, 35(9/10), 1099-1117.
Baumgartner, J. (2013). Internet Political Ads in 2012: Can Humor Mitigate Unintended Effects of Negative Campaigning? Social Science Computer Review, 31(5), 601-613.
Bos, L., Van Der Brug, W., & De Vreese, C. H. (2013). An experimental test of the impact of style and rhetoric on the perception of right-wing populist and mainstream party leaders. Acta Politica, 48(2), 192-208.
Boukes, M., & Boomgaarden, H. (2015). Soft News with Hard Consequences? Introducing a Nuanced Measure of Soft Versus Hard News Exposure and Its Relationship with Political Cynicism. Communication Research, 42(5), 701-731.
Brandstätter, H., & Opp, K. D. (2014). Personality traits (“Big Five”) and the propensity to political protest: Alternative models. Political Psychology, 35(4), 515-537.
Brooks, D., & Geer, J. (2007). Beyond Negativity: The Effects of Incivility on the Electorate. American Journal of Political Science,51(1), 1-16.
Dardis, F., Shen, F., & Edwards, H. (2008). Effects of Negative Political Advertising on Individuals' Cynicism and Self-Efficacy: The Impact of Ad Type and Message Exposures. Mass Communication and Society, 11(1), 24-42.
Dawes, C., Cesarini, D., Fowler, J. H., Johannesson, M., Magnusson, P. K., & Oskarsson, S. (2014). The relationship between genes, psychological traits, and political participation. American Journal of Political Science, 58(4), 888-903.
Elmelund-Præstekær, C. (2010). Beyond American negativity: Toward a general understanding of the determinants of negative campaigning. European Political Science Review, 2(1), 137-156.
Fridkin, K., & Kenney, P. (2008). The Dimensions of Negative Messages. American Politics Research, 36(5), 694-723.
Fridkin, K., & Kenney, P. (2011). Variability in Citizens’ Reactions to Different Types of Negative Campaigns. American Journal of Political Science, 55(2), 307-325.
Geer, J. G. (2006). In defense of negativity: Attack ads in presidential campaigns. University of Chicago Press.
Gross, J., & Johnson, K. (2016). Twitter Taunts and Tirades: Negative Campaigning in the Age of Trump. 49(4), 748-754.
Hameleers, M., & Schmuck, D. (2017). It’s us against them: A comparative experiment on the effects of populist messages communicated via social media. Information, Communication & Society, 20(9), 1425-1444.
Haselmayer, M., & Jenny, M. (2018). Friendly fire? Negative campaigning among coalition partners. Research & Politics, 5(3), Research & Politics, September 2018, Vol.5(3). Hill, R. P., Capella, M., & Cho, Y. N. (2015). Incivility in political advertisements: A look at
the 2012 US presidential election. International Journal of Advertising, 34(5), 812-829. Inglehart, R. F., & Norris, P. (2016). Trump, Brexit, and the rise of populism: Economic
Jagers, J., & Walgrave, S. (2007). Populism as political communication style: An empirical study of political parties' discourse in Belgium. European journal of political research, 46(3), 319-345.
Jamieson, K. H. (1992). Dirty politics: Deception, distraction, and democracy. Oxford University Press on Demand.
Jonker, J., & de Winther, W. (2019, 12-01). VVD: ‘nee’ tegen klimaatakkoord. Retrieved from:
https://www.telegraaf.nl/nieuws/3014754/vvd-nee-tegen-klimaatakkoord
Kaid, L. L., & Johnston, A. (1991). Negative versus Positive Television Advertising in US Presidential Campaigns, 1960-1988. Journal of communication, 41(3), 53-64.
Kenski, K., Filer, C., Conway-Silva, B., & Payne, J. (2018). Lying, Liars, and Lies: Incivility in 2016 Presidential Candidate and Campaign Tweets During the Invisible Primary. American Behavioral Scientist, 62(3), 286-299.
Kouwenhoven, A., & Logtenberg, H. (2017). Hoe Denk met ‘trollen’ tegenstanders monddood probeert te maken. Retrieved from:
https://www.nrc.nl/nieuws/2017/02/10/de-trollen-van-denk-6641045-a1545547
Lau, R., Sigelman, L., & Rovner, I. (2007). The Effects of Negative Political Campaigns: A Meta-Analytic Reassessment. The Journal of Politics, 69(4), 1176-1209.
Lee, & Xu. (2018). The more attacks, the more retweets: Trump’s and Clinton’s agenda setting on Twitter. Public Relations Review, 44(2), 201-213.
Mutz, D., & Reeves, B. (2005). The New Videomalaise: Effects of Televised Incivility on Political Trust. American Political Science Review, 99(1), 1-15.
Nai, A. (2018a). Fear and Loathing in Populist Campaigns? Comparing the Communication Style of Populists and Non-populists in Elections Worldwide. Journal of Political Marketing, 1–32.
Nai, A. (2018b). Going Negative, Worldwide: Towards a General Understanding of Determinants and Targets of Negative Campaigning. Government and Opposition, 1-26. Pattie, C., Denver, D., Johns, R., & Mitchell, J. (2011). Raising the tone? The impact of ‘positive’ and ‘negative’ campaigning on voting in the 2007 Scottish Parliament election. Electoral Studies, 30(2), 333-343.
Pattyn, S., Van Hiel, A., Dhont, K., & Onraet, E. (2012). Stripping the Political Cynic: A Psychological Exploration of the Concept of Political Cynicism. European Journal of Personality, 26(6), 566-579.
Pfau, M., & Burgoon, M. (1989). The efficacy of issue and character attack message strategies in political campaign communication. Communication Reports, 2(2), 53-61.
Reysen, S. (2005). Construction of a new scale: the Reysen likability scale. Social Behavior and Personality: An International Journal, 33(2), 201-208.
Roddy, B., & Garramone, G. (1988). Appeals and strategies of negative political advertising. Journal of Broadcasting & Electronic Media, 32(4), 415-427.
Roese, N. J., & Sande, G. N. (1993). Backlash Effects in Attack Politics 1. Journal of Applied Social Psychology, 23(8), 632-653.
Rooduijn, M., van der Brug, W., de Lange, S. L., & Parlevliet, J. (2017). Persuasive Populism? Estimating the Effect of Populist Messages on Political Cynicism. Politics and Governance, 5(4), 136-145.
Schenck‐Hamlin, W. J., Procter, D. E., & Rumsey, D. J. (2000). The influence of negative advertising frames on political cynicism and politician accountability. Human Communication Research, 26(1), 53-74.
Sniderman, P. M., Hagendoorn, L., & Prior, M. (2004). Predisposing factors and situational triggers: Exclusionary reactions to immigrant minorities. American political science review, 98(1), 35-49.
Toros, E. (2017). How to run the show? The differential effects of negative campaigning. Turkish Studies, 18(2), 297-312.
Vliegenthart, R. & Walter, A. (2010, 28-04). Vrij spel om negatief te gaan. Retrieved from:
https://www.groene.nl/artikel/vrij-spel-om-negatief-te-gaan
Walter, A. (2014a). Choosing the enemy: Attack behaviour in a multiparty system. Party Politics, 20(3), 311-323.
Walter, A. (2014b). Negative Campaigning in Western Europe: Similar or Different? Political Studies, 62(1_suppl), 42-60.
Walter, A. S., & Van der Brug, W. (2013). When the gloves come off: Inter-party variation in negative campaigning in Dutch elections, 1981–2010. Acta Politica, 48(4), 367-388. Walter, A. S., Brug, van der, W., & van Praag, P. (2014). When the stakes are high: Party
competition and negative campaigning. Comparative Political Studies, 47(4), 550-573. Wirz, D. S. (2018). Persuasion through emotion? An experimental test of the emotion-eliciting
Appendix Tweet 1: issue/civil
Tweet 2: issue/uncivil
Table 1. Manipulation check tweets ‘aim’
Table 3. Items ‘Populist attitudes’
Table 4. Items ‘Likeability of sender/target’
