Universiteit van Amsterdam
Mud-slinging in the Dutch polder
Examining the effects of (party- vs. candidate-targeted) negative
campaigning and exploring the moderating role of personality
Aldo van Grunsven, BSc. Student number: 6138705
Utrecht, October 2015 Supervisor: Yph Leltach
Abstract
Even though negativity in political campaigning is on the rise in most Western democracies, very little is known about its effect outside the US. Using a 2x2 experimental design (N = 184), the present study examined the effects of negative political campaigning on voter turnout, political trust, political efficacy, affect towards sponsor and affect towards target of negative ads in the Dutch political context and tested whether there are differences in impact of party- vs candidate-targeted attacks. Furthermore, it is one of the first studies to incorporate the Big Five personality traits as moderating factors of the effects of negativity.
The results showed that exposure to negative campaigning did not significantly decrease voter turnout and political trust. However, external political efficacy was significantly lower among participants exposed to negative ads. Furthermore, negative campaigning significantly lowered affect towards its sponsor (backlash), while its influence on affect towards the target of negativity was only marginal. No differences between effect sizes of party- vs candidate targeted attacks were found. Personality trait Openness moderated the influence of negativity on affect towards the sponsor of negative ads: participants scoring low on Openness were significantly more sensitive to negative campaigning in terms of their evaluation of the sponsor of attack ads, indicating that negative campaigns backlash for these types of personalities in particular. Overall, this study presents the first tentative evidence that ‘going negative’ does not work as intended in the Netherlands.
Introduction
With the slow decay of traditional parties and party identification in the years after the Second World War and the simultaneous rise of television, the traditional manner in which politicians present themselves to the electorate has shifted. Political campaigning in Western democracies has become increasingly professionalized (Gibson & Römmele, 2001;
Strömback, 2007), more costly (Fowler & Ridout, 2012), longer to the point of permanent campaigning (Strömback, 2007), and increasingly mediatized (Mazzoleni & Schulz, 1999).
Alongside the growth of professionalism came the rise of the negative campaign. Sometimes called mud-slinging or attack ads, negative political campaigning has become an increasingly common technique to win votes while simultaneously damaging opponents, particularly in the US. Geer (2012), for example, showed that the share of negativity in U.S. presidential election campaigns has increased by more than 50 percent in the last four decades. To illustrate what this means, during the Obama-Romney campaign of 2012 more than four out of five ads were either directly negative or made negative comparisons between candidates (Fowler & Ridout, 2012).
Yet, negative campaigns are not exclusive to the American two-party system. European countries too, have seen a steady growth in negativity in the last few decades (Walter & van der Brug, 2013), indicating a need for studies outside the US context: “Since the phenomenon of negativity seems to be universal to politics and election campaigns, it is imperative to care about [the determinants of] negativity beyond the US case” (Elmelund-Præstekær, 2010, p. 138). Recently, there have been some positive developments in this area, with a number of studies on negative campaigning conducted in European proportional election, or multi-party, systems (Elmelund-Præstekær, 2008; Hansen & Pedersen, 2008; Walter, van der Brug, & van Praag, 2014).
Unfortunately, thus far almost all of these studies have focused solely on answering why and when politicians utilize negativity. To my knowledge, only one study by Pattie, Denver, Johns and Mitchell (2011) specifically examined the effects negativity has on the electorate in a European multi-party system. However, their study used survey data, limiting
the possibilities to draw causal conclusions. This leaves a considerable knowledge gap in our understanding of negative campaigning: given the differences between two-party and multi-party systems, particularly in how voters approach elections (i.e. in terms of strategic voting) and how parties present themselves in electoral campaigns (i.e. taking into account the need for coalition partners) (Hobolt & Karp, 2010), the effects of negative campaigning as found in a two-party system cannot be expected to transfer directly to multi-party democracies
(Elmelund-Præstekær, 2010; Hansen & Pedersen, 2008). Moreover, while negative campaigning in two-party systems often focuses on the opposing candidate (Pinkleton, 1997), politicians in multi-party systems seem more inclined to attack opposing parties (Walter & van der Brug, 2013). With the ongoing personalization of politics, often ascribed to the increasing adherence to media logic (Campus, 2010), the question arises whether the growing attention to candidates in multi-party systems might result in differences in effect between attacking a candidate or attacking a party.
In this master thesis, I present one of the first studies to experimentally test the effects of negative campaigning in a multi-party democracy, while simultaneously contributing to the ongoing debate on the effects of negative campaigning in general by investigating what the effects of party vs. candidate-targeted negative campaigning are on the electorate (in terms of voter turnout, political trust, political efficacy) and on politicians (in terms of affect towards political parties and candidates).
Furthermore, this study takes a pioneering effort to study the moderating role
personality has in this process. It has long been recognized in individual-level media effects theories that the influence of media messages can be enhanced or reduced by individual characteristics (Bryant & Miron, 2004; Petty & Cacioppo, 1986). One important characteristic is our personality. Personality and political behavior are often seen as two sides of the same coin and intrinsically linked: if our personalities reflect what we do and what we like, certainly it must also reflect how we behave politically and how we react to different types of political communication. However, most studies on the effect of negative campaigning have primarily focused on the direct influence of negativity while often neglecting the fact that its impact
may depend on people’s personality. This is rather problematic: with professional political marketers increasingly catering towards the individual voter (Baines, Harris, & Lewis, 2002; Strömback, 2007), it has only become more important to increase our understanding of how personal differences can affect our reaction to negative political campaigning.
Literature indeed shows that there is good reason to assume our personalities can in fact play a major role in determining the impact of political campaigning. Studies have shown that personality affects one’s ability to be persuaded by political campaigning (Gerber, Huber, Doherty, Dowling, & Panagopoulos, 2013; Gerber, Huber, Doherty, & Dowling, 2011b) and that personal differences moderate the effectiveness of negative campaigning (Fridkin & Kenney, 2013). Gerber et al. even state that “Because the Big Five [personality dimensions] are posited to shape stimuli response more generally, researchers may find that these traits can improve the explanatory power of models predicting other outcomes that are of interest to political researchers” (2011b, p. 284). Following Gerber et al.’s (2011b) recommendation, the study presented here contains an exploratory analysis of the moderating influence of personality on the effects of negative campaigning in addition to the analysis of the main effect of negativity.
In sum, in this thesis I present an experimental study outside the US context which aims to answer the following questions: (1) What are the effects of negative campaigning on
both the electorate (in terms of voter turnout, political trust and political efficacy) and the politicians using it (in terms of affect towards sponsor and affect towards target), (2) are these effects different for party- vs candidate-targeted negative campaigns and (3) does personality moderate the effects of negativity?
Answers to these questions will add to existing knowledge on negative campaigning by specifically focusing on its effects in a multi-party system, an area of research that has thus far been overlooked in scientific literature. Secondly, this study expands on existing literature by making a distinction between party-targeted and candidate-targeted attacks and studying the differences in effect for these specific types of negative political advertisements. Lastly, this research adds much-needed input to existing theory on the effects of negativity by
making the first attempt to explore to what extent personality moderates the effects of negative campaigning on voter turnout, political trust, political efficacy, and affect towards political parties and candidates.
Theoretical framework Defining negative campaigning
Despite the large amount of attention in political (communication) science literature, there is still debate on what negative campaigning actually entails. While some scholars tend towards normative descriptions, describing negativity in political campaigning as “hostile and ugly” and “rhetorical, rather than substantive” (Ansolabehere, Iyengar, Simon, & Valentino, 1994, p. 829), others remain more neutral, separating ‘uncivil’ and ‘civil’ negativity (i.e. Brooks & Geer, 2007; Mutz & Byron, 2005). Yet others differentiate between negative ads and
comparative ads, with the latter setting the merits of the sponsoring candidate off against the faults of the targeted candidate (Fridkin & Kenney, 2013; Lau, Sigelman, Heldman, & Babbitt, 1999; Pinkleton, 1997).
In general, scholars agree that negative political ads, or attack ads, deviate from traditional, candidate-promoting ads in one important aspect: instead of specifically
increasing affect or likelihood to vote for the sponsoring candidate, negative ads are instead intended to damage the opposition and focus solely on the faults of the opposing candidates (Harrington & Hess, 1996). Following the examples of both Elmelund-Præstekær (2008) and Hansen and Pedersen (2008), I adopt an even more specific definition by Lau and Pomper (2001b, pp. 805–806), who specify negativity as: “only talking about the opponent – the deficient nature of his or her programs, accomplishments, qualification, associates and so on. Positive campaigning is just the opposite: talking about one’s own accomplishments, qualifications, programs, etcetera”. This definition offers a number of advantages. Firstly, it contains no inherently normative evaluation of negative campaigning. Secondly, the dichotomous nature of this definition is conducive in performing experimental research,
isolating negativity to a single variable (talking solely about opponent) and thus allowing for better manipulation of negativity in stimulus material (Lau & Rovner, 2009).
The effects of negative campaigning: The Demobilization hypothesis
While the popularity of negative campaigning is undeniable (Elmelund-Præstekær, 2008; Fowler & Ridout, 2012; Geer, 2012), there remains considerable debate in the scientific community on its effects. Even though most political campaigners keep insisting that negativity ‘just works’, in the last few decades, a number of scholars have pointed out that negative campaigning poses a potential danger to the workings of democracy.
In a widely cited experimental study, Ansolabehere et al. (1994) found evidence that negative ads (compared to positive ads) lower voting intention, decrease political efficacy and increase political cynicism. In a follow-up study using survey data, the authors found identical results, concluding that “negative advertising demobilizes voters” (Ansolabehere, Iyengar, & Simon, 1999, p. 901). According to the authors, there are three main reasons why negative campaigns might demobilize voters. First, negative campaigns could be successful in lowering voter turnout, by decreasing the probability of voting for the target of the attacks without simultaneously increasing the probability of voting for the sponsor of the attacks. Second, negative campaigning can influence the turnout of elections indirectly by having negative effects on political trust and political efficacy. Both political trust, defined as one’s faith in the efficient and honest functioning of government, and political efficacy, defined as the feeling that one’s actions can in fact influence the political process (Craig, Niemi, & Silver, 1990), are crucial ingredients for a healthy democratic system: political trust and efficacy play a role in determining to what extent the electorate is willing to cast its ballots and are thus vital for voter turnout. A decline in political trust, political efficacy and voter turnout can thus be detrimental to the functioning of democracy (Ansolabehere et al., 1994). Third, negative campaigns could lower evaluations of not only the target of attack ads, but also of its sponsor, a phenomenon known as ‘backlash’. This would decrease the probability of voting for either of them (Ansolabehere et al., 1999; Lau & Rovner, 2009).
Ansolabehere et al.’s demobilization hypothesis has been supported as well as disputed in a number of follow-up studies. Pinkleton, Um and Austin (2002) found tentative support for the negative influence of negativity on political efficacy, although “findings only bordered on statistical significance” (p. 23). Mutz and Byron (2005) showed that negativity significantly lowers political trust, yet stipulate that it can simultaneously promote interest in politics, as people seem to ‘like watching a good fight’. Others have found no evidence at all of negativity being detrimental to voter turnout (Garramone, Atkin, Pinkleton, & Cole, 1990; Lau et al., 1999) or political trust (Pinkleton et al., 2002), with some even reporting potentially positive effects of negativity on voter turnout (Brooks & Geer, 2007; Freedman & Goldstein, 1999).
In an ambitious effort to examine the legitimacy of the claim that negativity threatens democracy and to bring some closure to the debate on the demobilizing effects of negative campaigning, Lau et al. (1999; 2007) have conducted two expansive meta-analyses on experimental and survey-based studies on the effects of negativity. The authors find partial support for the claims made by Ansolabehere et al., with their findings likewise showing that both political efficacy and political trust suffer from negative campaigning. However, this evidence is still rather shaky: in the results of their meta-analysis, Lau et al.’s (2007) show that 25 studies found demobilizing effects of negativity, with only nine statistically significant results, while 29 studies (eight of which were significant) found effects in the opposite direction, showing that negativity does not affect voter turnout or might even improve it.
Given these contrasting findings, it is difficult to come to a definitive conclusion on the effects of negative campaigning on the functioning of the democratic system. Furthermore, it is important to understand that (almost) all studies on the effects of negativity have been conducted in two-party systems (Elmelund-Præstekær, 2010). Unfortunately, only very few studies have been conducted in countries with a multi-party system. Experimental studies in particular are almost none-existent. The present study attempts to address this knowledge gap by putting the demobilization hypothesis as put forth by Ansolabehere et al. to the test in a multi-party system. I assert that compared to traditional, positive campaigning:
H1 Negative campaigning demobilizes voters, decreasing voter turnout H2 Negative campaigning decreases political trust
H3 Negative campaigning decreases political efficacy
Besides finding some tentative evidence for the demobilizing influence of negativity on voter turnout, political trust and political efficacy, Lau et al. (1999, 2007) also find evidence for the ‘backlash effect’. The findings in their meta-analyses show that the decrease in affect for the sponsor of negativity (unwanted) can in some cases be even higher than the
decrease in affect for the targeted candidate (intended), due to the sponsoring candidate being perceived as unsympathetic and ‘punching below the belt’. The authors do warn to interpret these findings cautiously, explaining that the results from survey-based studies do not directly gauge the effect of going negative on affect for attackers and their targets (Lau et al., 2007).
This stipulation made by the authors indicates a general problem in negativity research. In real life, politicians are generally both the victim and the culprit of negative campaigning, making affect measures as performed in survey studies contaminated: when finding a significant drop in affect for a candidate, is this due to backlash from negative campaigning or the result of negative campaigning by the opposing candidate? Nevertheless, other survey studies have also found support for the backlash effect of going negative (Jasperson & Fan, 2002; Lau & Rovner, 2009; Skaperdas & Grofman, 1995), showing that affect towards the sponsors of ads significantly dropped when they employ negative campaigning. In European context, Pattie et al. (2011) report signs of backlash for politicians utilizing negativity as well. I therefore expect that:
H4 Negative campaigning decreases affect for the sponsor of negativity (backlash effect)
The present study, due to its experimental design, allows measurement of affect while maintaining clear distinction between target and sponsor of negative ads, thus increasing the validity of any results showing evidence of negativity backlashing on its sponsor.
So why do political campaigners insist that negative campaigning works? There are a number of explanations why negativity has the potential to work in the favor of a candidate. First of all, Fridkin and Kenney (2013) point out that the “negativity bias”, the tendency of individuals to give more attention to negative than positive messages, makes negative ads more apparent and memorable for the electorate, thus increasing the chances the message of the sponsor of a negative ad will be remembered. Furthermore, by appealing to feelings of fear and anxiety, negative campaigning can encourage reconsideration of prior beliefs, i.e. candidate choice, and stimulates information seeking, thus potentially improving political engagement (Brader, 2005). This process might explain why some authors, in contrast to the demobilization theory of Ansolabehere et al. (1999) have even found that negative
campaigning can actually increase voter turnout (i.e. Brooks & Geer, 2007; Lau & Pomper, 2001b).
In addition, studies have frequently and consistently shown that negative campaigning is successful in damaging people’s affect towards the targeted candidate (Fridkin & Kenney, 2004; Garramone et al., 1990; Lau et al., 1999; Pinkleton, 1997). Furthermore, people weigh negative communication more heavily than positive communication when forming attitudes. This includes assessment of likability of a political party or candidate (Pinkleton et al., 2002), thus further increasing the likelihood of negative campaigning damaging the targeted
candidate. Therefore, I expect that:
H5 Negative campaigning decreases affect for the target of negativity
Attacking the opposition: targeting party or candidate?
Although it is understandable that almost all US literature on negative campaigning
differences between the US and Europe in terms of the role that political parties play. First of all, most European countries feature multi-party systems instead of two-party systems. Second, parties in multi-party systems very often need to from coalitions. Voters have been shown to be distinctly aware that this means a vote on some parties might also mean a vote towards a specific coalition (Elmelund-Præstekær, 2008; Hobolt & Karp, 2010), changing the role that parties have in the political playing field. Third, although some do find signs of increase in the last few years (i.e. Campus, 2010), personalization of politics in Europe is still far from being at the same level as in the US. This further increases the importance of the role political parties play in multi-party systems.
These differences between two-party and multi-party systems raise the question whether the ‘traditional’, candidate-targeted negative ad has the same effect in systems that are much more party-focused. Moreover, could it be that party-targeted attack ads are much more influential in multi-party systems? This study aims to be one of the first to give tentative answers to these questions.
A thorough analysis of the literature showed that theory concerning party- vs candidate-based attacking is very limited. Again, this is likely due to the overwhelming proportion of US-based studies that only investigate the effect of candidate-targeted negativity. However, a number of these studies have shown that issue-based negativity, criticizing a candidate’s policies rather than his person, is more effective in decreasing affect towards the targeted candidate (Shen & Wu, 2002) and can sometimes even increase the intention to vote (Lau et al., 2007). These findings might indicate that attacking a party, an institution that is by its definition issue-based, is more effective in damaging the opposition and increasing voter turnout.
In addition, party-targeted negative campaigning can be seen as justified criticism of a party’s policies and standpoints (Walter & van der Brug, 2013), therefore increasing its negative influence on affect towards the targeted party. In contrast, candidate-targeted ads are often perceived as uncivil or bad-mannered (Brooks & Geer, 2007), thus potentially decreasing the negative effect on affect towards the targeted candidate as voters do not find
them credible or persuasive. Furthermore, this perception of candidate-targeted ads as uncivil and rude might potentially have a detrimental effect on political trust, efficacy and the will to vote, as people are disgusted by the manner in which politicians act towards one another.
Unfortunately, the considerations mentioned above remain highly speculative: issue-based attacking is significantly different from party-targeted negativity, as it still remains candidate-targeted; targeting a party might be seen as just as uncivil as attacking a candidate. As of yet, there is simply too little known about this subject. This severely limits the possibilities of formulating strong hypotheses. Therefore, I aim to study the differences in party- vs candidate-targeted negativity using three open research questions:
RQ1 Are there differences in the effects of party-targeted vs candidate-targeted negative campaigning on voter turnout, political trust and political efficacy? RQ2 Are there differences in the effects of party-targeted vs candidate-targeted negative campaigning on attitude towards the sponsor of negative ads?
RQ3 Are there differences in the effects of party-targeted vs candidate-targeted attack ads on attitude towards the target of negative ads?
Personality and politics: the role of the Big Five
In recent years, the study of personality has been dominated by the Five-Factor model, often colloquially called ‘the Big Five’, a framework for indexing personality on five main dimensions (McCrae & Costa, 1997): Extraversion, Agreeableness, Conscientiousness, Neuroticism and Openness to Experience. Besides offering an orderly and easy way to classify personalities, the Big Five have been shown to be stable over time (Borkenau & Ostendorf, 1998) and can be reliably measured across cultures (John & Srivastava, 1999).
Scholars have found evidence that the Big Five influence how we behave politically. Factors such as political ideology (Carney, Jost, Gosling, & Potter, 2008; Hirsh, DeYoung, Xiaowen Xu, & Peterson, 2010; Mondak & Halperin, 2008) and partisanship (Gerber, Huber,
Doherty, & Dowling, 2011a) are heavily influenced by our personalities. Furthermore, studies show that higher scores on Extraversion and Openness to Experience increase political efficacy and voter turnout (Cooper, Golden, & Socha, 2013; Gerber, Huber, Doherty, Dowling, et al., 2011; Vecchione & Caprara, 2009) and that Agreeableness correlates positively with political trust (Gerber, Huber, Doherty, & Dowling, 2011a).
However, despite the abundancy of literature on personality as predictor of political attitudes and behavior, remarkably little is known about how personality moderates the way people respond to different types of political communication. To illustrate, only very recently a study by Gerber et al. (2013) investigated for the first time whether Big Five traits moderate how people respond to different appeals to go voting. The authors reported that individuals scoring high on Openness to Experience are more likely to be persuaded by all types of appeal messages. This is not unexpected, as Open individuals tend to be more accepting of new information and more likely to process information that contrasts their own standpoints (Mondak & Halperin, 2008; Rammstedt & John, 2007).
Furthermore, although they did not measure personality differences by using the Big Five, Fridkin and Kenney (2013) studied whether the impact of negative messages on
candidate evaluation differs between individuals. The authors find that negative campaigning lowers affect towards sponsors of attack ads for “individuals with a low tolerance for incivility” (p. 322). When speaking in terms of the Big Five, this finding can be translated to “individuals with high scores on Agreeableness”: highly Agreeable personalities are conflict-avoiding and tend to dislike social comparison (McCrae & Costa, 1997; Mondak & Halperin, 2008).
In addition, in a study on how emotional cues influence the persuasiveness of political communication, Brader (2005) found that negative, fear-inducing messages can increase the persuasiveness of a political message. Personalities scoring high on Neuroticism tend to be more fearful and prone to anxiety (John & Srivastava, 1999; Mondak & Halperin, 2008), thus potentially increasing the persuasive force of negative campaigning for these individuals, as negative campaigning often cues fear (Elmelund-Præstekær, 2010).
Based on these findings, and given the fact that negative campaigning, which appeals against voting for a certain candidate, is very often seen as uncivil and uses fear to increase recall of its message (Hansen & Pedersen, 2008), I come to three tentative hypotheses:
H6 Highly Open individuals will be more sensitive to the effect of negativity on affect towards the target of negative campaigning
H7 Highly Agreeable individuals will be more sensitive to the effect of negativity on affect towards the sponsor of negative campaigning
H8 Highly Neurotic individuals will be more sensitive to the effect of negativity on affect towards the target of negative campaigning
Due to the scarcity of available literature on personality as moderator of the effects of negative campaigning, no hypotheses could be formulated on the moderating role of Conscientiousness and Extraversion. Furthermore, despite substantial effort, no literature was found on how the Big Five moderate the effects of negativity on political trust, political efficacy and voter turnout. This study will take a first step towards closing this knowledge gap by including all Big Five dimensions and examining their moderating effects on all outcome variables.
Method Participants and design
The experiment was constructed using the online digital survey-tool Qualtrics and distributed via hyperlink. Participants were recruited into the experiment within a three-week period through social networks and direct contact. Additionally, a snowball method was employed, whereby participants who had finished the survey were requested to spread the link to their own friends and family. In total, 207 participants were recruited into the sample with 186 participants completely finishing the questionnaire. To check whether the sponsor of the flyers had been recognized, participants were asked one multiple choice question about the
sender of the message. Only participants who had answered this question correctly (“PvdA”) were included in further analysis. Consequently, two participants were removed from the final sample (N = 184).
Participants’ age ranged from 18 to 93 years old (M = 33.4, SD = 14.0), with a slight female majority (55%). Most participants were highly educated, with 72 percent of the sample indicating at least one year of formal college education. In general, participants were left-wing (M = 4.490, SD = 1.39) and progressive (M = 4.96, SD = 1.23), and voted accordingly: over 66.4 percent of participants had voted on left-wing/progressive parties in the previous Dutch national election.
Using a 2x2 between subjects factorial design, participants were assigned at random to one of four conditions. Participants were exposed to political ads that were either positive or negative in tone (campaign type: positive vs negative) and targeted towards either a political party in general or towards the frontman of a political party (target: party- vs candidate-targeted).
Procedure
When participants clicked the hyperlink, they read a short consent form indicating the length of the survey and informing them about the subject of the study. Participants were pointed to their right to quit the survey at any time and asked to sign the consent form. Next, participants were asked to fill in their sex, age in years, level of education, the party they had voted for in the last Dutch national election, and their general political preference.
Participants were then instructed that they would have to carefully study two political flyers that were publicized during the 2012 Dutch national election campaign. Depending on their experimental condition, participants were exposed to either negative or positive flyers that were either party or candidate-targeted. To ensure thorough exposure to the stimulus material, participants were blocked from continuing the questionnaire for ten seconds per flyer. After exposure to the stimulus material, a manipulation check was conducted to ensure that participants had understood both the sponsor, the content and the tone of the flyers.
Next, the dependent variables were measured, followed by the assessment of the Big Five personality traits. The experiment concluded with thanking participants for their
contribution to the study and explaining that all the political flyers they had seen in the study were constructed and manipulated by the researchers and were never published. Finally, participants were asked for their e-mail address in case they were interested in learning of the results of the study.
Stimuli
To ensure realism and believability, the stimulus material in this study consisted of two political flyers per condition. There is a number of reasons for this choice: political flyers are a common sight during Dutch election campaigns, they are easily recognized by the public as ‘political communication’ and are frequently prime sources of information for the Dutch electorate (Brants & Van Praag, 2006). Furthermore, political flyers in the Netherlands often have a distinct ‘look’ that is associated with a certain party, making the sponsor of the message easily recognizable to participants. Two flyers were deemed necessary to deepen the impact and to increase the chances participants would recognize and respond to the tone and content of the message in the ads.
As an important aim of this study was to examine the effects of negativity on both the sponsor as well as the target of a negative campaigning, the sponsor and target needed to be easily identifiable parties that every participant would be able to recognize. Therefore, the two largest parties in the Netherlands were chosen as sponsor and target respectively. Traditionally and at the time this study took place, these parties are the PvdA (Partij van de Arbeid) and the VVD (Volkspartij voor Vrijheid en Democratie). The Dutch left-wing labour party PvdA was assigned the role of sponsor for a number of reasons. First of all, the PvdA was in the opposition during the election campaign of 2012 and would therefore be more likely to attack the incumbent party, which at that time was the Dutch right-wing liberal party VVD (Walter et al., 2014). Secondly, the PvdA was trailing just behind the VVD in the polls, making an attack on the frontrunner more realistic (Skaperdas & Grofman, 1995). Most importantly, the PvdA was the most negative party during the campaign of 2006 (Van Praag
& Walter, 2013), making this party more believable as a source of negative campaigning and increasing the likelihood participants would believe the political flyers to be real.
The political flyers were specifically created for this study using Adobe Photoshop (see Appendix A). To increase believability, meticulous care was taken to duplicate the ‘PvdA-style’, which include large visual stimuli, a specific font and a large red vertical stripe containing the slogan of the party. In all conditions, participants were shown two political flyers, with the first containing a short political message and the second only a short slogan.
To prevent differences between conditions occurring solely due to differences in persuasive power between the negative and positive political ads, a pre-test was conducted. This pre-test was held to make certain that the wording of the message and the slogan would be equally strong in terms of persuasiveness in all conditions, so that only the differences in tone of the message would affect participants, not the differences in wording. To ensure that this equality was achieved, the persuasiveness of the wording in the political ads was tested in a small focus group (N = 5). In the opinion of this focus group, the persuasive strength of the wording in the negative and positive conditions was about equal. However, some focus group members felt that a number of words used in the negative message were “too rude” or “indecent” and would therefore decrease the persuasive power of the message when
compared to the positive condition. Based on these statements and in cooperation with the focus group, slightly less uncivil messages were constructed and used in the final
experiment. Furthermore, care was taken so that the negative flyers attacking the candidate only focused on political issues, not on his personal live.
Positive stimuli. Depending on whether the group-condition was party- or candidate-targeted, the message flyer in the positive conditions consisted of a large visual of either the PvdA logo or PvdA-leader Diederik Samsom. A red banner containing the PvdA-slogan used during the 2012 election campaign, “The Netherlands, stronger and more social”, was
featured prominently. Furthermore, a short positive message (60 words) promoting the political goals of the PvdA/Diederik Samsom was constructed, containing lines such as “The PvdA/Diederik Samsom stands for a country in which all can find a decent paying job”. The
slogan flyer again featured a visual of either the PvdA logo or Diederik Samsom and the red banner containing the PvdA-slogan, but contained only one short additional slogan, “Time to give the economy a large boost”. Importantly, in all the flyers used in the positive condition, readers were told to “Vote PvdA”. This was added to ensure participants would understand who the sponsor of the message was.
Negative stimuli. The message flyers in the negative conditions were very similar to the ones used in the positive conditions in terms of general layout. However, visual stimuli instead consisted of a red ‘prohibited’-sign across a picture of either the VVD logo or VVD-party leader Mark Rutte, depending on the condition. Furthermore, the slogan featured in the red banner was changed to be negative: “Say NO to the VVD!/Rutte!”. The wording of the message was changed to contain a negative message, to lines such as “The cutbacks plans by the VVD/Mark Rutte will destroy the economy” and “You won’t get the truth from the VVD!/Mark Rutte!”. In the slogan flyer too the wording was changed to contain a negative message, “Telling coldblooded lies”, accompanied by a visual of either the VVD logo or Mark Rutte. As in the positive conditions, all flyers used in the negative conditions featured a prominent call to “Vote PvdA” to ensure participants recognized the sponsor of the message. Dependent Measures
The following dependent variables were measured in this study: voter turnout, political trust, political efficacy (internal and external), affect towards sponsor (party and candidate) and affect towards target (party and candidate) of political ads. The scales used in this study are included in Appendix B.
Voter turnout. A questionnaire taken from the American National Election Survey consisting of 6 items (ANES, 2013) was used to measure the likelihood of participants casting their vote in future elections. Items were translated to Dutch and scored on a five point Likert-scale ranging from 1 = completely disagree to 5 = completely agree. The questionnaire consisted of items such as “I think it is important to vote” and “I will definitely cast my vote in the next national election”. Principal component analysis confirmed the
one-dimensional structure of the measure (with one component explaining 66.64% of variance) and a reliable scale (α = .89) was constructed.
Political Trust. This measure was based on a questionnaire constructed for the ANES pilot (Craig et al., 1990) and consisted of seven items that were translated to Dutch. The questionnaire was scored on a five point Likert-scale ranging from 1 = completely disagree to 5 = completely agree and included items such as “Most public officials can be trusted to do what’s right” and “When government leaders make statements to the Dutch people on the television or in newspapers, they usually tell the truth”. Principal component analysis in the present yielded one component which explained 53.84% of variance. Internal consistency was good (.86).
Political efficacy. To measure this variable, seven items were again taken from a questionnaire created for the ANES (Niemi, Craig, & Mattei, 1991), translated to Dutch and scored on a five point Likert-scale ranging from 1 = “completely disagree” to 5 = “completely
agree”. This instrument differentiates between two sub-dimensions of efficacy: internal
efficacy, referring to beliefs about one's own competence to understand, and to participate effectively in, politics (4 items, i.e. “I consider myself to be well qualified to participate in politics”) and external efficacy, referring to beliefs about the responsiveness of governmental authorities and institutions to citizen demand (3 items, i.e. “I don't think public officials care much what people like me think”).
Principal component analysis (with Varimax rotation) in the present sample indeed showed the items to load on two components, explaining 64.32% of variance. However, one item (“Sometimes, politics and government seem so complicated that a person like me can’t really understand what’s going on”, reverse scored) loaded equally weak on both
components (rotated factor loading below .45) and was therefore removed from further analysis. This problem was not unexpected, as Niemi et al. (1991) found the same problem with this item in their study. Internal (4 items, α =.74) and external efficacy (2 items, α = .77) were constructed as separate variables.
Affect towards sponsor and affect towards target of political ads. After exposure to the stimulus, participants were asked to indicate their feelings towards the sponsor and target of the flyers, with both being scored through two items. To score affect towards the sponsor of the political flyers, participants were requested to indicate how they felt about (1) PvdA and (2) PvdA-leader Diederik Samsom on temperature scales ranging from 0 = very
negative to 100 = very positive. To score affect towards target, participants were requested
to indicate how they felt about (1) VVD and (2) Mark Rutte on temperature scales ranging 0 =
very negative to 100 = very positive. The measures of affect towards target in the positive
conditions served as reference point. Moderator
Personality. Personality was assessed with the Big Five traits using the BFI-10, a very short questionnaire consisting of 10 items. The BFI-10 was developed by Rammstedt and Oliver (2007) specifically for use in instances where time is limited. The BFI-10 was tested and compared to longer, more elaborate measures for the Big Five traits and was found to yield equivalent and valid results.
Participants were asked to indicate to what extent they agreed (1 = completely
disagree, 5 = completely agree) with ten statements starting with “I see myself as someone
who…” and ending with sentences such as “…is reserved” or “…tends to be lazy”. Each of the Big Five traits was measured using 2 items, which were combined to assign participants a score on that particular trait. Principal component analysis indeed yielded 5 components that explained 70.50% of variance. Items measuring Neuroticism, Conscientiousness and Openness to Experience loaded as expected on their respective components. This was also the case with the items measuring Extraversion and Agreeableness, however, one item loaded on both components. Similar results have been found by Rammstedt and John (2007), therefore this was deemed as an acceptable outcome. Scales for each Big Five dimension were constructed by computing the means of their respective two items.
Control variables
Previous studies (Gerber, Huber, Doherty, Dowling, et al., 2011; Lau et al., 2007; Mutz & Byron, 2005) have shown that voter turnout, political trust, political efficacy and attitude towards political actors are associated with sex, age (in years), education and political ideology. These four variables were therefore included in all analyses as controls in order to isolate the independent contribution of negative campaigning, personality and their
interaction terms on the dependent variables.
Education was measured by asking participants to indicate the highest level of education they had followed for at least a year, ranging from 1 = primary school to 6 =
university. Participants’ ideology was measured using four items that tapped two
sub-dimensions, left-wing ideology and progressive ideology. To measure left-wing ideology, participants were asked to indicate on seven-point scale to what extent they identified themselves as “right-wing vs left-wing” and “liberal vs socialistic”. Progressive ideology was measured by asking participants to indicate on seven-point scale to what extent they identified themselves as “traditional vs nonconventional” and “conservative vs progressive”. Principal component analysis indeed yielded the two components, explaining 87.26% of variance.
Results Preliminary analyses
To check whether manipulation was successful, two items scored on a five point Likert-scale (1 = completely disagree to 5 = completely agree) were combined to measure participants’ perception of negativity in the flyers. These items were “The tone of these flyers was generally positive” (reverse scored), and “The message in these flyers was presented in a negative way”. A one-way analysis of variance between participants exposed to positive (M= 2.21, SD = .73) versus negative flyers (M = 4.47, SD = .65) showed that the negative flyers were perceived as significantly more negative, F(1, 182) = 485.90, p < .001, confirming that manipulation of the stimulus material was successful. Additional one-way ANOVA showed no
significant differences between the perceived level of negativity of party-targeted vs candidate-targeted negative flyers (F = 0.06, p = n.s.).
ANOVA and Chi-square test confirmed that randomization had been successful: no significant differences between the four groups in terms of sex (χ²(3, N =184) = 3.71, p = .29), age (F(3, 180) = 1,50, p = .22), education (F(3,180) = 1.16, p = .33), left-wing ideology (F(3,180) = .72, p = .54) and progressive ideology (F(3,180) = .29, p = .84) were found. Effects of (party vs candidate-targeted) negativity on dependent variables
To determine the effect of negative campaigning on voter turnout (H1), political trust (H2), political efficacy (H3) and affect towards target (H4) and sponsor (H5) of negative ads, eight univariate analyses of variance separate for each dependent variables were conducted. In all eight ANOVAs, a 2 (campaign type: positive vs negative) x 2 (target: party vs
candidate) between-subjects design was adopted. Sex, age, education, left-wing and
progressive ideology were added as covariates. The results of the analyses are presented in Table 1.
First of all, the results showed no significant differences in voter turnout and political trust between the positive and negative conditions, indicating that compared to positive campaigning, negative campaigning does not demotivate the electorate to go to the polling station nor does it decrease trust in politics in general. Therefore, H1 and H2 are not supported. However, effects of negativity were found on political efficacy. More specifically, the results showed that negative campaigning significantly (p<.01) lowers external political efficacy. Internal political efficacy was not affected by negative campaigning. Given these contrasting findings, support for H3 is only partial.
Importantly, the findings confirm that negative campaigning can backlash, as affect towards the sponsor of negative ads was significantly lower in the negative conditions compared to the positive conditions: both affect towards the sponsoring party (PvdA) as well as affect towards the sponsoring candidate (Samsom) decreased significantly (p<.05) when participants were exposed to negative campaign. Therefore, H4 is fully supported.
Remarkably, the results show that negativity does not seem to work as intended: exposure to negative campaigning only marginally decreases participants’ affect towards the target of negative ads. Affect towards both party (VVD) as well as candidate (Mark Rutte) was not significantly (p<.10) lower for participants exposed to negative flyers. Therefore, H5 was not supported.
[Insert Table 1 here]
In addition to studying the main effects of negative campaigning, the second aim of this study was to explore whether there are differences in targeting political parties vs political candidates in political attack ads. The results showed no significant main effect of party-targeted vs candidate-party-targeted campaigning in terms of voter turnout, political trust and political efficacy nor were any significant effects found for affect towards the sponsor and target of negative ads, indicating that it does not seem to matter whether party or candidate is the focus of a negative attack. However, the interaction effect on affect towards the
targeted candidate was significant (p<.05). Additional analyses of variance showed that there was no significant difference in effect of party- vs candidate targeted flyers in the negative condition (F(1,93 = .203, p = n.s.). Only between participants in the positive condition were there significant differences in affect towards targeted candidate, with participants exposed to promoting Samsom flyers (M = 55.76, SD = 23.26) scoring significantly higher (F(1,89) = 6.98, p<.01) than participants exposed to flyers promoting the PvdA (M = 42.38, SD = 26.95). However, given the fact that this study aims to investigate the effects of negative
campaigning, and that the measurement of affect towards targeted candidate in the positive conditions functioned solely as a reference point and a means to contrast the scores for affect towards targeted candidate in the negative conditions, this finding was not of much interest to this study.
Overall, the results of the ANOVAs show that negative campaigning does not affect voter turnout or political trust, but it does have negative effects on external political efficacy. Furthermore, negative campaigning results in backlash against the sponsor of negativity yet does not decrease affect towards the target of attack ads. Lastly, findings show that there are
no significant differences in effect between party- and candidate-targeted negative campaigning.
The moderating role of personality on the effect of negative campaigning
To examine to what extent the Big Five dimensions of personality moderate the effect of negativity on the dependent variables in this study, eight hierarchical multiple regressions were conducted.
[Insert Tables 2-9]
Sex, age and education, left-wing and progressive ideology were entered as control variables in Step 1. All variables (except the dichotomous variable sex) were centered to avoid collinearity problems and make the outcome of the regressions more easily
interpretable. In Step 2, a dichotomous variable measuring type of campaigning (negative = 1) was entered into the model. Entering this variable in Step 2 allowed separate assessment of the increase in R-square that negative campaigning is responsible for. Step 3 followed with the introduction of the centered Big Five measures. Finally, the interaction terms, constructed by multiplying each of the Big Five personality traits with the dichotomous variable measuring negativity, were introduced in Step 4. Evidence for moderation by any of the Big Five was determined only when results showed a significant increase in R-square after introduction of the interaction terms. Furthermore, the unstandardized coefficient of the interaction term also needed to be significant.
The results consistently showed education and ideology to be significant predictors for voter turnout, political trust and political efficacy. Furthermore, adding negativity to the model contributed significantly to the model predicting external political efficacy (F(1, 177) = 5.70,
p<.05) and explained an additional 2.9% of variance, which was consistent with the earlier
findings in the ANOVA. Introduction of the Big Five increased R-square significantly
(F(5,172) = 4.87, p<.05) only for the model predicting political trust, with an additional 1.8% of variance explained: the coefficient shows that higher scores on Agreeableness predict higher scores on trust. However, when the interaction terms were introduced, the R-square of the models predicting voter turnout, political trust and political efficacy did not increase
significantly nor were any of the interaction terms significant. Therefore, I conclude that personality does not moderate the effect of negativity on any of the dependent variables measuring demobilization of the electorate.
Analyses of the hierarchical multiple regressions predicting affect towards sponsor and affect towards target of negative ads confirmed left-wing ideology as a significant positive predictor for affect towards the sponsor and a significant negative predictor for affect towards the target of political flyers. This finding was expected, as the sponsor of the ads was a left-wing political part while the target was right-left-wing. Adding negativity to the model contributed significantly to the model predicting affect towards sponsoring party (F(1, 177) = 4.66, p<.05) and explained an additional 2.2% of variance. This was consistent with the earlier findings in the ANOVA which showed that negative campaigning decreases affect towards a candidate sponsoring an attack ad. Furthermore, negativity also contributed to the model predicting affect towards the sponsoring candidate (F(1, 177) = 4.66, p<.05), explaining an additional 3.4% of variance. Again, this finding was consistent with the results of the ANOVA.
Introducing the Big Five as predictors did not yield significant R-square increase in any of the models predicting affect towards sponsor, nor in any of the models predicting affect towards the target of the negative ads. However, with the introduction of interactions in Step 4,
significant effects for the interaction term of Openness x negativity were found for the models predicting affect towards sponsoring party (ΔR = .05, ΔF(5,167) = 2.26, p<.05) and affect towards sponsoring candidate (ΔR = .06, ΔF(5,167) = 2.49, p<.05). To examine this effect in detail, the interaction between Openness and negativity was plotted for both models using a macro published online by Jeremy Dawson (2015). I plotted the simple slopes for participants who scored ± 1 SD from the mean level of Openness to Experience in my sample (see Figure 1).
[Insert Figure 1]
Surprisingly, Figure 1 shows that highly Open individuals’ (1 SD above mean) evaluation of the sponsoring party is unaffected by campaign tone (B = -4.02, p = n.s.). However, the effect of negativity on affect towards the sponsoring party is significant (B =
-15.00, p<.001) for people with low scores (1 SD below mean) on Openness, indicating that backlash from negative campaigning is particularly problematic for personalities with low scores on Openness.
Figure 1 shows a very similar effect of the interaction term of Openness and negativity. Again, highly Open individuals are not significantly affected by negative campaigning in terms of their affect towards the sponsoring candidate (B = 2.95, p = 54). In contrast, negative campaigning does have a significant negative effect (B = -18.18, p<.001) on participants scoring low on Openness, again indicating that backlash of negative
campaigning is particularly salient among people who score low on Openness to Experience. In terms of the constructed hypotheses about the moderating role of personality on the effects of negativity in political campaigning, the results from the regression analyses are quite clear: higher scores on Openness did not moderate the influence of negative
campaigning on affect towards the target of the negative ads. Furthermore, Agreeableness did not have a significant moderating influence on the effect of negative campaign on affect towards sponsor. Lastly, Neuroticism did not moderate affect toward the target of negative ads as hypothesized. Therefore, H6, H7 and H8 are rejected.
Discussion
The results of the research presented in this master thesis contribute to our knowledge of the effects of negative campaigning in a number of ways. Not only was this study one of the first to conduct experimental research on this subject outside a two-party context, it pioneers a new model for studying negativity by incorporating personality as a moderator of its effects.
First, in contrast to a number of authors warning about the demobilizing effects of negativity (Ansolabehere et al., 1999, 1994; Lau & Pomper, 2001a), I found that willingness to vote is not harmed by exposure to negative campaigning, nor did negativity have harmful effects on political trust. However, while negativity did not influence how participants felt about their own capabilities to participate in politics (internal political efficacy), the results did show that exposure to negative campaigning significantly lowered people’s feelings about
how well the government and politicians in general respond to demands of the electorate (external political efficacy), corroborating findings in earlier literature (Lau et al., 2007). The findings in this study thus support claims that voters are somewhat resilient against the demobilizing effects of negative campaigning (Brooks, 2006; Pinkleton et al., 2002). In addition, scholars have advised caution when using of negative campaigning, finding evidence of attack ads backlashing against their sponsor (Pattie et al., 2011; Walter & van der Brug, 2013) This is precisely what was found in this study: negative campaigning significantly decreased affect towards the sponsor of attack ads with both evaluation of the sponsoring party as well as evaluation of the sponsoring candidate being significantly lower among participants exposed to negativity. Surprisingly, no significant effect of negativity on affect towards the target of negative campaigning was found. Whereas most studies in two-party systems agree that negative campaigning decreases affect for the target of negativity (Lau & Pomper, 2001a; Lau & Rovner, 2009; Pinkleton, 1997), the results of this study indicate that this might not necessarily be the case in multi-party systems: only marginal negative effects of negativity were found on participants’ evaluations of both the targeted party and the targeted candidate.
Secondly, this study differentiated between the effects of party-targeted negativity vs candidate-targeted negativity, finding no evidence that either one is more effective in damaging the opponent or more detrimental to the workings of democracy. Interestingly, attacking a candidate was also not significantly different from attacking a party in terms of causing backlash. This was surprising as targeting candidates personally is often seen as uncivil and below-the-belt (Lau et al., 2007) and was found to lower evaluation of the sponsor in previous studies (Garramone et al., 1990). The results of the present study indicate that attacking a candidate can be seen as fair game as long as it is issue-based, such as in the stimulus material in this study.
Thirdly, I found that the moderating influence of personality on the effects of negative campaigning was largely absent in the present sample: no significant interactions between negativity and the Big Five were found for voter turnout, political trust, political efficacy and
affect towards the target of negativity. The lack of significant results could be due to the use of a short and admittedly limited measure of personality, the BFI-10 (Rammstedt & John, 2007). On the other hand, this study did support earlier findings on the main effects of
personality, with Agreeableness predicting political trust (Gerber, Huber, Doherty, & Dowling, 2011b) and Conscientiousness predicting voter turnout (Mondak & Halperin, 2008),
suggesting that the instrument did work as intended. Therefore, a more likely explanation is that the lack of significant findings is due to the small sample size (low power). While the present study had enough power to detect medium effect sizes within a 95% confidence interval (Cohen, 1992), detecting moderation effects might have required a larger sample, as finding interactions in general is notoriously difficult (McClelland & Judd, 1993). Finally, it should be noted that the present study used stringent tests by controlling for a number of variables including ideology, and all Big Five dimensions and their interactions were included in the same regression models.
Still, this study did find two interaction effects between personality trait Openness and negative campaigning. Openness, frequently found to be the strongest Big Five predictor of political attitudes and behaviour (Osborne & Sibley, 2012), was shown to moderate the effects of negativity on affect towards both the sponsoring party as well as the sponsoring candidate. The results revealed that participants scoring low on Openness were significantly more sensitive to negative campaigning in terms of their evaluation of the sponsor of attack ads. In contrast to highly Open individuals, whose evaluation of sponsor remained
unaffected, those low in Openness expressed greater signs of backlash, evaluating both sponsoring candidate as well as sponsoring party significantly lower when exposed to negative campaigning. This finding somewhat contrasts the results of Gerber et al. (2013), who found highly Open individuals to be more sensitive to political messages. However, a crucial differences between the present study and Gerber et al.’s is that the present study looked at negative messages, whereas Gerber et al. studied positive messages. As higher scores on Openness are associated with more in-depth thinking and tolerance to ambiguity (Gerber, Huber, Doherty, & Dowling, 2011b; McCrae & Costa, 1997), it is possible that those
scoring low on Openness evaluated the negative messages less proficiently and were thus more likely to come to the conclusion that negative campaigning is simply uncivil and rude, resulting in decreased affect towards the sponsor. Highly Open individuals might have been more inclined to think deeper about the argumentation in the attack ads and conclude that attacking the target was justified, leading to less dislike towards the sponsor of negative ads. In summary, based on these results, I can cautiously conclude that individual personality characteristics indeed can determine one’s reaction to negative campaigning, specifically in terms of evaluating politicians and political parties, with low-Open individuals being more sensitive towards negative campaigning.
In interpreting the results found in this study, it is important to keep in mind a number of factors that could limit the generalizability of the findings. First, though efforts were made to generate a diverse sample, participants in this study consisted largely of students and high-educated, left-wing voters, which makes generalizing to the general population somewhat difficult. Therefore, before definite conclusions can be drawn, this study should be replicated with a larger and more heterogeneous sample. Second, measuring the effects of one-time exposure to negative ads in a controlled experimental setting cannot index all of the effects negativity causes in the long-term (Lau & Rovner, 2009). Third, participants were only exposed to one single sponsor of negativity, as the sponsors of the ads used in this study needed to be kept identical. This was done to make comparison of sponsor and target evaluation possible between the positive and negative conditions. Future research could focus more explicitly on whether specific characteristics of sponsors of attack ads (i.e. extreme right-wing vs left-wing, new vs established, etc.) influence the effects of negativity.
Notwithstanding these limitations, this study presented the first tentative evidence that ‘going negative’ does not work as intended in the Netherlands. While not necessarily a demobilizing force, negative campaigning, whether party- or candidate targeted, does not seem to work to the advantage of the politicians utilizing it, and might even work against them, especially for voters with low scores on Openness.
References
ANES. (2013). User’s guide and codebook for the ANES 2012 Time Series Study. Ann Arbor, MI and Palo Alto, CA; The University of Michigan and Stanford University. Ansolabehere, S., Iyengar, S., & Simon, A. (1999). Replication experiments using aggregate
and survey data: the case of negative adversiting and turnout. American Political
Science Review, 93(4), 901–909.
Ansolabehere, S., Iyengar, S., Simon, A., & Valentino, N. (1994). Does attack advertising demobilize the electorate? American Political Science Review, 88(4), 829–838. http://doi.org/10.2307/2082710
Baines, P. R., Harris, P., & Lewis, B. R. (2002). The political marketing planning process: improving image and message in strategic target areas. Marketing Intelligence &
Planning, 20(1), 6–14. http://doi.org/10.1108/02634500210414710
Borkenau, P., & Ostendorf, F. (1998). The Big Five as states: how useful Is the Five-Factor model to describe intraindividual variations over time? Journal of Research in
Personality, 32(2), 202–221. http://doi.org/10.1006/jrpe.1997.2206
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. Brants, K., & Van Praag, P. (2006). Signs of media logic half a century of political
communication in the Netherlands. Javnost - The Public, 13(1), 25–40. http://doi.org/10.1080/13183222.2006.11008905
Brooks, D. J. (2006). The resilient voter: moving toward closure in the debate over negative campaigning and turnout. The Journal of Politics, 68(3), 684–696.
Brooks, D. J., & Geer, J. G. (2007). Beyond negativity: the effects of incivility on the electorate. American Journal of Political Science, 51(1), 1–16.
Bryant, J., & Miron, D. (2004). Theory and research in mass communication. Journal of
Communication, 54(4), 662–704. http://doi.org/10.1093/joc/54.4.662
Campus, D. (2010). Mediatization and personalization of politics in Italy and France: the cases of Berlusconi and Sarkozy. International Journal of Press/Politics, 15(2), 219–
235. http://doi.org/10.1177/1940161209358762
Carney, D. R., Jost, J. T., Gosling, S. D., & Potter, J. (2008). The secret lives of liberals and conservatives: personality profiles, interaction styles, and the things they leave behind.
Political Psychology, 29(6), 807–840. http://doi.org/10.1111/j.1467-9221.2008.00668.x
Cohen, J. (1992). A power primer. Quantitative Methods in Psychology, 112(1), 155–159. http://doi.org/10.1038/141613a0
Cooper, C. a., Golden, L., & Socha, A. (2013). The Big Five personality factors and mass politics. Journal of Applied Social Psychology, 43(1), 68–82.
http://doi.org/10.1111/j.1559-1816.2012.00982.x
Craig, S. C., Niemi, R. G., & Silver, G. E. (1990). Political efficacy and trust: A report on the NES pilot study items. Political Behavior, 12(3), 289–314.
http://doi.org/10.1007/BF00992337
Dawson, J. F. (2015) "Interpreting interaction effects." Interpreting Interaction Effects. Jeremy Dawson, n.d. Web. Oct. 2015. <http://www.jeremydawson.co.uk/slopes.htm>.
Elmelund-Præstekær, C. (2008). Negative campaigning in a multiparty system.
Representation, 44(1), 27–39. http://doi.org/10.1080/00344890701869082
Elmelund-Præstekær, C. (2010). Beyond American negativity: toward a general
understanding of the determinants of negative campaigning. European Political Science
Review, 2(01), 137. http://doi.org/10.1017/S1755773909990269
Fowler, E. F., & Ridout, T. N. (2012). Negative, angry, and ubiquitous: Political advertising in 2012. The Forum, 10(4), 51–61. http://doi.org/10.1515/forum-2013-0004
Freedman, P., & Goldstein, K. (1999). Measuring median exposure and the effects of
negative campaign ads. American Journal of Political Science, 43(October), 1189–1208. Fridkin, K. L., & Kenney, P. J. (2004). Do negative messages work? The impact of negativity
on citizens’ evaluations of candidates. American Politics Research, 32(5), 570–605. http://doi.org/10.1177/1532673X03260834
Fridkin, K. L., & Kenney, P. J. (2013). Variability in citizens’ reactions to different types of negative campaigns. American Journal of Political Science, 55(2), 307–325.
http://doi.org/10.1111/j.l540-5907.2010.00494.x
Garramone, G. M., Atkin, C. K., Pinkleton, B. E., & Cole, R. T. (1990). Effects of negative political advertising on the political process. Journal of Broadcasting & Electronic Media,
34(3), 299–311.
Geer, J. G. (2012). The news media and the rise of negativity in presidential campaigns. PS:
Political Science & Politics, 45(3), 422–427. http://doi.org/10.1017/S1049096512000492
Gerber, A. S., Huber, G. a, Doherty, D., Dowling, C. M., & Panagopoulos, C. (2013). Big Five personality trais and responses to persuasive appeals: Results from voter turnout experiments. Political Behavior, 35(4), 687–728.
Gerber, A. S., Huber, G. A., Doherty, D., & Dowling, C. M. (2011a). Personality traits and the consumption of political information. American Politics Research, 39(1), 32–84.
http://doi.org/10.1177/1532673X10381466
Gerber, A. S., Huber, G. A., Doherty, D., & Dowling, C. M. (2011b). The Big Five personality traits in the political arena. Annual Review of Political Science, 14(1), 265–287.
http://doi.org/10.1146/annurev-polisci-051010-111659
Gerber, A. S., Huber, G. a., Doherty, D., Dowling, C. M., Raso, C., & Ha, S. E. (2011). Personality traits and participation in political processes. The Journal of Politics, 73(03), 692–706. http://doi.org/10.1017/S0022381611000399
Gibson, R., & Römmele, A. (2001). A party-centered theory of professionalized campaigning.
Journal of Composite Materials, 6(4), 31–43. http://doi.org/0803973233
Hansen, K. M., & Pedersen, R. T. (2008). Negative campaigning in a multiparty system.
Scandinavian Political Studies, 31(4), 408–427.
http://doi.org/10.1111/j.1467-9477.2008.00213.x
Harrington, J. E., & Hess, G. D. (1996). A spatial theory of positive and negative campaigning. Games and Economic Behavior, 17(2), 209–229.
http://doi.org/10.1006/game.1996.0103
Hirsh, J. B., DeYoung, C. G., Xiaowen Xu, & Peterson, J. B. (2010). Compassionate liberals and polite conservatives: associations of agreeableness with political ideology and
moral values. Personality and Social Psychology Bulletin, 36(5), 655–664. http://doi.org/10.1177/0146167210366854
Hobolt, S. B., & Karp, J. a. (2010). Voters and coalition governments. Electoral Studies,
29(3), 299–307. http://doi.org/10.1016/j.electstud.2010.03.010
Jasperson, A. E., & Fan, D. P. (2002). An aggregate examination of the backlash effect in political advertising: The case of the 1996 U.S. Senate race in Minnesota. Journal of
Advertising, 31(1), 1–12. http://doi.org/10.2307/4189203
John, O., & Srivastava, S. (1999). The Big Five trait taxonomy: History, measurement, and theoretical perspectives. Handbook of Personality: Theory and Research. New York: The Guilford Press. Retrieved from
https://books.google.com/books?hl=en&lr=&id=iXMQq7wg-qkC&oi=fnd&pg=PA102&dq=the+big+five%3B+history,+measurement&ots=uD9739kxj8 &sig=E3QyNT-OhjreeQNNmykU_2YJbh0
Lau, R. R., & Pomper, G. M. (2001a). Effects of negative campaigning on turnout in U.S. senate elections, 1988–1998. The Journal of Politics, 63(03), 804–819.
http://doi.org/10.1111/0022-3816.00088
Lau, R. R., & Pomper, G. M. (2001b). Negative campaigning by US Senate candidates. Party
Politics, 7(1), 69–87. http://doi.org/10.1177/1354068801007001004
Lau, R. R., & Rovner, I. B. (2009). Negative campaigning. Annual Review of Political
Science, 12(1), 285–306. http://doi.org/10.1146/annurev.polisci.10.071905.101448
Lau, R. R., Sigelman, L., Heldman, C., & Babbitt, P. (1999). The effects of negative political advertisements: A meta-analytic assessment. American Political Science Review, 93(4), 851–875. http://doi.org/10.2307/2586117
Lau, R. R., Sigelman, L., & Rovner, I. B. (2007). The effects of negative political campaigns: A meta-analytic reassessment. Journal of Politics, 69(4), 1176–1209.
http://doi.org/10.1111/j.1468-2508.2007.00618.x
Mazzoleni, G., & Schulz, W. (1999). “Mediatization” of politics: A challenge for democracy?
