CAMPAIGN NEGATIVITY AND ITS EFFECTS ON VOTERS ENGAGEMENT. Fruzsina Nábelek. Corvinus University of Budapest

CAMPAIGN NEGATIVITY AND ITS EFFECTS ON VOTERS ENGAGEMENT Fruzsina Nábelek Corvinus University of Budapest Paper presented at the ECPR General Conference, 6-9 September 2017, Oslo, Norway Recent campaigns around the world seem to show a success of populist political rhetoric characterised by anti-politics, negativity and sometimes intentionally deceiving information provided to the voters. Critics about the lack of substance of political debate are becoming the main focus of evaluation of the political process of democracies although these worries have always been existing about modern political campaigns and their media representation. The crucial question behind these worries is how this style of political communication affects the citizens and their decisions, especially of voting and participating that are crucial elements that define the democratic process. One of the main critics about the development of political communication is that informative, policy-focused campaigns are being replaced by negative, enemy-seeking ones. Negativity in this sense substitutes substance of political debates, potentially denying voters the possibility of making well-informed decisions which follow their interests. Additionally critics imply that a highly negative, occasionally uncivil political debate might decrease the credibility, the sense of responsivity, and even the legitimacy of politicians by making voters stay away from politics. The effect of campaign negativity on voters engagement both towards the political system generally and towards competing parties, however, is not as evident as general critics consider as negative effects (demobilization, dealignment, ill-informed electorate) and the effectiveness (i.e. rising political engagement towards the campaigning party or candidate) are constantly debated. Research on the topic often show contradictory results, some of them providing cases of negativity decreasing voter turnout, of lower levels of partisanship and higher level of passive, indecisive and uninformed citizens, or of counterproductive effects on the party using negativity, while other results show a rising level of participation and partisanship of voters. All these contradictions show that effects of campaign negativity on voters decisions are not straightforward. On the other hand inconsistent results imply that negativity generally does not have the same effect in every communication situation. Experimental results strengthen this implication as they show that different types of negative information trigger different reactions of the subjects. Taking into consideration these differences the research examines whether the level of campaign negativity have an effect on an individual voters decision to participate at the elections. The effects of different types of messages is examined in comparative perspective, combining cross-national data on campaigning and post-election survey data of 8 European countries (Czech Republic, Denmark, Germany, the Netherlands, Poland, Portugal, Sweden, United Kingdom). Campaign negativity in the examination is used in the broadest sense: 1

attacking an opponent (a party or a person) on policy or on character. The research examines voters general engagement towards the political system measured by the (self-reported) participation at the previous elections, as voting at the elections can be considered the most basic way of participating in a society s political life. THEORETICAL BACKGROUND AND HYPOTHESIS DEFINITION OF NEGATIVE CAMPAIGNING AND RESEARCH ON ITS POSSIBLE EFFECT ON ELECTORAL TURNOUT Campaign negativity or negative campaigning has several definitions both in the academic literature and the public discourse. In the latter negative campaigning is mostly identified by mud-slinging, harsh, personal and occasionally uncivil attacks of the opponents. Media often reports on negative campaigning talking about the campaigns becoming more aggressive or nasty, while politicians blame each other with negativity, violation of campaign norms, or with distracting voters from more important questions by negative messages. Negative campaigning in this sense is compared to positive, more preferable and acceptable messages. Research on negative campaigning, however, uses a less normative approach, one that at the same time is closer to the interpretation of the users of negative campaigning (politicians and their consultants), as it considers negative campaigning a political communication tool that attack the opponents based on its arguments, issues, behaviour, characteristics or past performance. The aim of the negative campaign is generally to make the target seem less suitable for the position that is at stake at the election while explicitly or implicitly claim that the attacker would be a better choice. As Benoit (1999) argues positive and negative messages have different functions but the same goal of convincing voters: while positive appeals want to prove to the electorate that the candidate or party is the best to be elected, negative attacks use statements about the opponent that try to show that the other candidate or party is unfit to any position. In this sense the research uses the notion of negativity as any reference of a political actor to another political actor using a negative tone. Research on the possible effects of negativity can be considered to be the part of broader research interests such as the questions about the efficacy of political campaigns on one hand, and the effects of political communication, and especially of media coverage of political events on the other. Scientific research on the effects of campaigning show mixed results. The minimal effect theory argues that campaigning is not able to change significantly voters choices (Campbell, 2000; Holbrook, 1994; Shaw, 1999) but serves mainly to activate existing political predispositions and make them electorally relevant (Finkel, 1993: 1), thus, it fulfils more a mobilizing than a convincing one. On the other hand other results show the strong influence of agenda-setting, priming and framing on voters way of judging politics (McCombs & Shaw & Weaver, 1997; Iyengar & Simon, 1993; Entman, 1993), the effect on knowledge on policy issues (Norris & Sanders, 2003), on perception of candidate images (Kaid & Holtz-Bacha, 1995), and on perception of issue-ownership (Bowler et al. 1992) etc. 2

proving that campaigning does matter although the importance and strength of these effects may remain questionable (Farrell & Schmitt-Beck 2002). The practical proof of the importance of campaigns is that the audience and the politicians consider them important, the latter even willing to spend growing amount of resources on campaigning. Media coverage of campaigning and generally of politics, on the other hand is often criticized stating that the media focus on negativity, scandals and tabloid news on the expense of more important issues. The most pessimistic critiques, in summary referred as media malaise theory (Robinson, 1967; Kennett, 1999), state that the way media cover political contest creates disengaged and cynic voters who chose to stay away from politics which weakens the very core of the democratic process. Research on negative campaigning probably became prevalent reacting to these arguments in the 90s, focusing on the effects of negativity of political campaigns and of their media coverage especially on political participation. Ansolabehere and Iyengar s demobilization theory (1994, 1995) argues that negative campaigning has a notable effect on the electorate mainly on their decision to participate. As they argue while positive campaigning strengthens existing preferences and provides a motivation to vote, negative campaigning does not fulfil this aim but creates uncertainty about the original preferences and at the same time gives no motivation to vote neither to the supporters of the attacker, nor to the uncertain voters resulting in lower levels of turnout. Thus, according to the theory of Ansolabehere and Iyengar, negativity creates uncertain and undermotivated citizens, disengagement towards political actors and the entire political system, and it causes a declining trust, making voters feel less capable to influence the political process. The demobilization theory has received a lot of attention both in the American public discourse both in academia as it scientifically confirmed the existing poor judgement of negative campaigning. If negative campaigning actually decreases voter turnout and engagement of citizens, critics claiming that negativity destroys the quality of democracy, could be true (Brooks, 2006). Therefore, the demobilization hypothesis confirms that negativity is not only a question of campaign strategy but a phenomenon threatening the entire political system, which may create the need to regulate the use of negative campaigning (Mayer, 1996). Ansolabehere and Iyengar s work has undoubtedly meant a powerful foundation of the scientific research of negative campaigning, however their results has been questioned on several basis. First, methodological critics articulated the limits of experimental design. The data analysis of the authors which completed the experimental results were also critized. Some of the reactions showed the flaws of data collection (Finkel & Geer, 1998; Wattenberg & Brians, 1999) and handling of data on voter turnout (Brooks, 2006). Second, several authors suggest the involvement of new variables to see more accurate results. Freedman and Goldstein (1999 and 2002) and Finkel and Geer (1999) argue that the involvement of exposure to negative ads is necessary in the analysis of effects. Lau and Pomper (2002) claim that the expected results should be considered as well, as those who are behind in the contest tend to use more negative messages. Krupnikov (2011) states that timing of negative messages has an effect as the decision to vote is made in two phases: first, the choice of the preferred candidate and second, the decision to participate. In the first phase negative 3

messages can help the voter, while in the second phase negativity can increase uncertainty which might lead to abstention. Third, the broader effects on the political system are questioned as well. As Norris (2002) argues participation levels, activism, interest in politics are fluctuating in the United States and no steady decline can be shown. Parallel to these critics later research does not confirm the demobilization hypothesis. Research conducted using different methods such as aggregated data on participation, surveys on individual intention to participation, experiments and combination of these methods either showed that negative campaigning had no significant effect on participation (Finkel & Geer, 1998; Lau et al, 1999; Sigelman & Kugler 2003; Brooks, 2006) or that the opposite of demobilization effect is true, thus, that negative campaigning raises the level of participation (Wattenberg & Brians, 1999; Goldstein & Freedman, 1999, 2002; Lau & Pomper, 2002; Geer & Lau, 2005). Theoretical arguments about the possible mobilizing effect of campaign negativity claim that, unlike the hypothesis of Ansolabehere and Iyengar, negative information does provide a motivation to vote by questioning the performance of the incumbent and by showing the possible threats of a poor electoral decision that need to be avoided by the voters. More nuanced approach in the research of negative campaigning seem to strengthen these ideas as they show that voters do not consider negative messages less informative than positive ones, and they do use them to make their judgements about the candidates (Pinkleton, 1997; Brooks & Geer, 2007; Stevens et al. 2008; Sides & Lipsitz & Grossmann, 2010; Fridkin & Kenney, 2011). On the other hand some groups of voters are shown to be more responsive to negative messages: although indecisive voters generally find negative messages less acceptable and informative they are influenced by them more than voters with strong party or candidate preference. As Zaller (1992) argues voters with strong existing preferences make a selection processing the information about the preferred candidate or party: they tend to accept information that fit their preconceptions while they perceive less those that are against them. Regarding to campaign negativity this means that voters with stable preferences are affected less by negativity especially when it is against their preferred party contradicting to their preconceptions. On the other hand independents make less selection between the pieces of information available meaning that they might perceive more negative messages and at the same time they might be more open to these as existing preferences do not shape their perception of the campaign. At the same time, it is also demonstrated that negative messages have a higher level of memorability during the campaign than positive ones, thus, negative messages possibly influence voters decisions easier than positive messages (Lau et al. 2007). This corresponds with the results of cognitive psychology that show the existence of the so called negativity bias i.e. the higher impact of negative events, information, feeling etc. on evaluations, on considerations of risks, and on decision making than positive ones (see Nai & Walter S. 2015). On the other hand there are differences between the effects of different types of negative messages. Generally uncivil and personal type of negativity receive less approval from the voters, possibly causing a backlash effect on the attacker although type of messages are still shown to be used by the voters. 4

Previous research, thus show that negativity might have an effect on voters behaviour regarding to participation and on their actual vote as well. However these effects differ when we consider different election situation, different types of negative messages and different groups of voters. Considering these, the following hypothesis can be made. First, generally considered the amount of negative messages has no demobilizing effect on one s decision to participate at the election. If voters tend to use negative messages in their decision the same way than positive messages the level of negativity in a campaign should not explain the intention of voting. On the other hand if they consider them more important than positive ones, negativity might mean a motivation for voting. Second, the level of personal or integrity type of negative messages might affect participation. It is generally confirmed in the literature that too personal negative campaigning is refused by the voters meaning that these can provoke strong emotions of the voters like disapproval or even disgust. If these type of messages cause a backlash effect they might motivate some voters to participate to vote against the sponsor of the attack. However, a general backlash effect might appear as well against the whole process of election as this kind of debate might undermine the credibility of the participants and the perceived importance of the elections causing some voters to not participate at all. Third, the level of negativity might affect the decision to participate among voters without an existing party preference. As an alternative hypothesis it can be assumed that if there is an existing effect of negative campaigning on participation it should be stronger among independents than among partisans as they do not have strong pre-existing concepts about the political actors that are hardly changed by campaigning, and at the same time they might be more exposed to negative messages as well. Finally, about partisans it can be assumed that if negativity has an effect on their participation a) the level of negativity used by their party have a mobilizing effect but b) the level of negativity used against their party will have no effect on their decision to vote. In the first case the negative messages used by their own party provide incentives to vote as they are intended to show the level of threat that an opponent s win might mean. On the other hand negative campaigning against one s party will have no effect as partisans tend to ignore pieces of information that are against their existing preconceptions. CASE SELECTION AND DATA Examining these hypotheses I use cross-national data on campaigning and participation on 15 European elections. The research on the possible effects of campaign negativity on political behaviour shows inconclusive results depending not only on the research methods used (Fridkin & Kenney, 1999; Martin, 2004) but on the cases examined as well. While generally most of the research has focused on the US campaigns, examinations on European campaigns has recently appeared as well (for an overview see Nai & Walter, 2015), however these are still failing to clear the contradictory results seen before. One reason can be that the research of negative campaigning, as most of the research on the field of political communication, tends to focus on specific cases: on a single country, a campaign, a party or a person. Comparative research on political communication and more specifically on campaigning is often a topic-focused (see for example Swanson & Mancini, 1996 on mediatization, Plasser & Plasser, 2004 on professionalization, Farrell & Schmitt-Beck on campaigns, Nai & Walter, 5

2015 on negative campaigning etc.) description of these single cases which as Holtz-Bacha and Kaid (2011) argues limits the possibility of generalization. As they point out generalization in these studies is not only difficult because of methodological limitations (mostly they use data collected with different methods and based on different understandings of the same concept) but because campaigns, as special parts of the political process depend much on their contexts. To avoid these limitations the research uses cross-national data on campaigning of 8 European countries based on content analysis of newspaper coverage of the 13 election campaigns examined (see Table 1). The data collection was conducted by the Comparative Campaign Dynamics Project (Debus et al. 2016) which examined newspaper coverage of 19 national election campaigns in 9 European countries (Czech Republic, Denmark, Germany, Hungary, the Netherlands, Poland, Portugal, Sweden, United Kingdom) 1. In each country country-teams conducted content analysis of articles related to the two latest national elections in two national newspapers. To avoid the selection bias of media outlets, in each country the major left and right leaning newspapers were selected 2. The period of examination was the fourweek period before the election s day in case of each election. For this period every article of front pages were coded and a random sample of the inner-pages articles were analysed as well. To provide intercoder reliability, in every country-team each sampled article were coded independently by three coders and coders were asked to evaluate how confident they were in their coding decisions. Based on this each element of the content-analysis that appears in the data is based on (1) agreement of the three coders, (2) agreement of two coders, if both of them were at least medium confident about their coding or (3) coding of one coder who was fully confident in her coding. In case of every article general information were recorded (election year, newspaper, title of the article, page etc.) and the subject(s) of the article was coded as well. The subjects of the articles were the most important parties participating in the election, and each article could have more than one subject, each of them coded separately with the help of an online coding questionnaire. Thus, the unit of observation in the content analysis was an actor making statements in an article. Regarding to each coding questions, coders were asked to provide citations from the article to explain their coding decisions making it easier to decide cases where the coders were uncertain about the coding. In the questionnaire about each subject it was coded who was speaking on behalf of the subject, what policy issue or issues were addressed (if there were any), whether the subject was speaking about spending in connection with the policy issue(s) mentioned, and what issue-related or general valence content appeared in their statement. Regarding to valence content categories like honesty/integrity of the party or the party leader, competence of the party or the party leader, party unity and leader charisma could be coded (and more than one valence content could be coded for each issue or subject). Direction, i.e. positive/neutral/negative, of the valence categories were coded as well. Issues and valence contents were not only recorded for the subjects talking about themselves but about every other party that they were making statements about in the 1 In this analysis those election campaigns are used for which post-election surveys are available. 2 For the list of newspapers examined see Appendix 1. 6

article. Finally, at the end of the coding survey journalist framing was analysed as well: general tone of the article towards the subject (positive/neutral/negative) and dominant issue of the article were identified. Raw data of the content analysis were restructured in three databases: subjects talking about themselves (database: Self), subjects talking about other political actors (Other), and database of journalist framing of the election (Journalist) each containing statements as cases. In the analysis I will use data on subjects statements about their own policy positions and valence (Self) which by definition can be considered as positive messages of the election campaign. On the other hand I will use the dataset on statements about the opponents which in most cases can be considered as negative messages (Other). The two datasets are aggregated on party and election level creating four measurements of campaign negativity for each election examined: 1. the proportion of negativity of each election, 2. the proportion of integrity (party or leader) and leader charisma type of negative messages (referred in the analysis as negative-personal messages) 3. the proportion of negativity used by each party at each campaign, 4. and the proportion of negativity used against each party at each campaign. These indicators will serve as independent variables in the analysis, thus, it will be examined whether they have an effect on one s decision to participate. The indicators of campaign tone are added to individual level data of post-election surveys. In most of the cases included in the analysis the sources of the survey data are the 3rd and 4th (advance release) modules of Comparative Study of Electoral Systems (CSES 2015 & 2017). The CSES research programme provides a common survey of each participating countries of elections between 2006 and 2011 and between 2011 and 2016. In case of the recent election of the Netherlands (DPES, 2012) and the 2010 election in UK (Whitely & Sanders, 2010) I use post-election surveys that were conducted in the same manner as in CSES but are not included (yet) in the pooled database of the 4th module, therefore, in these cases the national election studies are used. In case of Denmark, Sweden and Portugal, however, adequate survey data is only available for one election each (see Table 1). The dependent variable of the analysis will be a dichotomous variable on whether the respondent has voted based on the question whether he or she has cast a ballot at the recent election. 7

1. Table - Election campaigns examined with data source of the dependent variable and control variables, and the number of individual cases included in the analysis 3 Czech Republic 2010 CSES Module 3 N=1857 Denmark 2007 CSES Module 3 N=1442 Germany 2009 CSES Module 3 N=2095 the Netherlands 2010 CSES Module 3 N=2153 Poland 2007 CSES Module 3 N=1817 2013 CSES Module 4 N=1653 2013 CSES Module 4 N=1889 2012 Dutch Parliamentary Election Study 2012 N=1677 2011 CSES Module 4 N=1919 Portugal 2009 CSES Module 3 N=1316 Sweden 2014 CSES Module 4 N=832 United Kingdom 2010 The British Election Study 2010 N=1577 2015 CSES Module 4 N=1567 WHAT EFFECT PARTICIPATION? CONTROL VARIABLES OF THE ANALYSIS Campaign communication is only one aspect that might influence voters participation at the elections. In the research on individual participation and electoral turnout several factors appear to have a mobilization effect. These in the analysis will be included as control variables. According to Franklin s (2001) summary on what affects electoral turnout and differences of turnout between political systems three groups of factors can be distinguished that play a role in the level of participation on country and individual level. First, instrumental motivations or in other words institutional context define the rules of elections of which some might motivate a higher turnout than others. Second, some of the individual resources of the electorate make it possible to invest in the costs of participation like time, financial costs of participation or the ability and motivation of information-seeking. Third, the level of political mobilization regarding to an election affects participation as well. The mobilization factor generally includes mobilization efforts of the candidates and parties, discussion of politics among smaller groups of individual, campaign communication, and potentially the level of negative campaigning as well. 3 The final database includes 21 794 cases. 8

Among the institutional rules all those electoral situations that make the value of a vote higher stimulate participation. As Downs (1957) argues one element of the explanation of participation at the elections might be if it seems probable that casting a vote will count in the outcome of the elections, even though the probability of a single vote making any difference is extremely small. Voting in this sense is not a rational act however a voter s perception of the importance of his or her vote might be different. Generally it is argued that the more proportional the electoral system is, the lower is the chance that one s vote will be perceived as wasted as even the votes casted on smaller parties can result in a mandate (Blais & Carty, 1990; van der Eijk et al. 1996). Therefore, in the analysis among institutional explanations a variable on the electoral system (proportional or majoritarian) will be included. Another institutional factor related to higher stakes of a vote is the competitiveness of the race, in which case the more competitive the race is the higher the value of a vote is, thus, the higher is the motivation to participate at the elections (Blais & Dobrzynska 1990; Jackman & Miller, 1995). In the analysis I will measure the competitiveness of the race by the margin of the winner and runner-up party. If the margin was high, we can assume that there was a clear winner meaning that the stakes of the election probably seemed lower during the campaign. Although, institutional or country level factors are proved to be stronger influencers of turnout (Franklin, 2001) some individual level, socio-economic characteristics of voters are shown to have an effect on participation. Inter-relating factors like age, level of education, income or employment are proved to influence the intention to vote: the older, the more educated, and those who have higher income and better employment status are more likely to vote than the younger, less educated, and those who have lower income and worse employment status (see Powell, 1986; Rosenstone & Hansen, 1993; Miller & Shanks, 1996). Besides socio-economic factors, some other individual characteristics can be determinant in participation. On the other hand, Zaller (1992) argues that the individual level of political interest and information-seeking influences how much one is affected by political campaigning: more informed voters are claimed to be less influenced by the campaign. At the same time they are expected to be more involved in politics (see Norris, 2000 s virtuous circle), therefore they are more likely to participate. To measure informedness of voters in each post-election survey included in the analysis three questions on political knowledge were asked from the respondents. The control variable of informedness measures how many correct answers the respondent have given to these questions. The respondents satisfaction with democracy is used as a control variable as well as we can assume that those who do not feel satisfied with the democratic process are less likely to vote. Finally, the strength of partisanship is included in the analysis as stronger partisanship is expected to decrease the intensity of campaign effects. Following the hypotheses, the effect of negativity and different types of negative messages on the individual self-reported participation at the election is examined among the whole electorate and among smaller groups independents and partisans, including the control variables regarding to institutional characteristics of the election and to individual characteristics of the respondents. As having voted is a dichotomous variable (0= have not voted, 1=voted) for the analysis binary logistic regressions are used. The cases of the 9

combined database are weighted by a combination of the design weight (correcting for the randomness of sampling) provided to each post-election survey and a weight based on the different sample sizes of the dataset. As I use country level variables and the observations within a country are not fully independent from each other standard errors are clustered by countries in the analysis. RESULTS First, in a base model the role of the control variables are examined. The odds ratios reported in Table 2 show how much a unit of change in the control variable effects the odds of having participated at the election. An odds ratio higher than one shows a positive effect meaning that the variable raises the odds of participation while an odds ratio smaller than one refers to a negative, in this case demobilizing effect. In the base model the examination of the control variables shows that basic demographic variables except for gender affects whether the respondent has voted at the elections or not. Among the level education shows the strongest effect: a higher level of education increases the odds of voting by 1.5, meaning that a person who completed a lower secondary school is 1.5 times, while one that has completed higher secondary school is 3 times more likely to have voted than a respondent who only finished the primary. Other individual level factors examined in the model show positive effect as well. Those who are more informed about politics are more likely to vote and those who are more satisfied with the working of democracy in their country are more willing to participate as well. On the other hand, as expected, being partisan increases the probability of voting as well, among those who claimed that they have felt closer to a party it was almost four times more likely to vote at the election than among those who claimed that they had not felt close to any party at all. Among the institutional factors the electoral system seems to have the strongest effect while the time since the election shows significant but relatively weak effect on the intention to vote. The competitiveness of the race seems to increase slightly the probability of voting as well: the smaller the margin was between the winning and the runnerup party, the more likely the respondents chosed to participate. After examining the base model the level of negativity (Model 1) and the level of personal type negative (Model 2) messages were included in the analysis. Examining all groups of voters, however, negativity shows no significant effect on voting in neither of the models. These results imply that examining individual voters, negative campaigning in general has no effect on participation. Personal type of messages that are generally considered to be harsher and to have a more alienating effect show no effect either suggesting that voters, when deciding to vote, do not consider this kind of campaigning in a different way than other styles of campaigning. 10

Table 2 Binary logistic regression results explaining participation with the proportion of negative and negative-personal type messages Variables Base Model1 Model2 Gender 0.88 0.90 0.86 (0.11) (0.10) (0.11) Age 1.03** 1.03** 1.03** (0.004) (0.04) (0.00) Employed 1.10** 1.08* 1.08* (0.07) (0.04) (0.05) Education 1.49** 1.48** 1.45** (0.17) (0.17) (0.17) Household income 1.14** 1.15** 1.15** (0.02) (0.02) (0.02) Partisan 3.61** 3.51** 3.69** (0.60) (0.63) (0.57) Informed Satisfaction with democracy Proportional electoral system Time since last election Margin between 1st and 2nd party Negative messages Negative personal messages Constant 1.51** (0.17) 1.43** (0.13) 1.79** (0.37) 1.02* (0.86) 0.92** (0.04) 0.02** 1.51** (0.17) 1.44** (0.13) 1.42 (0.44) 1.02** 0.90* (0.05) 0.96 (0.05) 0.10** (0.21) 1.50** (0.16) 1.43** (0.13) 1.42* (0.26) 1.02** 0.92* (0.04) 0.92 (0.05) 0.07** (0.07) N 15859 15859 15859 Pseudo R 2 0.18 0.19 0.19 Binary logistic regression reporting odds ratios. Robust standard errors clustered to country in parenthesis. * p<.1; **p<.05 Although negative campaigning generally does not seem to have an effect on participation it is still interesting to examine whether different groups of voters respond in a different way than the average of voters. First, the theory suggests that those who do not have stable party preferences are more responsive to campaign messages. Analysis of the independent voters in the countries examined, however, shows that regarding to negative campaigning that is not the case. Just as the average of voters, independents are effected by socio-economic, individual and country level factors while the proportion of negative a personal type negative messages had no effect on whether they had decided to vote at the elections. 4 4 For the results of the models regarding to independents see Appendix 2. 11

Examining the voters who claim to have some partisan attachment shows the same results regarding to negative messages and negative-personal type messages. The amount of negative messages used against and used by the preferred party shows no significant effect on partisan s participation either. 5 Partisans, however, who perceive themselves not very convinced adherents to a party are different from those who claim themselves convinced or very convinced supporters of a party. While the latter two groups are not effected by negative campaigning the decision of voting in case of those who are not strong supporters of their party, besides other individual and country level factors, were influenced by negative campaigning although the strength of the effect is rather small (see Table 3). Examining the effect of the proportion of negative campaign messages in general it can be said that with the growth of campaign negativity the odds of having participated at the elections decreases by only 0.08. Personal type of messages have a slightly stronger demobilizing effect implying that not very convinced partisans might consider these type of messages differently than negativity in general. Negativity used by the supported party has a significant but a very limited negative effect on participation in the analysis as well. At the same time, the supported party being attacked shows no effect on not convinced partisans either. 5 See Appendix 3 12

Table 3 Binary logistic regression results explaining the participation of not very convinced adherents of a party Participation of not very convinced adherents of a party Variables Model1 Model2 Model3 Model4 Gender 0.84 0.84 0.90 0.92 (0.93) (0.89) (0.10) (0.16) Age 1.03** 1.03** 1.03** 1.03** (0.07) Employed 0.91** 0.92 0.98 0.94 (0.13) (0.13) (0.13) (0.17) Education 1.28** 1.27** 1.30** 1.33** (0.15) (0.15) (0.16) (0.18) Household income 1.17** 1.15** 1.15** 1.14** (0.05) (0.05) (0.05) (0.04) Informed Satisfaction with democracy Proportional electoral system Time since last election Margin between 1st and 2nd party Negative messages Negative personal messages Negative messages used by the preferred party Negative messages targeted against the party Constant 1.48** (0.15) 1.30** (0.14) 0.62** (0.07) 1.01 0.87** (0.06) 0.92** (0.02) 6.55 (8.05) 1.48** (0.15) 1.32** (0.15) 0.78** (0.08) 1.02** 0.88* (0.06) 0.89** (0.03) 1.20 (1.00) 1.45** (0.15) 1.37** (0.13) 0.85 (0.14) 1.02* 0.87** (0.06) 0.98** (0.11) 0.66 (0.39) 1.47** (0.18) 1.37** (0.12) 0.96 (0.14) 1.02* 0.87** (0.05) 0.99 0.31** (0.15) N 2912 2912 2912 2662 Pseudo R 2 0.11 0.11 0.11 0.11 Binary logistic regression reporting odds ratios. Robust standard errors clustered to country in paranthesis. * p<.1; **p<.05 CONCLUSION The effects of negative campaigning on political engagement are constantly debated in the literature. Studies on the effect on participation show either demobilizing or mobilizing effects or no effect at all. At the same time these studies generally focus on one country, the majority of them on elections in the United Stateds. Individual country studies, however, limit the possibility of generalization, thus the purpose of this paper was to examine the effects of negativity in cross-national perspective and on cases outside the US context. The dataset of Comparative Campaign Dynamics provided content analysis conducted with the same 13

methods in several European countries and elections. From the datasets campaign data on each election for which post-elections surveys were available were used in the analysis making possible the examination of the effects of negativity for 13 election campaigns. The results show that the demobilizing effect of negativity in general can not be detected analysing several election campaigns. Contrary to the literature the effect of personal type negative messages that are generally considered more rejected by voters show no effect either. At the same time it was assumed that different type of voters might respond to negative campaigning in different ways. Independents can be expected to meet with more negative campaign messages, and to be more likely to refuse these causing a backlash effect of staying at home. Partisans however were expected to be mobilized by the negativity of their party as negative campaigning in this case aims to activate their existing negative emotions about the opponent and to call attention to the importance of voting. However, neither independents, nor partisans were found to be influenced by negative campaigning regarding to their participation at the elections. In the case of a smaller group of partisans, those who claim to be not very convinced supporters of their party, a demobilizing effect can be detected. Although the effect is small, this implies that this group of voters is different from convinced partisans and is more proned to be influenced by campaign negativity. These results in suggest that generalization based on individual country studies on the effects of negative campaigning has its limitations as examining different cases might show contradictory result. Examining more (and European) countries the demobilization effect is very limited compared to other individual and country level factors. On the other hand, the results imply that there are differences not only between the electoral contexts but between the different groups of voters as well which suggests that in the analysis of the effects of campaign communication this difference should be taken into account as well. 14

BIBLIOGRAPHY Ansolabehere, Stephen Iyengar, Shanto Simon, Adam Valentino, Nicholas (1994): Does Attack Advertising Demobilize the Electorate? In: American Political Science Review, Vol. 88, No. 4. 1994., p. 829-838. Baumann, Markus & Gross, Martin (2016): Where Is My Party? Introducing New Data Sets on Ideological Cohesion and Ambiguity of Party Positions in Media Coverage. Mannheimer Zentrum für Europäische Sozialforschung: Working Paper. Available: http://www.mzes.unimannheim.de/d7/en/publications/report/where-is-my-party-introducing-new-data-sets-onideological-cohesion-and-ambiguity-of-party-positions-in-media-coverage Benoit, W. L. (1999). Seeing spots: A functional analysis of presidential television advertisements, 1952-1996. Greenwood Publishing Group. Blais, A., & Carty, R. K. (1990). Does proportional representation foster voter turnout?. European Journal of Political Research, 18(2), 167-181. Blais, A., & Dobrzynska, A. (1998). Turnout in electoral democracies. European Journal of Political Research, 33(2), 239-261. Blais, A. (2006). What affects voter turnout?. Annu. Rev. Polit. Sci., 9, 111-125. Bowler, S., Broughton, D., Donovan, T., & Snipp, J. (1992). The informed electorate? Voter responsiveness to campaigns in Britain and Germany. In Electoral Strategies and Political Marketing (pp. 204-222). Palgrave Macmillan UK. Brooks, Deborah Jordan (2006): The Resilent Voter: Moving Toward Closure in Debate over Negative Campaigning and Turnout. In: The Journal of Politics, Vol. 68, No. 3. 2006. Aug., p. 684-696. Brooks, Deborah Jordan Geer, John G. (2007): Beyond Negativity: The Effects of Incivility on the Electorate. In: American Journal of Political Science, Vol. 51. No. 1. 2007. Jan p. 1-16. Campbell, A., Converse, P. E., Miller, W. E., & Donald, E. (1966): The american voter. University of Chicago Press. Campbell, J. E. (2000). The American Campaign. College Station: Texas A&M University. Carraro, Luciana Castelli, Luigi (2010): The Implicit and Explicit Effects of Negative Political Campaigns: Is the Source Really Blamed? In: Political Psychology, Vol. 31. No. 4. 2010, p. 617-645. Claassen, R. L. (2011). Political Awareness and Electoral Campaigns: Maximum Effects for Minimum Citizens?. Political Behavior, 33(2), 203-223. The Comparative Study of Electoral Systems (www.cses.org). CSES MODULE 3 FULL RELEASE [dataset]. December 15, 2015 version. doi:10.7804/cses.module3.2015-12-15 15

The Comparative Study of Electoral Systems (www.cses.org). CSES MODULE 4 FOURTH ADVANCE RELEASE [dataset]. April 11, 2017 version. doi:10.7804/cses.module4.2017-04- 11 Denny, K., & Doyle, O. (2009). Does voting history matter? Analysing persistence in turnout. American Journal of Political Science, 53(1), 17-35. Downs, A. (1957). An economic theory of political action in a democracy. Journal of Political Economy, 65(2), 135-150. Entman, R. M. (1993). Framing: Toward clarification of a fractured paradigm. Journal of communication, 43(4), 51-58. Farrell, D. M., & Schmitt-Beck, R. (2002). Do political campaigns matter?: campaign effects in elections and referendums. London ; New York: Routledge. Finkel, S. E. (1993). Reexamining the" minimal effects" model in recent presidential campaigns. The Journal of Politics, 55(1), 1-21. Finkel, Steven E. Geer, John G. (1998): A Spot Check: Casting Doubt on the Demobilizing Effect of Attack Advertising. In: American Journal of Political Science, Vol. 42. No. 2. 1998. Apr., p. 573-595. Franklin, M. N. (2001). The dynamics of electoral participation. In: Leduc, L. Niemi, R. Norris, P. (2001): Comparing democracies 2: Elections and Voting in Global Perspective Thousand Oaks CA: Sage, 148-168. Freedman, Paul Goldstein, Ken (1999): Measuring Media Exposure and the Effects of Negative Campaign Ads. In: Americal Journal of Political Science, Vol. 43. No. 4. 1999. Oct, p. 1189-1208. Fridkin Kahn, Kim Geer, John G. (1994): Creating Impressions: An Experimental Investigation of Political Advertising on Television. In: Political Behavior, Vol. 16. No. 1. 1994. March, p. 93-116. Fridkin Kahn, Kim Kenney, Patrick J. (1999): Do Negative Campaigns Mobilize or Suppress Turnout? Clarifying the Relationship between Negativity and Participation. In: American Political Science Review, Vol. 93. No. 4. 1999. Dec., p. 877-889. Fridkin, Kim L. Kenney, Patrick J. (2011): Variability in Citizens Reactions to Different Types of Negative Campaigns. In: American Journal of Political Science, Vol. 55. No. 2. 2011. April, p. 307 325. Geer, John G. (2008): In Defense of Negativity. University of Chicago Press, Chicago. Gervais, B. T. (2014). Following the news? Reception of uncivil partisan media and the use of incivility in political expression. Political Communication, 31(4), 564-583. 16

Goldstein, Ken Freedman, Paul (2002): Campaign Advertising and Voter Turnout: New Evidence for a Stimulation Effect. In: The Journal of Politics, Vol. 64. No. 3. 2002. August, p. 721-740. Green, D., & Shachar, R. (2000). Habit Formation and Political Behaviour: Evidence of Consuetude in Voter Turnout. British Journal of Political Science, 30(4), 561-573. Holbrook, T. M. (1994). Campaigns, national conditions, and US presidential elections. American Journal of Political Science, 973-998. Holtz-Bacha, C. & Kaid, L. L. (2011): Political Communication across the World: Methodological Issues Involved in International Comparisons. In: Bucy, E. P. & Holbert L. (eds.): The Sourcebook for Political Communication Research. Methods, Measures, and Analytical Techniques. London ; New York: Routledge. Iyengar, S., & Simon, A. (1993). News coverage of the Gulf crisis and public opinion: A study of agenda-setting, priming, and framing. Communication research, 20(3), 365-383. Jackman, R. W., & Miller, R. A. (1995). Voter turnout in the industrial democracies during the 1980s. Comparative Political Studies, 27(4), 467-492. Krupnikov, Yanna (2011): When Does Negativity Demobilize? Tracing the Conditional Effect of Negative Campaigning on Voter Turnout. In: American Journal of Political Science, Vol. 55, No. 4. 2011. Octo, p. 796-812. Lau, Richard R. Sigelman, Lee Heldman, Caroline Babbitt, Paul (1999): The Effects of Negative Political Advertisements: A Meta-Analytical Assessment. In: American Political Science Review, Vol. 93. No. 4, 1999. Dec., p. 851-875. Lau, Richard R. Pomper, Gerald M. (2002): Effectiveness of Negative Campaigning in US Senate Elections. In: American Journal of Political Science, Vol. 46. No. 1. 2002. Jan., p. 47-66. Lilleker, Darren G. (2006): Key Concepts in Political Communication. SAGE, London. Maisel, L. S. (2012). The negative consequences of uncivil political discourse. PS: Political Science & Politics, 45(03), 405-411. Martin, P. S. (2004). Inside the black box of negative campaign effects: Three reasons why negative campaigns mobilize. Political psychology, 25(4), 545-562. Mayer, William G. (1996): In Defense of Negativity. In: Political Science Quarterly Vol. 111. No. 3. 1996, p. 437-455. McCombs, M. E., Shaw, D. L., & Weaver, D. H. (1997). Communication and democracy: Exploring the intellectual frontiers in agenda-setting theory. Psychology Press. Mutz, D. C., & Reeves, B. (2005). The new videomalaise: Effects of televised incivility on political trust. American Political Science Review, 99(01), 1-15. 17

Nai, A., & Walter, A. S. (Eds.). (2015). New perspectives on negative campaigning: Why attack politics matters. ECPR Press. Norris, Pippa (2000): A Virtous Circle. Political Communications in Postindustrial Societies. Cambridge University Press, New York. Norris, P., & Sanders, D. (2003). Message or medium? Campaign learning during the 2001 British general election. Political communication, 20(3), 233-262. Pinkleton, Bruce (1997): The Effects of Negative Comparative Political Advertising on Candidate Evaluations and Advertising Evaluations: An Exploration. In: Journal of Advertising, Vol. 26. No. 1. 1997. tavasz, p. 21- Plasser, F., & Plasser, G. (2002). Global political campaigning: A worldwide analysis of campaign professionals and their practices. Greenwood Publishing Group. Powell Jr, G. B. (1986). American voter turnout in comparative perspective. The American Political Science Review, 17-43. Roddy, Brian L. Garramone, Gina M. (1988): Appeals and Strategies of Negative Political Advertising. In: Journal of Broadcasting and Electronic Media, Vol. 32. No. 4 1988. ősz, p. 415-427. Robinson, M. J. (1976). Public Affairs Television and the Growth of Political Malaise: The Case of" The Selling of the Pentagon". The American Political Science Review, 70(2), 409-432. Rosenstone, S. J., - Hansen, J. (1993). Mobilization, Participation, and Democracy in America. New York: Macmillan. Shaw, D. R. (1999). A study of presidential campaign event effects from 1952 to 1992. The Journal of Politics, 61(2), 387-422. Sides, J., Lipsitz, K., & Grossmann, M. (2010). Do Voters Perceive Negative Campaigns as Informative Campaigns? American Politics Research, 38(3), 502 530. https://doi.org/10.1177/1532673x09336832 Sigelman, Lee Kugler, Mark (2003): Why is Research on the Effects of Negative Campaigning so Inclusive? Understanding Citizens Perception of Negativity. In: The Journal of Politics, Vol. 65, No. 1. 2003. Febr., p. 142-160. Stevens, Daniel Sulivan, John Allen, Barbara Alger, Dean (2008): What s Good for the Goose is Bad for the Gander: Negative Political Advertising, Partisanship and Turnout. In: Journal of Politics, Vol. 70. No. 2. 2008. April, p. 527-541. Stichting Kiezersonderzoek Nederland SKON; Centraal Bureau voor de Statistiek - CBS; Kolk, Dr. H. van der (Universiteit Twente); Tillie, Prof.dr. J.N. (Universiteit van Amsterdam); Erkel, P. van; Velden, M. van der (Universiteit van Amsterdam); Damstra, A. (Universiteit 18

van Amsterdam) (2012): Dutch Parliamentary Election Study 2012 - DPES 2012. DANS. https://doi.org/10.17026/dans-x5h-akds Swanson, D. L., & Mancini, P. (1996). Politics, media, and modern democracy: An international study of innovations in electoral campaigning and their consequences. Greenwood Publishing Group. Van der Eijk, C., Franklin, M., & Marsh, M. (1996). What voters teach us about Europe-wide elections: What Europe-wide elections teach us about voters. Electoral Studies, 15(2), 149-166. Wattenberg, Martin P. Brians, Craig Leonard (1999): Negative Campaign Advertising: Demobilizer or Mobilizer? In: American Political Science Review, Vol. 93. No. 4. 1999. Dec., p. 891-899. Whiteley, P.F. and Sanders, D., British Election Study, 2010: Face-to-Face Survey [computer file]. Colchester, Essex: UK Data Archive [distributor], August 2014. Zaller, John (1992): Nature of Mass Opinion. 19