The Effects of Populist Rhetoric: The Dynamics of Attention-Seeking from Inside and Outside of Government

Transcription

1 The Effects of Populist Rhetoric: The Dynamics of Attention-Seeking from Inside and Outside of Government Paper to be presented at the ECPR Joint Session in Münster (D), 2010 Mirjam Ryser University of Bern and the University of North Carolina, Chapel Hill DO NOT CITE WITHOUT THE PERMISSION OF THE AUTHOR Abstract What are the effects of populist communication within a setting where populist parties hold seats in the executive branch? Little is known about the effects of populist communication and the strategies of populist parties in public office. In this paper, we approach the phenomenon of populist communication from a political psychological perspective and argue that appeals to the emotions of fear and enthusiasm inherent in populist rhetoric boost attention. Data collected in an experiment within a 2009 campaign on the highly controversial initiatives against the construction of minarets and against arms exports in Switzerland allow us to test for the effectiveness of left wing outsider and right wing insider populist communication. We measure the effects of manipulated populist scripts on a) subjects attention, b) information seeking, and c) persuasion with regard to candidate choice and issue voting. We find that a) outsider left wing populism tends to increase attention, but older people tend to be more attentive to such emotional appeals; b) traditional right wing populism decreases subsequent interest in the initiatives, especially in the older population; c) younger people are prone to vote for right wing populist candidates and in line with such messages. Authoritarian predispositions in the youth have an effect only on their candidate choice. Our results suggest that future research on political communication might benefit from considering different personality characteristics in experimental settings. Thanks to all of the participants of the online survey. My special thanks to Skyler Cranmer, Marco Steenbergen, Todd Hartmann, David Blatter. 1

2 Introduction Political communication is a rising subfield within the study of populism. Scholars have examined diverse topics, such as populist party programmes (e.g. De Raadt et al., 2004), the discourse of a given party (e.g Armony & Armony, 2005), and the media as a central forum for populists (e.g. Calhoun, 1988; Mazzoleni, 2003). More recently, research on continental European populism has expanded the scope of analysis from right wing parties to the entire party spectrum (e.g. Jagers & Walgrave, 2007; Ryser, 2010). These contributions are important since they teach us about the logic underlying populist ideology (Hakwins, 2009), the diverse populist strategies different parties apply (Jagers & Walgrave, 2007), and about the contextual dependency of populism (Ryser, 2010). Despite this emerging literature addressing the agenda-setting side of populist communication, little is known about the effects of such rhetoric: we do not know whether, how, and under which circumstances populist rhetoric is more likely to persuade individuals, or influence their attitudes. The contribution of the present study is to address the underdeveloped field of the effects of populist rhetoric. Building on the emotions and political communication literature from a functionalist perspective (e.g. Lazarus, 2001), we argue that populist rhetoric usually appeals to different emotions, namely anxiety and enthusiasm. This argument is based on a growing, although US-centric, literature on political communication, which demonstrates the importance of emotions in political communication (Brader, 2005; 2006; Marcus et al., 2005). This study is particularly salient because it contributes to the under-researched setting, in which a populist party is an established member in the executive branch: we analyze campaigns of two 2009 referenda against the construction of minarets and arms exports in Switzerland where a so-called populist party, the Swiss People s Party, holds seats in the executive. Particularly interesting is the fact that the referendum against the construction of minarets was mainly initiated by members of the right wing Swiss People s Party, while the referendum on arms exports was mainly launched by opponents in the radical left. Thus, these cases provide the unique opportunity to assess and compare the effects of populist communication of a populist party in public office to outsider populism and to contrast rightist vs. leftist populism. This study is the first of its kind in so far as it conducts an experiment with populist speeches as a treatment. Experiments are well established methods that provide causal inferences and are especially well suited to the study of emotions, since they capture short-term responses that often escape awareness (Brader, 2006: 76) and can hardly be measured by standard survey techniques. We begin with a description of the two referenda in order to provide some background information and show that they fit the purpose of this study. Second, we discuss the existing literature regarding the question of how individuals might be influenced by populist rhetoric and derive expectations. Specifically, this section describes the mobilization potential of populism, defines populism and its relation 2

3 to emotional appeals and argues that emotions influence information processing and persuasion. Third, we explicate the research design and methods. We then discuss the results, and conclude with an assessment of the results of the study. Popular initiatives in Switzerland: against the construction of minarets and against arms exports The popular initiatives 1 in Switzerland against the construction of minarets and a ban on arms exports from Switzerland were voted on November 29th, The minaret initiative contained an amendment banning the construction of new minarets and was accompanied by a far-reaching, emotional campaign which received wide spread attention both domestically and internationally. For example, the right wing launched a controversial poster campaign showing a scarfed woman in front of a Swiss flag from which minarets rise like missiles. Campaigning was heavily marked by cultural arguments, pitting genuine Swiss against foreign islamic values. However, the acceptance of this initiative by 57.5 percent of the voters ran against the tendency of Swiss initiatives not to receive the required majority of both voters and cantons to pass into law. The result also ran against most of the expectations and predictions of Swiss pundits (e.g. gfs.bern, Longchamp et al., 2009). The Federal Department of Justice and Police explained the acceptance of the vote with marked fears and uncertainties among the population (cp. Federal Department of Justice and Police, 2009). Retrospectively, opinions of commentators on the referenda continued to diverge. Some commentators debated the role of younger women who were assumed to be mobilized by populist appeals to a perceived threat of Islam towards women. Other studies pointed out that people with little education and partisans of middle parties were among the decisive voter demographics in support of the initiative (Hirter & Vatter, 2010). Still others suggested that high levels of support for the ban among conservative ideologues, the lower class, and the rural population contributed decisively to the outcome (Moser, 2009: 3). In contrast, the debate about the proposed arms exports ban received less public attention, although the pro and contra side launched substantial campaigns. Arguments of the proponents centered around the humanitarian values of Switzerland and the danger of the arms industry, while opponents emphasized the danger of job losses. The failure of the initiative with only 31.8 percent of yes votes was largely attributed to the lack of mobilization beyond leftist voters (Hirter & Vatter, 1 The political system of Switzerland differentiates between the direct-democratic instruments referendum and initiative. The former can be used to abort certain legislation by the government. The latter refers to proposed legislation initiated by a certain number of signatures from Swiss citizens. 3

4 2010). These cases present a splendid opportunity to examine the effects of populist rhetoric. First, as already mentioned, the initiative against the construction of minarets passed in a landslide, while the initiative banning arms exports failed. Retrospectively, we may attribute this result to success and failure in campaigning respectively. However, we do not know much about the specific mechanisms and effects of political communication and on what point particular messages succeed or fail. Such subtile mechanisms are difficult to detect with self reports in survey research. However experimental manipulation offers an avenue to measure this particularities. In this paper, we will test the effects and contributions of populist rhetoric, by testing individual party messages. Second, the two initiatives allow us to assess the efficacy of left vs. right wing populism. The popular vote on arms exports was initiated by the leftist outsider alliance Bündnis gegen Kriegsmaterial-Exporte consisting of members of different parties from the extreme left wing (such as the Alternative Liste (AL) and the young socialists (JUSO)) and organizations, such as the GsoA (Group for a Switzerland without an Army) which can be considered as extreme left populist. This provides an opportunity to study left wing populism, which is generally stronger among outsider parties (cp. Mudde, 2004: 549). In contrast, the minaret initiative was launched by a committee, consisting mainly of members of the populist incumbent Swiss People s Party. 2 Furthermore, this initiative seems to be in line with the current spirit of populist right wing parties, in so far as muslims have become one of their special enemies (Mudde, 2007: 78ff). Third, these cases afford us the opportunity to study the efficacy of populist discourse by an incumbent party (Swiss People s Party) and an outsider coalition (Bündnis gegen Kriegsmaterial-Exporte), which has not been addressed by research so far. While populism has been considered a short-term phenomenon, the incorporation of populist parties in executive branches such as the Lega dei Ticinesi (LDT), and the Swiss People s Party (SVP) in Switzerland, or the Forza Italia (FI) in Italy, presents a new challenge for research. The few theories on populist communication examining their relationship with the media, have been mostly based on the assumption that the populist party will fail (e.g. Stewart et. al., 2003). Only a few studies consider the setting of populist incumbent parties (e.g. Albertazzi, 2009; Ivarsflaten, 2008). Fewer studies contribute to the question of whether and how other parties react to populist incumbent parties. Mudde (2004: 563) argues that the established parties will pursue a strategy of simultaneous inclusion and exclusion. They try to exclude the populist actors from incumbency, but will include populist issues and communication in order to fight off the challenge. Indeed, Ryser (2010) finds in her empirical analysis of political speeches that populist rhetoric is not only a prerogative of the right wing Swiss People s Party, but that christian-democrats are also likely to employ this 2 14 out of 16 committee s members are in the Swiss People s Party. The remaining 2 are in the christian-conservative party Eidgenössisch-Demokratische Union (EDU). 4

5 rhetorical strategy. So far, results have been mixed, but seem to indicate that governmental incorporation of populist parties does not inevitably lead to their moderation as proposed by some (e.g. Mény & Surel, 2002). In particular, in the case of the Swiss People s Party a strategy of opposition within government is pursued (Albertazzi, 2009). Thus, questions about whether and how parties adapt their communication strategies once in government and how other partisan and outsider actors react to this challenge have remained widely unanswered. Thus, our suggested approach to the study of political communication is to broaden the scope of analysis from the populist actor itself to the interdependency of actors acting, reacting, and inter-acting with one another. Individuals reactions to populist messages Mobilization and populist issues The mobilization potential of populism is suggested by its very definition and can be easily seen in populist campaign efforts. The political communication literature generally treats populism as a political communication strategy, thus allowing for the possibility that populist rhetoric might be employed by different parties and cover different issues (e.g. Jagers & Walgravae, 2007). While definitions of populism vary (cp. Weyland, 2001), inclusion in an in-group and exclusion of an out group are generally considered intrinsic components of populism. An in-group ( the people ) is commonly portrayed as homogenous and under attack by one or several out-groups ( the others ) (e.g. Albertazzi and McDonnell, 2008a: 3; Jagers and Walgrave, 2007: 322). This definition does not restrict the use of populist rhetoric to right wing parties; it allows it to be used across the ideological spectrum (e.g. Taggart, 2000). For example, left wing populists might conceive of the working-class ( the people ) as being threatened by economic elites and right wing populists might portray native born citizens ( the people ) as being invaded by immigrants and criminals. The political efficacy of appeals to cultural traits is supported not only by work on populism, but also by a broad literature on public opinion, European Integration, and election studies (e.g. Bornschier, 2005; Kriesi, 2005; Kriesi et al., 2005; Statham & Koopmans, 2009). Kriesi (2005) evaluates the defense of the national sovereignty as a response to globalization and argues that mobilization based on cultural traits is much easier than economics. This is so because national identity constitutes the smallest common denominator, whereas globalization s losers are heterogenous and hard to unify. Thus, it is not surprising that immigration issues generally top the agenda of right wing populist parties. However, Statham and Koopmans (2009: 461) take issue with the tendency of Kriesi (2005) to reduce Euroscepticism to cultural nationalism by pointing out that the idea of a homogenous, national civic community with common values corresponds more to a civic-nationalism than to xenophobia. This national civic community is more 5

6 likely to appeal to a cross-class population and not to a specific voter segment. This refinement of Kriesi (2005) does not diminish the importance of cultural cues in populist communication and according mobilization, but rather sustains it. However, recent research in political psychology also demonstrates that not every group cue mobilizes citizens equally (Brader et al., 2008). The populist Swiss People s Party successfully mobilizes voters using debates on asylum and immigration issues which attract a considerable amount of public attention (Albertazzi, 2009). For example, while the the Swiss People s Party s leader Christian Blocher had a seat in the executive in 2006, the party proposed legislation tightening asylum and immigration laws, and a leftist attempt at a referenda to overturn the legislation failed (cp. Ryser, 2010). After the dismissal of Christoph Blocher from the Federal Council in 2007, the party launched several campaigns which attracted attention beyond the borders of Switzerland. The most visible of these campaigns were a referendum to accelerate the expulsion of foreign criminals (this legislation is currently under parliamentary processing) and the now infamous (and successful) initiative banning the construction of minarets. By contrast, leftist organizations have been less successfully in mobilizing voters so far. The Bündnis gegen Kriegsmaterial-Exporte is an issue organization which launched the initiative against arms exports. It is composed primarily of individuals with partisan ties to the extreme-left wing and/or ties to other ecological or humanitarian organizations. The principal mobilization force within this organization, the GSoA (Group for a Switzerland without an Army), mobilizes voters around the issue of army disbandment; the GSoA has often proposed legislation calling for the compete dissolution of the Swiss military. The GSoA has been substantially less successful than the Swiss People s Party in terms of initiative turnouts. That said, they were able to collect enough signatures to launch several, ultimately unsuccessful initiatives to abolish the Swiss army. The fundamental question raised by this discussion is Why do individuals react to populist rhetoric? Based on the arguments that cultural appeals are inextricable from populist communication and themes around immigration and integration seem to mobilize voters successfully, we expect that cultural issues and appeals related to the minaret initiative affect individuals to a stronger degree than the humanitarian and economic issues related to the arms initiative. We also hypothesize that party communication by the incumbent populist party affects voters to a higher degree than outsider populist communication by the left wing. In the following, we argue that emotional appeals defined by Brader (2006: 68) as any communication that is intended to elicit an emotional response from some or all who receive it and taken here to be an essential feature of populist communication - attracts attention and does much to sway voters. 6

7 Populist rhetoric and affective persuasion Emotional appeals are an essential part of populist rhetoric. While scholars dating back to Aristotle emphasized that emotional reactions to rhetoric are crucial to persuade people (cp. Neblo, 2007: 39), many contemporary scholars argue that emotions - especially politics of fear - are an essential part of populist communication. Such populist communication seeks to elicit enthusiasm in order to define the in-group and mobilize voters (Canovan, 1999; Meyer, 2006: 81; Mudde, 2007: 63ff; Albertazzi & McDonnell, 2008a: 2). However, studies elucidating the function of emotions in populism are conspicuously absent from the literature (Ryser, 2010). We argue here that populist rhetoric appeals simultaneously to emotions of enthusiasm and threat. As defined above, populism fosters inclusion by creating an in-group; this manifests in communication by appeals to emotions of enthusiasm. Since the in-group is portrayed as under attack by some sort of enemy, populist rhetoric frequently uses threat devices. While it is certainly the intent of populist parties to persuade voters, the efficacy of such rhetoric is unclear. Fortunately, social and political psychology provides a voluminous literature on how emotional appeals stimulate attention and contribute to persuasion. As such, we draw heavily on this literature in order develop a theory of emotive appeals in populist rhetoric. In so doing, we must consider the effects of affective appeals on information processing (attention) and persuasion. Information processing and persuasion A well-established literature emphasizes the crucial role of emotions in information processing as well as the consequential persuasion and mobilization of citizens (Hutchings et al., 2006: 166). In the introductory chapter of their seminal political psychological work The Affect Effect, Neuman and colleagues (2007) compile a non-exhaustive list of 23 named theories, concepts and models addressing the relationship between cognition and emotion. While it is impossible to give an adequate review of those works in a single paper, many of these approaches draw heavily, or are explicitly based, on a dual-process approach. This approach posits that individuals process information either in an effortful and motivated way in which they evaluate messages based on relevant content or that they process information in a non-systematic, low-effort way and pay little attention to the contents of communications. In the latter situation, so-called peripheral cues, such as the rhetorical style or attractiveness of the message sender, become increasingly important (for a summary see Wirth & Schramm, 2005). But what is the role of emotions when individuals are confronted with politics? The seminal theory of Affective Intelligence by Marcus and colleagues (2000) distinguishes between two distinct emotional systems in information processing: the dispositional and the surveillance system. The theory posits that the dispositional system encourages issue approach and operates on the positive dimension of enthusiasm, relying on routine information and fostering interest in a given cam- 7

8 paign. In contrast, the surveillance system is guided by issue avoidance and the negative dimension of anxiety, directing citizens away from longstanding beliefs and simultaneously enhancing their motivation to learn. In such situations individuals tend to yield to the force of arguments and persuasion may be facilitated (cp. Huddy et al., 2005). However, the theory of Affective Intelligence tells us little about the effects of campaign communication and emotional reactions (Hutchings et al., 2006: 167). Only few studies experimentally examine the effect of campaigns and according individual reactions. The studies which do, examine only candidate voting by manipulating anxiety and enthusiasm separately (e.g. Brader, 2005; 2006). Specifically, Brader (2005) finds that enthusiasm fosters interest in campaigns and stimulates voting, but has no effect on the recall of campaign related messages or persuasion. The author finds a reverse pattern for fear appeals. However, it is less clear how mixed emotions typical in populist rhetoric influence information processing and lead to persuasion. Political psychology has long conceived of emotions as a bipolar structure where individuals are assumed to move away from one pole when approaching the other. Only recently have researchers recognized the individuals potential for simultaneously experiencing positive and negative feelings toward the same object (Neblo, 2007; Oceja & Carrera, 2009; Steenbergen & Ellis, 2006). Nonetheless, the question of what the behavioral consequences of mixed emotions are, has not been answered. A functionalist perspective, such as that maintained by Lazarus (2001), provides us with some hints: according to his argument emotions are interdependent, and thus, their labels positive and negative are often misleading. A positive emotion, such as hope, can only occur in the presence of the negative experience of threat (Lazarus, 2001). For example, expecting an upcoming knee surgery, we feel threatened, but simultaneously hopeful that the surgery will go smoothly. Applying this idea to a concrete populist message means that the appeal to negative emotions should catch our attention and allow the party to make its point persuasively (Kern, 1989, cited in Brader, 2006: 155 refers to this strategy as get em sick, get em well ). Thus, conceiving of populism as an attention-booster, we hypothesize that populist messages generally elicit a higher degree of attention and thus, foster issue approach. Besides attention, information seeking is another important aspect of individual information processing and is an antecedent of persuasion. So far, we have argued that populist messages will generally attract attention. However, from a functionalist perspective, attention is directed towards the claims of the populist message; this renders the consultation of other information sources redundant. This argument is supported by literature on populism and authoritarianism, which emphasize the mobilization potential of populist parties (e.g. Canovan, 1999; Di Tella, 1997). The vast literature on authoritarianism suggests that in threatening situations, certain individuals are more likely to search for security from as powerful perceived others in order to reduce fear and insecurity (e.g. Feldman & 8

9 Stenner, 1997; Oesterreich, 2005). 3 Although such reactions are associated with an authoritarian personality structure, such predispositions are assumed to be common in regions where populist parties are popular (Di Tella, 1997). Therefore, we have good reason to suspect that populist rhetoric fosters a tendency to rely on its sender, especially for individuals with higher authoritarian predispositions; thus rendering the need for subsequent information gathering unnecessary. Consequentially, the populist sender will attract more voters and is generally more persuasive than a non-populist sender. While most of the experimental studies in political communication and psychology which examine information processing have neglected to take age differences into account, 4 studies of populism acknowledge that its unconventional style of campaigning also tends to draw in younger, less politically aware, and traditionally apathetic voters (Heinisch, 2003: 124). The importance of age related patterns is supported by studies of emotion by consumer research. Recent findings show that the motivation of elderly individuals approaching information and the way they process information differs considerably. Older people tend to focus more on affective than factual information and their attentive ability, memory, and capacity to process information is generally decreased (Drolet et al., 2007: 212; Isaacowitz et al., 2000; Williams & Drolet, 2005: 344). Scholars such as Hashtroudi et al. (1994: 160) and Isaacowitz et al. (2000) even link the combination of inefficient information processing and the enhanced importance of values and experiences in older people to a stronger incorporation of irrelevant (emotional) information in working memory. In line with that finding, Williams and Drolet (2005) show that older consumers recall more information from affective ads. It has also been acknowledged that the propensity of individuals to accept mixed emotions differ with character traits such as culture and age (Basseches, 2005). Specifically, older age groups tend to synthesize and acknowledge conflicting emotions to a higher degree than younger adults. Why this effect occurs remains an open question (Williams & Aaker, 2002), but it has been suggested that individuals maturation seems to partially explain this pattern (Williams & Drolet, 2005: 344). As such, we expect age-realted differences in our results. 3 Similarly, some authors argue that populist right wing parties are largely supported by insecure individuals (Mudde, 2007: 223). However, a clear separation from authoritarian predispositions seems to be difficult due to endogeneity. 4 Age differences have been examined in related subfields. Scholars, such as Visser and Krosnick (1998) and Krosnick and Alwin (1989) found differences in attitude strength according to life stages; though they point out the complexity and object dependency of the phenomenon. However, the finding that early adulthood is a period of relative plasticity is consistent across various studies (Visser and Krosnick, 1998: 1407). Also Sears (1986: 523) argues in his characterization of college students that their traits vary systematically from other age groups and that those individuals possess stronger information processing skills. 9

10 Methods Overview To evaluate our hypotheses, we conducted an experiment with Swiss citizens shortly before the votes on minarets and arms exports in November Subjects were exposed to a treatment condition after having completed a pretest which included measures such as political attitudes, authoritarian predispositions, some demographics and current mood among others. An additional battery of questions was presented to the subjects after the experimental exposure. Sample Subjects were recruited through several venues, such as s, facebook, student s listserves, corporate listserves, snowball sampling, and an announcement on a Swiss version of Craig s list ( The subjects were given the opportunity to participate in a lottery, competing for cash prices and movie tickets. A total of 671 participants (Swiss citizens over 18) completed the online survey experiment. Of these subjects, 90 were dropped because they left the survey before receiving the treatment, leaving us with a final sample of 581 subjects. While an experimental setup provides us with high internal validity due to the random assignment of a treatment, the question of external validity has to be considered. While the sample is similar to the Swiss population of 2009 on the key demographic dimension of gender (47.8 percent males and 52.2 percent females vs percent males and 50.8 percent females in the Swiss population), there are notable differences in education (49.3 percent of the sample had at least a bachelor s degree compared to 24.9 percent of the Swiss population) and age (sample mean of vs. a population mean of 41.2 years at the end of 2008) (Bundesamt für Statistik, 2009). Experimental manipulation We use populist speech as a treatment. As argued above, the combination of conflicting emotions are expected to be an effective means of communication in terms of boosting attention and persuasion. Participants completing the survey were randomly assigned to one of eight treatment conditions; an additional group served as a control group. The eight treatment conditions result from manipulating each pro and contra side of the two initiatives with a populist tone (2 x 2 x 2 design). Manipulating the contra side is necessary due to possible negativitiy effects. The populist scripts were designed according the populist technique of fostering fear and a subsequent creation of a sense of unity among the perceived in-group. The treatments were loosely inspired by a viral video named Heaven or Hell produced by the Swiss People s Party before the federal elections in This 10

11 video reflects the populist s strategy: footage of youth and immigrants rioting is followed by images of an idealized Switzerland with showing beautiful landscapes, goats, and festivities. The treatments scripts were edited so that they contained the same number of arguments. We manipulated the style of argumentation and populist cues (for an example, see Table 1; for a full translation of the scripts, see the Appendix). The arguments were selected from each pro and contra camp of the two popular votes and were adapted to fit the populist style. Each script contains two arguments, one referring to threats to the Swiss constitution, the other to physical or financial threats. The arguments were derived from the websites of both of the initiative committees and their opponents. Manipulation check TABLE 1 ABOUT HERE Following the advice of Brader (2005), we included a battery of questions towards the end of the survey in order to check whether and to what degree the treatment elicited emotions. Assessing the arousal of emotions is important, as the literature has increasingly emphasized that distinct emotions may result in different behavioral reactions (e.g. Steenbergen & Ellis, 2005). 5 Usually, self-reports are used because one cannot infer the type and intensity of emotion experienced on the basis of behavioral or physiological measures (Wirth & Schramm, 2005: 12). However, there are also some warnings warranted. First, much of our experiences are unconscious and a reliable and veridical assessment is difficult (Lodge et al., 2006: 28). Second, emotions are not recalled equally. According to a review by Wirth & Schemer (2005: 21), there is evidence to suggest that positive emotions are recalled better than negative ones. Individuals also tend to underestimate the intensity of mixed emotions (Aaker et al., 2008) and age is likely to influence their reports (cp. Isaacowitz, 2000). TABLES 2a, 2b, 3 ABOUT HERE Tables 2a and 2b present the entire sample s means for different emotions by populist scripts as well as the results from an ANOVA testing all populist scripts against their placebos. The negatively loaded emotions (i.e. anxiety, threat, worry, and anger) are experienced to a higher degree by individuals exposed to a populist script. Of the three positively loaded emotions (i.e. enhancement, enthusiasm, and hope), populist scripts elicit on average significantly less hope. Differences according to age groups and issues, are reported in the Appendix. 5 We conducted a pre-test in which scripts were randomly distributed among a small student sample (N = 13). We found that 9 out of 13 populist scripts elicited a self-reported emotional reaction compared to only 5 out of 13 non-populist messages (M pfeeling = 1; M npfeeling = 0.5; t = , p =.0033). 11

12 Building on attribution theory (Fletcher and Ross, 1985), we also included a question asking which objects specific emotions are directed at; the idea here is that different stimuli might generate different emotions (e.g. Marcus et al., 2005: 170). Table 3 shows that party communication can elicit emotions towards objects which the messages do not allude to. Specifically, emotions are often directed towards the sender of the message itself. The fact, that emotional messages draw attention to personal characteristics instead of more substantial political issues is not a new finding (cp. Marcus et al., 2005: 19). Significant differences between the populist scripts and their placebos are only found for worry (p <.05). Populist scripts seem to elicit worry against the message sender to a higher degree than their placebos, while the placebos tend to elicit emotions towards contents of the scripts (i.e. arms, muslims, initiatives). In sum, the self-reported emotional arousals caused by populist messages seem to rather tap into what are often labeled negative emotions. However, they do not necessarily need to follow the intended direction of the message, since emotions - especially worry - are sometimes directed towards the party itself. Results This paper tests whether emotional appeals by populist parties inside and outside of the executive branch (which we call insider and outsider populism respectively) affect behavioral reactions in the population differently. Three aspects of behavioral reactions towards populist speeches are considered. The first, the extent to which subjects recall the content of the presented messages measures whether populist scripts foster citizens attention. Second, we measure information seeking in order to assess whether or not subjects were motivated to search for additional information. Both of these measures are important antecedents for persuasion. Third, this study assesses the persuasiveness of populist messages. Attention to populist messages Do subjects pay more or less attention to populist messages? If yes, are there any differences in effects between insider and outsider populist discourse? As argued above, populist messages do appeal to emotions, and since emotional events are recalled better compared to neutral ones (Klauer, 2004, cited in Wirth & Schramm, 2005: 20), we expected to find that recall is enhanced by populist message contents. In order to measure recall, we counted the recall of arguments (keywords) and split them into quartiles ranging from 1 to 4. Comparing the populist to all nonpopulist scripts, cross-tabulations do not reveal statistically significant differences at first glance. Although recent experimental studies in political communication have only begun to focus analysis on different age groups, we provided theoretical motivation above for such focus (e.g. Drolet et al., 2007). As such, we split 12

13 the sample according to whether the subjects were 30 years old or younger. 6 A preliminary calculation shows that younger individuals recall information from the scripts generally better on average than subjects in the older sample (the average of the younger subset s five point scale interest scale is 2.54; the one from the older subset is 2.01; F = (1, 452) = 21.90; p < ). 7 This finding is in line with Sears (1986) argument that younger individuals possess stronger information processing skill. Considering the idea of age related patterns in the context of populist communication reveals differences between populist and non-populist messages. As shown in Table 4, older individuals process populist scripts more carefully than non-populist scripts (F = (1, 160) = 2.741; p < 0.1). 8 As can also be seen, younger individuals process populist scripts and their placebos in similar fashions. Consider now the question of whether insider and outsider populist communication affects individuals differently. Table 5 shows that scripts tailored towards convincing individuals to vote in favor of the initiative against the export of arms, fosters vigilance to a significantly higher degree than their placebos for the entire sample (mean of 2 vs. 2.4; F = (1, 137) = 4.79; p < 0.1). In contrast, populist scripts in favor of the initiative against the construction of minarets do not lead to different attentional reactions. 9 The results indicate that populist communication is not equally effective. Instead, it seems to be the case that issues and individual characteristics affect the efficacy of populist rhetoric. We find that emotional cues foster attention, especially in older individuals, as predicted by findings from consumer research (Williams & Drolet, 2005). Additionally, populist outsider (left wing) messages tend to generally catch the public s attention to a higher degree. A possible explanation might be the difference in the salience of the actual campaigns: the campaign on the minaret initiative received considerably more media attention than the one banning arms exports (cp. Hirter & Vatter, 2010). 6 This subsample also differs in terms of an enhanced political interest: on a 4 point scale (ranging from 0 to 3) the average of the older sample amounts to 2.6, compared to the younger sample s average of 2.2 (F = (1, 520) = 34.50; p < ). The correlation of enhanced political interest with increasing age is a well-established finding in literature (Hirter & Vatter, 2010: 3). 7 Although the data do not allow us to measure individuals exposure time to the treatment, the older subset spent significantly more time completing the survey (31 minutes) than the younger subset (27 minutes) (F = (1, 520) = 4.00; p = 0.06). 8 Considering the different issues, with the exception of the initiative against the construction of minarets, where we do not find statistically significant differences, the older age-group is more attentive to populist messages. 9 Considering the two age groups, the described pattern holds. However, we find, interestingly, that populist communication can also lead to a reduction in individual s attention. Populist scripts opposed to the initiative against the export of arms lead to a reduced attention in individuals, equal or below 30 years (means of 2.7 vs. 2.2 respectively; F = (1, 61) = 4.00; p < 0.1). 13

14 Information seeking TABLES 4, 5 ABOUT HERE A second individual reaction towards political messages is the subsequent arousal of interest and a desire to search for additional information. The extent to which an individual decides to search for further information on a given subject generally depends on emotional reactions (e.g. Brader, 2005; Marcus et al., 2005). We included a measure at the end of the experiment which asked individuals whether they wished to receive an information packet concerning the upcoming vote. As argued above, we expect that populist messages foster a passive attitude in the need to search for additional information due to the reliance on the populist sender because the sender portrays himself as a provider of security (e.g. Oesterreich, 2005). Indeed, the results displayed in Table 6 support this claim: populist messages generally decrease interest. While roughly 60 percent of the subjects who were exposed to a populist manipulation indicated a desire for further information, 68 percent of the placebo group did so (note that exactly 50 percent of the control group (N = 62) reported interest in additional information). That much said, considering the four groups exposed to different types of populist manipulations leads to more nuanced findings: the manipulated scripts in favor of the initiative against the construction of minarets are the only scripts which elicit statistically significant differences in interest. While 53 percent of the individuals receiving a populist treatment indicated a desire for more information, 72 percent of the corresponding placebo group made such an indication (Table 7). In line with previous research (e.g. Sears, 1986), younger individuals are generally more inclined to search for additional information than older individuals: 68 percent of the younger individuals requested further information while only 52 percent of the older sample did so (Pearson Chi-Square (1, 521) = 12.4; p = 0.001). The results suggest again that populist messages have differential effects on younger and older populations. While there was barely an impact on the younger subset, 10 we find similar (significant) differences within the older subset as we do for the entire sample on the minaret initiative. In sum, populist messages in favor of the anti-minaret initiative foster passivity towards further information; particularly in older individuals. The literature on authoritarianism suggests that individuals tend to rely solely on the populist sender (e.g. Feldman & Stenner, 1997; Oesterreich, 2005). However, self-reported emotional attributions suggest that worry is associated with populist messages senders (cp. Table 3). TABLES 6, 7 ABOUT HERE 10 The closest effect to a conventional threshold is found for the initiative against the initiative against arms export (81 percent of 37 individuals exposed to a non-populist script requested further information, while 64 percent of 25 individuals exposed to a populist message made such an indication; Pearson Chi-Square (1, 62) = 2.3; p = 0.13). 14

15 Persuasion We measured two aspects of persuasion: whether subjects consider the message sender as worthy of support by vote and whether they would vote in line with his message. TABLES 8a, 8b, 9 ABOUT HERE The first question we should address is whether participants would vote for the party who they assumed had posted the message. Subjects did not know which party composed the message, but were asked after the presentation of the scripts, whether or not they would vote it. Pitting the populist against the non-populist scripts does not reveal a statistically significant pattern (not reported). 11 However, splitting the sample by age group and issue leads to further insights: subjects 30 or under show significant but varying patterns in differences between the populist scripts and placebos for both initiatives (see Table 8a and 8b, note that based on the limited data available, the results have to be interpreted cautiously. Note also that the high percentage of votes opposed to the initiative against the construction of minarets and in favor of the initiative against arms exports is related to the voting behavior of the over-educated sample). The results shown in Table 8a suggest that populist scripts for the anti-minaret initiative lead to more support for the initiative compared to placebos. The opposite pattern is revealed in Table 8b: the populist versions of the scripts for the arms export ban initiative foster less party support than their placebos. While it is not surprising that younger individuals tend to be prone to persuasion, it remains an open question how we might explain this reverse effect on the different issues. According to theory, subjects with authoritarian personalities seem to have stronger reactions to populism (e.g. Mudde, 2007: 219ff). There is much debate about the measurement of authoritarianism and its relation to different ideologies. We included an established measure of child-rearing values (cp. Feldman & Stenner, 2003), as we believe it to capture authoritarian predispositions unrelated to expressed ideological preferences. Table 9 displays estimates of a logistic regression for the younger subset. This regression includes, some controls and an interaction term with a measure for authoritarian predisposition and minaret-populism. The interaction term reveals a significant effect, while the same application to the initiative against arms export does not yield a significant pattern. This suggests that younger individuals with higher authoritarian predispositions confronted with populist cultural scripts are more likely to vote for the sender party. However, the same mechanism does not apply to left wing outsider populism (not reported), where subjects evaluate a nonpopulist communication style as more convincing for party support independent of authoritarian predispositions. 11 Also analyzing the populist messages by the different voting issues does not reveal any statistical significant differences. 15

16 The second question we address is whether individuals vote in line with the presented message contents. Combining all populist scripts and pitting them against their placebos does not reveal significant differences. Again, the division by age groups matters: Table 10a suggests that younger people are more likely to vote in line with the populist messages compared to their placebos. Conversely, Table 10b shows that this pattern does not hold for the older age group. This result is in line with theory, which emphasizes the plasticity of younger individuals (e.g. Sears, 1986). Next, we examine the different effects of the initiatives in the entire sample. While the scripts for the initiative against arms exports do not elicit any statistically significant differences in reported voting compared to their placebos (not reported), we find significant differences for the scripts in favor of the initiative against the construction of minarets. Table 11 reports that individuals exposed to such populist scripts are generally more likely to vote in favor of the initiative than individuals exposed to a placebo. Applying the same model to the age subsets shows that this pattern only applies to the younger age group (see Table 12a and 12b). We included the same covariates in a logistic regression model as before (see Table 13), although we use a variable measuring whether subjects vote in line with the party message (self report) as the dependent variable. However, the estimate for the interaction term between authoritarian predispositions and exposure to a populist script in favor of the minaret initiative falls far from a conventional statistically significant threshold. The result that authoritarian predispositions interacted with exposure to the populist anti-mineret message contribute positively to party choice but not for issue voting might be an important first finding for further experimentation. 12 In sum, populist messages seem to be more persuasive for younger subjects. We also discovered that the salient right wing populist communication during the campaign in favor of the initiative against the construction of minarets tends to be more persuasive compared to its left wing counterpart, which, however, tends to boost attention. The finding that specific messages with emotional language induce different behavioral reactions within an emotionally loaded campaign is remarkable and highlights the importance thereof: emotional language combined with salient heuristics - in the present case minarets - leads to different behavioral reactions than a placebo script referring solely to a heuristic itself. Furthermore, our findings suggest different relationships and functions of authoritarian predispositions. Authoritarian predispositions seem to affect the younger subset s candidate choice, but not issue voting. These results are suggestive, but replication with different subject samples is necessary to establish a robust result. TABLES 10a, 10b, 11, 12 ABOUT HERE 12 For scripts either opposed to the initiative against the construction of minarets or opposed to the initiative against arms export, we do not find any significant differences, neither for the whole sample nor for any of the two subgroups. 16

17 Conclusion We took a first step in analyzing the effects of outsider and insider populist communication. The simultaneous 2009 votes on the two initiatives against the construction of minarets and against arms exports in Switzerland allowed us to examine such effects. The initiative against the construction of minarets was largely promoted by insider right wing populist communication, while the initiative against arms exports provided us with a case of left wing outsider populism. We collected data from an online experiment shortly before the vote and manipulated scripts of populist speeches based on arguments within a real-world campaign. Highlighting the role of emotions in populist communication and designing our research accordingly, we examined three behavioral reactions towards populist messages. Our results revealed subtile mechanisms of individuals reactions. We found that populist messages do not affect voters equally, but issues and traits (age and authoritarian predispositions) matter. It appears that outsider populist (left wing) messages tend to catch the public s attention to a higher degree than insider populism; a result that we suspect to be affected by different levels of media attention and individual s saturation with populist right wing discourse in the real world. Our results also indicate that populist messages foster attention especially in individuals over 30; a finding which is supported by marketing research (Williams & Drolet, 2005). However, right wing insider populist communication on the initiative against the construction of minarets fosters disinterest, especially in older individuals. We could not conclusively answer, whether this pattern is mediated by the more intense attention older individuals typically dedicated to the message, or the emotional language used by minaret-proponents in the real-world campaign and over-saturation of such discourse, interacted with the enhanced political interest of older individuals, which fosters apathy. While the results of this study suggest a special influence on the older subset with regard to information processing, we cannot say conclusively that heightened attention persuades individuals subsequently. Younger individuals tend to vote more for and in line with the populist sender advocating the initiative against the construction of minarets. A first attempt to explain such reactions with individuals traits, suggests that authoritarian predispositions help to explain candidate s choice, but not issue voting. This shows that populist communication need not necessarily ties to marked authoritarian predispositions in order to affect issue voting. This might be an important clue for explaining differences in candidate and issue voting; it may also be the case that anti-immigrant parties do not have an entrenched constituency (Mudde, 2007: 225). Experimental methods are a fruitful means by which to detect the subtle mechanisms of emotions, inherent in populist rhetoric, which cannot be detect by survey research. Though we have taken an important first step, many questions remain. Our results suggest that future studies in populist communication should focus on traits, such as age and gender. With regard to gender, Mudde s (2007: 113) 17