Graduate School of Communication
To Be or Not To Be Legitimate:
Towards a Better Understanding of the News Media’s Role in Shaping
the Electoral Success of Populist Parties
Master’s Thesis
by
Lena-Maria Böswald
Student ID: 12027332
Master’s Programme Communication Science
Supervision: Dr Lukas Otto
June 28, 2019
Word Count: 8,129
1Abstract
The electoral success of populist parties on both the left and the right end of the political spectrum has prospered over the last decade – and the media has largely been held accountable for it. However, we still know little about the effect news framing can have on citizens’ voting intention, especially on the implicit individual level. The current study adds to the existing literature on media and populism by investigating whether news framing which explicitly blames a populist party – be it from the left or right side of the political spectrum – to be a threat to democracy, therefore depriving it of its legitimacy and questioning its equal status in parliament, can immediately evoke positive or negative discrete emotions amongst the reader and how this emotional response can subsequently affect one’s implicit intention to vote for a populist party. Starting from the premise that this causal relationship is
subordinate to one’s personal political ideology, this two-study experiment explores to which extent political ideology influences the degree of anger and satisfaction evoked by party-critical news framing. Further, the impact of anger elicited by the news coverage on one’s implicit support for a populist party is assessed. The results of the analyses display that political ideology only moderates the relationship between news framing and emotional response at the left-wing populist party level: The more right-leaning one is, the larger is the effect of critical news framing on satisfaction as emotional response. Interestingly, no significant results were found at the right-wing populist party level.
Towards a Better Understanding of the News Media’s Role in Shaping the Electoral Success of Populist Parties
In respect of elections, populist parties and politicians – be it from the left or right side of the political spectrum – have recently been very successful. Over the past decade,
moderate and radical populist parties have entered national and regional parliaments across the globe: In the Netherlands, for instance, Geert Wilder’s anti-immigration Partij voor de
Vrijheid (Freedom Party) became the second largest party in the general elections of 2017 –
and two years later, his Eurosceptic counterpart, the Forum voor Democratie (Forum for Democracy), gained most of the popular votes in the Dutch provincial elections. On the other side of the pond, Donald Trump’s plain but emotionally charged use of populist rhetoric resulted in nothing less than his surprising success in the 2016 US presidential election. In Germany, despite the country’s shadow of the past and continuing process of accounting for its history, the right-wing populist party Alternative für Deutschland (Alternative for Germany) reached its peak in late autumn 2018 by having entered each of the 16 state parliaments of the Republic. On the other side of the political spectrum, Die Linke (The Left) has already been a constant member of the German Bundestag since 2007.
This global populist success story has not gone unnoticed – neither by the press nor media scholars. Ever since the rise of populism in the Western hemisphere, researchers have put emphasis on the media’s relevance in getting populist viewpoints across and in pushing populist ideas forward (e.g. Boomgaarden & Vliegenthart, 2007; Krämer, 2014; Mudde, 2004; Rooduijn, 2014). Yet, the role news media coverage can play in actually fostering the electoral success of populist parties has largely been neglected in empirical research, with Hameleers, Bos and de Vreese (2018) being a rare exception. Studies do, however, outline that news media play a large role in shaping people’s beliefs, opinions and sentiments about an issue of societal relevance (e.g. Walgrave & De Swert, 2004; Lecheler & de Vreese, 2010) as they have a significant effect on how we perceive society as a whole (Luhmann, 1996).
Despite growing interest in the topic among media scholars, but only few examples supporting the assumption that exposure to populist messages through news media does indeed affect populist voting behaviour (e.g. Bos, van der Brug, & de Vreese, 2011), there clearly is a lack of empirical research exploring explicitly how populist actors are represented, challenged or criticised in media coverage and to what degree this particular form of news framing can have an immediate effect on citizens’ attitudes and subsequent voting
behaviours towards populist parties. Hameleers, Bos and de Vreese (2018), for instance, focus on populist rhetoric deliberately induced by the media, therefore stressing the media’s responsiveness to the demand side of the audience, but neglect how populist parties are in fact covered and conceptualised in the news and how this might implicitly affect the outcome at the polls. As a consequence, an open question still remains whether media framing which does not resort to populist rhetoric, but plays with a party cue only, can also play a significant role in shaping the electoral success and strengthening of populist parties.
This research aims to bridge this gap by investigating how news framing which explicitly blames a populist party to be a threat to democracy, therefore depriving the party of legitimacy and questioning its equal status in parliament, can immediately affect people’s implicit intention to vote or not to vote for a populist party. However, the affect based effects in question may not be direct, as another variable might mediate the relationship. Various scholars have already referred to emotionality as an inherent component of populism (e.g. Mazzoleni, Steward, & Horsfield, 2003; Plasser & Ulram, 2003; Meyer, 2006; Bos, van der Brug, & Vreese, 2010; Hameleers, Bos, & de Vreese, 2016). Emotions, especially discrete positive and negative emotions, have proven to not only affect opinion formation but also voting behaviour of citizens in political communication research cross-nationally (e.g. Brader, 2005; Kühne, 2012; Kühne & Schemer, 2013; Lecheler, Schuck, & de Vreese, 2013;
Lecheler, Bos, & Vliegenthart, 2015). As a result, this research is guided by the following research question:
RQ: To what extent does the evaluative framing of news about populist parties affect people’s emotional response and how does this influence subsequent populist voting intentions?
Therefore, the central aim of this study is to focus on how citizens’ voting intentions are affected by news framing once a populist party is represented as a threat to democracy in the news, placing focus on Germany, a country where two (moderately) successful populist parties from both ends of the political spectrum are present – the right-wing populist party AfD and the left-wing populist party Die Linke. Germany appears as an interesting case study not only because of contemporary Germany’s experience with populism, but also due to the country’s ongoing public debate focusing on the media’s role in strengthening the rise of right-wing populism (Gäbler, 2018; Liesching & Hooffacker, 2019).
Theoretical Framework
What Is (Media) Populism?
Populist parties differ in their orientations and political manifestos amongst countries, but all of them, at least to a certain degree, act up to one principle: they pledge to return the “sceptre of democracy” to “the people” (Albertazzi & McDonnell, 2008, p. 2) in order to fulfil the people’s demand for sovereignty. While there are various views on and definitions of populism, the ambiguous concept, in this study, is defined as a “thin-centered ideology” (Rooduijn, de Lange, & van der Brug, 2014, p.1), given its lack of core values (Taggart, 2004) and dependency on a host ideology of greater substance, such as nationalism or socialism, in order to survive and to be effective. Populism centres upon one very plain, but central concept: the blameless in-group, us, versus the culprit out-group(s), them (Mudde, 2004).
Taking previous research into account, populist rhetoric presumes an inherently antagonistic relationship between the homogeneous and virtuous people and the corrupt other (Mudde, 2004; Taggart, 2004; Albertazzi & McDonnell, 2008). This out-group can take the form of either the “corrupt elite” (Mudde, 2004, p. 543) or the “dangerous others”
(Rooduijn, 2014, p.7) who pose a threat to society – or both (Albertazzi & McDonnell, 2014). Against this background, the core idea of populist rhetoric can be considered to be framing attributions of responsibility and blame to everyone who is not part of the innocent in-group (Hameleers, Bos, & de Vreese, 2016; Hameleers, Bos, & de Vreese, 2018) – a concept which can also, even inadvertently, be implemented by the media.
Previous research has emphasised the central role of the media in enhancing the support for populist parties (Mazzoleni, 2008; Krämer, 2014; Roduijn, 2014). Whether this assumption is correct or not is still to be confirmed. However, this much is certain: Populist parties have gradually been on the rise – and the media, according to various scholars, is partially to be held accountable for their success (e.g. Bos, Lefevere, Thijssen, & Sheets, 2014; Hameleers, Bos, & de Vreese, 2018). Regarding the media’s role in shaping the success of populist parties, two explanatory suggestions have been floated that blend seamlessly into each other. Mazzoleni (2008) and Krämer (2014), for instance, refer to the concept of media populism by determining a process by which the news media actively use elements of populist rhetoric, style and their ideas to respond to popular audience tastes and demands. Due to the fact that “mediatised attributions of blame are highly attractive and persuasive” (Hameleers, Bos, & de Vreese, 2017), polarising and emotionally charged rhetoric can not only serve as a character trait of the universal populist lingua, but also be adapted and induced by the media itself (Hameleers, Bos, & de Vreese, 2018).
Secondly, it can be stated that news media do in fact offer fertile ground for populist rhetoric, given the media’s dependency on media logic (Brants & van Praag, 2017) and news values (Harcup & O’Neill, 2017). The tone, style and issues used, pushed and owned by populist parties play perfectly into the hands of news media’s submission to professional news values (Adriaansen & van Praag, 2010) that adhere to market demands. Media populism can hence also be interpreted as “news-value populism” (Bos & Brants, 2014, p. 707), the media’s receptiveness to populist rhetoric – or, its unintentional response to audience demands when giving disproportional attention to populist rhetoric (Bos & Brants, 2014) that takes emotionally charged issues to the centre stage. Therefore, the media can
actively help disseminate populist messages and simultaneously increase the visibility, publicity and legitimacy of populist actors by offering them a communication platform (Sheets, Bos, & Boomgaarden, 2015; Esser, Stepińska, & Hopmann, 2016).
The underlying question in former research has thus ever been whether the media predominantly disseminate populist messages particularly intended by politicians or convey a journalistic interpretation of what has been happening and how this can affect one’s political behaviour. What has never been touched upon, however, is the media’s interpretative coverage of populist parties and its impact on populist vote decisions.
Media Logic and News Framing
Hence, this research goes one step further by taking both options into consideration while drawing on one supplementary element: the media’s representation, or framing, of populist parties in the news. Generally speaking, framing theory suggests that how an issue is characterised and presented to the audience has an effect on the intrinsic choices people make about how to process and situate this information in a broader paradigm of meaning (Entman, 1993). News media organisations select and present information in a particular way when covering events and issues, thereby constructing social and political reality which significantly influences their audiences’ perspectives on the world. The way in which the media decides to portray a certain topic or event can thus effectively shape and form public opinion (de Vreese, 2005). This decisive power in the hands of news media, commonly defined as frame setting by Scheufele (1999), either clearly highlights some aspects of a social issue or neglects others by only selecting aspects that are of relevance for news media coverage (Lecheler, Bos, & Vliegenthart, 2015) – and can thus easily have an effect on public opinion formation and political behaviour. The public excessively relies on how the media present news to them to process information and news framing “can systematically affect how recipients of the news come to understand […] events” (Scheufele, 1999, p. 107).
According to Lecheler and de Vreese (2019), the effect a news frame has on its audience is dependent on how a frame activates long-term knowledge constructs that are
already available and accessible in a person’s mind and which are connected to our
evaluations of political issues and events (Price & Tewksbury, 1997). Hameleers, Bos and de Vreese (2017) even argue that journalists can actively translate abstract societal issues into “meaningful patterns of interpretation” (p. 4) through news framing.
With respect to populism in the media, Walgrave and De Swert (2014) outline that not only through issue coverage, but also through framing can news media shape the electoral success of a populist party. This research hence assumes that the manner of presenting populist parties in the news can shape the public’s perception. Therefore, it is expected that specific news framing of populist parties can eventually affect people’s vote decisions during election periods. As a consequence, news frames about populism will be more thoroughly examined in the following.
News Framing and Populism
News framing has, surprisingly, only been scarcely examined by former studies interested in the relationship between the media and populism (noteworthy exceptions are e.g. Bos, Lefevere, Thuijssen, & Sheets, 2014; Hameleers, Bos and de Vreese, 2018). Hameleers, Bos and de Vreese (2017), for instance, analyse how populist blame attributions are embedded in journalistic reporting styles. Hameleers, Bos and de Vreese (2018) further emphasise the media’s ability to imitate effects of populist rhetoric through news framing by using attributions of blame, which, in turn, positively affect populist voting behaviour.
However, while Hameleers, Bos and de Vreese (2018) focus on the effect of populist rhetoric deliberately induced by the media on populist voting behaviour, therefore adopting Krämer’s (2014) definition of media populism, they disregard how populist parties can be covered and conceptualised in the news and how this might implicitly affect the audience’s voting intention. Further, Bos, Lefevere, Thijssen and Sheets (2014) specifically place focus on party cues embedded in media coverage about right-wing populist parties, but only highlight the interplay of issue ownership by populist parties, its media coverage and
parties as democratically elected, but illegitimate entities which are not to be taken seriously, into consideration. Sheets, Bos and Boomgaarden (2015) also suggest that elitist or anti-immigrant media cues indirectly affect populist voting intention amongst the audience. A prerequisite here is, again, that news media resort to populist rhetoric itself, without actually taking media’s coverage about populist parties into account.
This research aims at filling this gap, given that prior investigations by Vliegenthart, Boomgaarden and van Spanje (2012) have already considered populist party cues to be important for the effects of media coverage on policy support. Further, former research has shown that different journalistic styles influence the audience’s perception of populist parties (Hameleers, Bos, & de Vreese, 2017), with an interpretative and annotating writing style offering the most fertile soil for populist attitudes to strike roots, given its tendency to frame issues in black and white – an inherent character trait of populist communication.
Interestingly, Wettstein, Esser, Schulz, Wirz and Wirth (2018) argue that members of populist parties are evaluated more negatively than other politicians by the media and are actively challenged by journalists, due to the media’s adherence to the social responsibility model of journalism. Assuming that party-critical coverage can evoke different emotional responses amongst readers from different ideological standpoints, as a sole party cue can already comprise adverse or favourable emotional reactions by means of the topic’s intrinsic
emotionality (Bos, van der Brug, & Vreese, 2010; Hameleers, Bos, & de Vreese, 2016; Wirz et al., 2018), this study will focus on evaluatively framed, party-critical coverage of populist parties as a potential catalysator of populist voting behaviour through the mediating role of emotions.
News Framing and Emotions as Mediators
The study of emotional responses in the context of news framing is relatively recent, and as such, there is only little theorising. Previous research has, however, depicted that news framing effects on political behaviour are, to a large extent, dependent on individuals’ emotional responses to a particular frame they are exposed to (Myers, Nisbet, Maibach, &
Leiserowitz, 2012; Lecheler & de Vreese, 2019). Gross (2008) even states that news framing effects are predominantly existent because of the power of news frames to evoke emotional responses.
When a message comprises information that serves as a cue to a core relational theme of a discrete emotion (e.g., hostile response to an imminent threat for anger; positive confirmation for satisfaction), a specific emotion (i.e., anger, satisfaction) is elicited. Former research displays that an affect-based response to an emotional cue that was
subconsciously processed by the individual through the exposure to news framing can have an effect on how one’s attitude, opinion formation and political behaviour unfolds (Holm, 2012; Lecheler, Schuck, & de Vreese, 2013; Lecheler, Bos, & Vliegenthart, 2015). The role of emotions in news framing is hence of particular interest with respect to the already emotionally charged concept of populism (Wirz, 2018) – a topic whose prominence in the media has already caused heated public debates and which is likely to elicit emotional responses amongst citizens.
Previous research has already characterised emotionality as an intrinsic element of populism (e.g. Mazzoleni, Steward, & Horsfield, 2003; Plasser & Ulram, 2003; Meyer, 2006). Taking its media representation into account, Wirz (2018) has discovered that populist content in the news may in fact elicit discrete emotions in the absence of populist style – simply by covering an issue or cue associated with populist movements. Further, Lecheler, Schuck and de Vreese (2013) show that discrete emotions play a role as mediators of news framing effects. As a consequence, it can be assumed that framed news coverage of populist parties, no matter on which side of the political spectrum they are situated, evokes a greater emotional response, be it a positive or a negative one, than neutrally written and fact-driven news coverage. As a consequence, the following hypothesis is included in the research process:
H1a: Negative news framing of right-wing populist parties will elicit more emotional response than news coverage without a frame.
H1b: Negative news framing of left-wing populist parties will elicit more emotional response than news coverage without a frame.
Starting from the premise that the effect of news framing on one’s degree of emotional response is, however, subordinate to one’s personal political ideology, the following section will discuss the effect one’s personal political standpoint – and
subconscious preference for populist parties – can have on the way news coverage about a populist party is perceived.
Political Ideology as a Moderator
Generally speaking, one can assume that a negative news frame about a populist party results in increased negative emotions. However, it can be expected that the effect of news framing on emotional response is moderated by “positional agreement” (Bos, Lefevere, Thijssen, & Sheets, 2014, p. 11) as news media coverage with a party cue present does in fact have a stronger effect on voters who already, even if partially, agree with the position voiced by a populist party. This is pillowed by appraisal theories of emotions (Lazarus, 1991), arguing that different assessments of a scenario can trigger different discrete emotions that “color judgments in a manner consistent with the emotion’s underlying appraisal pattern” (Rico, Guinjoan, & Anduiza, 2017, p. 445). A cross-country panel study by Müller et al. (2017) indicates that the higher the dose of exposure to populist news coverage, the more it enhances prior agreement and disagreement with populism. Sheets, Bos and
Boomgaarden’s (2015) research further displayed that populist media cues operate differently for experimental subjects on opposite ends of the ideological spectrum. Nabi (1999) argues that different frames can evoke different emotional responses amongst different individuals, depending on which knowledge constructs are recalled by the frame. Wirz’s (2018) study further depicts that the experience of both positive and negative
emotions to populist messages is stronger for people who, at least to a certain degree, agree with populist ideology.
While anger activates the disposition system and leads individuals to rely on heuristics, ergo instinctive routines, confrontational responses and habitual commitments, satisfaction as a nuance of enthusiasm is prone to reinforce people’s habitual behaviours and existing political attitudes (Lerner & Keltner, 2001; Ladd & Lenz, 2008; Brader & Marcus, 2013). Although populism, especially right-wing populism, has received a large amount of media attention, previous large-scale content analyses display that the media’s coverage of populist parties is often rather negative and sceptical instead of neutral and factual (Esser et al., 2017; Wettstein et al., 2018).Given anger’s intrinsic morality aspect, focusing on the perception that an event is unfair or illegitimate, it is thus expected that more anger is elicited amongst people who (partially) agree with the party’s ideology when a populist party is presented in a negative light. Respectively, it can be assumed that more satisfaction is evoked once people do not (partially) agree with the party’s ideology, given that framing confirms their prior attitudes, leading to attitudinally congruent persuasion (Stroud, 2008). Against this background, the following four hypotheses can be constructed:
H2a: Negative news framing of right-wing populist parties will elicit more anger amongst more right-leaning participants than news coverage without a frame. H2b: Negative news framing of left-wing populist parties will elicit more anger amongst more left-leaning participants than news coverage without a frame.
H3a: Negative news framing of right-wing populist parties will elicit more satisfaction amongst more left-leaning participants than news coverage without a frame.
H3b: Negative news framing of left-wing populist parties will elicit more satisfaction amongst more right-leaning participants than news coverage without a frame.
Arguably, anger is the emotion most connected to the spread of populism – mainly because the attribution of blame, populism’s core theme, “strongly resonates anger’s underlying appraisals” (Rico, Guinjoan, & Anduiza, 2017, p. 455). But what role does anger play in explaining people’s support for populism?
Emotions and Voting Behaviour
Emotional appeals have proven to be able to create democratically desirable behaviour (e.g. Brader, 2005). Kühne (2012) has, for instance, examined that emotions induced by the media can indeed significantly influence political attitudes and behaviour. Broad research by Weber (2012) and Lecheler, Schuck and de Vreese (2015) further depicts that voters exposed to anger and enthusiasm appeals are more willing to vote than voters exposed to other emotional appeals or no appeals at all. According to Brader (2006), anger increases the willingness to participate in politics, but also strengthens less careful forms of information processing, as anger induces individuals “to follow more superficial
considerations and stereotypes” (Rico, Guinjoan, & Anduiza, 2017, p. 449). A study by Kieslich and Marcinkowski (2019) even indicates that citizens exposed to news framing eliciting anger show greater willingness to support populistic demands. Rico, Guinjoan and Anduiza (2017) further found out in their three-wave panel study that anger is a key
motivating factor associated with support for populism – at the individual level and over time – as populism is in principle polarising, therefore offering fertile soil for the action tendency of anger. Given that, in accordance with appraisal theories of emotions (Lazarus, 1991), we expect anger with respect to the framed news coverage to be the emotional response that resonates most with populism per se, it is assumed that the more angry participants feel after reading the article, the more they are in fact willing to vote for a populist party – be it on the left or the right side of the political spectrum.
As a consequence, the following hypotheses will further guide the research process (for conceptual model, see Appendix A4):
H4a: The more anger is elicited by the news coverage, the more participants are willing to vote for a right-wing populist party.
H4b: The more anger is elicited by the party, the less participants are willing to vote for a right-wing populist party.
H5a: The more anger is elicited by the news coverage, the more participants are willing to vote for a left-wing populist party.
H5b: The more anger is elicited by the party, the less participants are willing to vote for a left-wing populist party.
Previous psychological research has already displayed that (political) attitudes can be tremendously influenced by unconscious processes (Greenwald, Poelman, Uhlmann, & Banaji, 2009). Nevertheless, most research only measures the explicit component of political attitudes and subsequent behaviour – even if studies have shown that self-reports may distort actual voting intention (Bos, Sheets, & Boomgaarden, 2018), because “participants who feel they hold a minoritarian position prefer to give insincere answers” (Roccato & Zogmaister, 2010, p. 271). Generally speaking, implicit attitudes are prone to correlate with their explicit counterparts in situations of non-sensitive attitudes (Greenwald, McGhee, & Schwartz, 1998) – but the more sensitive a topic, the less are implicit and explicit attitudes likely to correlate. Support for populist parties can hence be seen as a highly controversial and socially sensitive topic and might therefore prompt less honest self-reporting amongst participants. As a result, it is highly likely that citizens have in fact less negative attitudes towards or sympathise much more with populist parties implicitly than they report explicitly, as measures of support for populist parties which are considered to be illegitimate by both the media and a large share of the public can substantially underestimate the amount of explicit support for reasons of social desirability (Arzheimer, 2009). After all, the electoral success of populist parties clearly depicts that these parties do indeed find an echo amongst citizens who are entitled to vote.
Given that the measurement of implicit attitudes has already proven to be a good predictor for voting behaviour in former communications research (e.g. Roccato & Zogmaister, 2010; Sheets, Bos, & Boomgaarden, 2015; Bos, Sheets, & Boomgaarden, 2018), a participant’s attitude towards a populist party and hence his or her implicit populist
voting intention is to be examined by means of implicit measures, allowing for automatic and uncontrolled associations towards the party.
Methods
Design
The experiment was designed as a 2x2 between-subjects factorial design (frame/control right-wing populism x frame/control left-wing populism) in which the
manipulating factor news framing as between-subjects variable was examined at four levels, those being the right-wing populism framing condition, the right-wing populism control condition, the left-wing populism framing condition and the left-wing populism control condition.
Participants
The sample reflected variation regarding age, gender, political ideology and level of education, although it has an overrepresentation of higher educated adolescents prone to vote for the Green party (45.4%) (see Appendix A1). A total of 178 German speaking experimental subjects were recruited for the following experiment. Every participant was acquired by means of a convenience sample through the broad personal and academic offline and online network of the researcher and online survey platforms. As a consequence, no random selection process was executed, resulting in a non-probability sampling
procedure. Participants who did not give informed consent, only partially filled out the survey and took longer than 3123 seconds (more than 50 minutes) to fill out the questionnaire were excluded from the study prior to testing (n = 37).
As a consequence, 141 experimental subjects (95 women, 46 men) at the age of 19 to 70 took part in the experiment (MAge = 28.96, SDAge = 10.11, MdAge = 25.00). Participants were mostly highly educated (n = 136), with only a marginal number of participants coming from a lower educational background (n = 5). 85.1% of all experimental subjects were born in Germany, whereas 14.9% of the participants spoke German on native speaker level. When
asked about their media consumption habits, participants ranked online media as one of their primarily used media outlets, followed by social media and television. Radio, print media and podcasts were, in comparison, used less frequently (see Appendix A2). Furthermore,
participants mostly assessed the quality of German news media coverage as high (M = 6.22,
SD = 1.06) (see Appendix A3).
Stimuli
After answering the pre-questionnaire, participants in all conditions were presented with a newspaper article (see Appendix B). In the right-wing populism frame condition, experimental subjects were exposed to an article about the rise of the Alternative for Germany. I manipulated the article by framing the party as illegitimate and a threat to democracy, therefore inacceptable as a coalition partner. As a result, I depicted the right-wing populist character of the party, placing focus on its ties to right-right-wing extremist groups. The article discussed the entry of the AfD into parliament, to which degree it has affected the tone of debate and democratic process in the Bundestag and how the party is using
emotional rhetoric to provoke its political opponents. In the control condition, the same news story was presented. However, this version did not portray the party as a threat to
democracy; it merely outlined its development and current position in the German parliament. Further, the word populist was not used explicitly once. In the left-wing populism frame condition, participants were exposed to a similar article, this time covering the rise of The Left. I manipulated the article again by framing the party as a threat to democracy and not being capable of forming a coalition. As a consequence, I outlined the left-wing populist character of the party, putting emphasis on its party history and ties to left-wing extremist groups. The article covered the establishment of The Left as opposition party, to which degree it has influenced the democratic process in parliament and how the party is using motions in the Bundestag to provoke its political opponents. In the control condition, participants were exposed to the same news story. Nevertheless, this version did neither depict the party as a threat to democracy, nor did it mention the explicit word populist once. In all conditions, participants were exposed to an article published on the same webpage of
the liberal German daily newspaper Der Tagesspiegel (The Daily Mirror) written by the same author.
Each of the four articles was of roughly equal length (350 – 400 words) to ensure high internal validity of the experiment. No visual material was used to guarantee that the wording of the newspaper article is the only factor affecting the emotional response of the
participants, given that research has already displayed that including visuals and text in communication elicits stronger emotions amongst the audience (Maier, Slovic, & Mayorga, 2017).
Procedure
After taking a proactive interest in taking part in the experiment, participants were sent the study’s URL. The self-completion online experiment took approximately fifteen minutes to complete (M = 1007.25, SD = 560.33). Participants were recruited over a sixteen-day time frame between May 23, 2019 and June 8, 2019. Given that several Multiple Target Implicit Association Tests (MT-IAT) do not run in one survey, every participant was at first randomly assigned by the online survey software Qualtrics to one of two studies at group level: the right-wing populism study including the IAT for the AfD and the left-wing populism study including the IAT for Die Linke. Prior to the exposure to the stimulus material, every experimental subject gave informed consent and confirmed his or her voluntary participation in the online experiment. No rewards were given for contributing to the research.
Participants in both studies were first asked demographic questions, then answered questions about their media consumption and about their own political alignment and loyalties. Afterwards, participants of this between-subjects post-test only online experiment were again randomly assigned to one of in total four different experimental treatments at the individual level: frame right-wing populism (n = 35), control right-wing populism (n = 36), frame left-wing populism (n = 32) and control left-wing populism (n = 38). The minimum forced exposure time to the stimulus material was 45 seconds (M = 117.47, SD = 83.01).
After reading the article, participants were asked to take the MT-IAT to measure their implicit populist attitudes. The MT-IAT component was used via IATgen and run in Qualtrics software to ensure a smooth transition between the questionnaire and the IAT.
Subsequently, if of German decent, participants were asked to state their explicit voting intention. Lastly, attendants were debriefed after filling out the questionnaire and thanked for their cooperation.
Manipulation Check
For the right-wing populism study, the manipulation check was successful.1 In the framing condition, participants were significantly more likely to believe the newspaper article framed the AfD as a threat to democracy (M = 6.26, SD = 0.74) than in the control condition (M = 4.22, SD = 0.29), t (47.34)= 6.38, p < .001, 95% CI [1.39, 2.68], d = 3.63. For the left-wing populism study, an independent sample t-test displayed comparable results.2 In the framing condition, participants were significantly more likely to believe the newspaper article framed Die Linke as a threat to democracy (M = 4.44, SD = 1.66) than in the control
condition (M = 2.29, SD = 0.98), t (48.42)= 6.42, p < .001, 95% CI [1.48, 2.82], d = 1.58.
Implicit Populist Attitudes as Dependent Variable
In order to test participants’ implicit attitudes towards a populist party, an Implicit Association Test in accord with Greenwald, McGhee and Schwartz (1998) for both the right-wing and the left-right-wing populist party was conducted3, comparing “AfD” (blue, Alexander
Gauland, migration, Björn Höcke, nationalism, lying press) and “FDP” (yellow, free market economy, Christian Lindner, globalisation, liberalism, entrepreneurship) targets on a “Good”
(happy, peace, wonderful, love, laughter, joy) versus “Bad” (agony, failure, hurt, awful, cruel,
evil) dimension. Same goes for “Die Linke” (red, Sahra Wagenknecht, social justice, Gregor Gysi, comrades, Rosa Luxemburg, socialism), respectively. Amongst those participants who
took the right-wing populist association test, MT-IAT D-scores ranged from -1.50 to 0.98 (MAfD = -0.54, SDAfD = 0.45), showing a negative skew towards the AfD on an implicit level. Respectively, amongst those participants who took the left-wing populist association test,
MT-IAT D-scores ranged from -0.85 to 1.15 (MLinke = 0.13, SDLinke = 0.48), showing a slight positive skew towards Die Linke on an implicit level.
Mediation
Emotional Response. After the exposure to the news article, participants were asked
to indicate which emotions they experienced while reading the article. A widely recognised scale for the measurement of emotions during media exposure is the modified differential affect scale by Renaud and Unz (2006). This scale provides a three-item measurement for discrete emotions. However, the scale does not include items for satisfaction. Therefore, these items were developed according to the measurement of contentment provided by the M-DAS model. Both anger and satisfaction were measured with three items for which the participants rated how much they felt these emotions on a 7-point scale ranging from 1 (“not at all”) to 7 (“very much”). Anger (furious, angry, irritated, M = 4.36, SD = 1.53, Cronbach’s α = .91) and satisfaction (content, encouraged, reinforced, M = 4.51, SD = 1.29, Cronbach’s α = .724) both proved to be reliable enough. I also measured fear with three items (anxious, afraid, frightened, M = 4.78, SD = 1.41, Cronbach’s α = .79), given that it’s possible that the emotional frames induce fear in addition to anger and satisfaction.5
Moderated Mediation
Political Ideology. The moderator was tested by asking participants on a 11-point
self-placement scale (0 = left, 10 = right) where their own political identification would fall (M = 3.68, SD = 1.80). A 11-point self-indicator scale was selected to provide participants with a neutral midpoint (e.g. Kroh, 2007; Zuell & Scholz, 2016). Higher scores on the scale indicate a more right-leaning view, whereas lower scores indicate a more left-leaning standpoint.
Results
Randomisation Check
Prior to analysis, both left-wing and right-wing populist party data sets were merged into one. Three between-conditions randomisation checks on the control variables gender6,
age7 and level of education8 were conducted, revealing no significant differences between all four experimental conditions regarding the variables at hand.
News Framing and Its Effect on Emotional Response
To test whether negative news framing of right-wing populist parties elicits more emotional response than news coverage without a frame, an independent sample t-test was conducted (Figure 1), with Condition AfD as independent variable and Emotional Response as dependent variable.9 The findings displayed that there was no significant difference in level of emotional response between the frame condition (M = 4.03, SD = 0.82) and the control condition (M = 4.17, SD = 0.56), t (69) = -0.87, p = .387, 95% CI [-0.48, 0.19]. Thus, H1 is rejected.
Figure 1. News framing and its effect on emotional response. Note. The dependent variable
ranges from 1 (“low emotional response”) to 7 (“high emotional response”).
To examine whether negative news framing of left-wing populist parties evokes more emotional response than news coverage without a frame, another independent sample t-test was performed (Figure 1), with Condition Die Linke as independent variable and Emotional
Response as dependent variable.10 The results showed that there was no significant
4.03 4.17
4.94
5.13
Frame AfD Control AfD Frame Die Linke Control Die Linke
difference in level of emotional response between the frame condition (M = 4.94, SD = 5.13) and the control condition (M = 5.13, SD = 0.96), t (68) = -0.87, p = .389, 95% CI [-0.65, 0.26]. Hence, H2 cannot be supported.
Political Ideology and Its Effect on Emotional Response to News Framing
To test hypotheses 2a and 2b, two linear regression analyses were conducted (Table 1). For the right-wing populism study11, news framing, b = -0.32, t = -0.54, p = .592, 95% CI [-1.49, 0.86], and political ideology, b = 0.08, t = 0.76, p = .453, 95% CI [-0.12, 0.27], did not significantly predict anger as emotional response. On average, participants exposed to the
AfD framing condition scored 0.32 lower on the anger scale than participants in the control
condition. With one unit increase on the left-right ideology scale, the predicted anger increased by 0.08. The interaction effect between news framing and political ideology did also not yield significant results, b = 0.00, t = 0.04, p = .965, 95% CI [-0.26, 0.28]. The effect of news framing on anger as emotional response was hence not more positive or negative for more right-leaning participants. These results are not supportive of H2a.
Table 1
Regression models predicting the effect of news framing on anger as emotional response moderated by political ideology
Variable Model I b Model II b
(Constant) 3.30* (0.43) 5.99* (0.52)
Framing -0.32 (0.59) -1.19 (0.76)
Political Ideology 0.08 (0.10) -0.11 (0.13)
Framing x Political Ideology 0.00 (0.14) 0.15 (0.20)
Adjusted R² -0.007 0.030
F for change in R² 0.85 1.70
N 71 70
* p < .050 Note. Unstandardized regression weights. Standard errors between brackets.
For the left-wing populism study12, news framing, b = -1.19, t = -1.56, p = .123, 95% CI [-2.72, 0.33], and political ideology, b = -0.11, t = -0.86, p = .396, 95% CI [-0.37, 0.15], did not significantly predict anger as emotional response. On average, participants exposed to the Die Linke framing condition scored 1.19 lower on the anger scale than participants in the
control condition. With one unit increase on the left-right ideology scale, the predicted anger decreased by 0.11, indicating the more left-leaning a participant was, the more anger was elicited. The interaction effect between news framing and political ideology was also not significant, b = 0.15, t = 0.74, p = .464, 95% CI [-0.26, 0.56]. These results do not support H2b.
To test hypotheses 3a, a linear regression analysis was performed (Table 2). 13
Table 2
Regression models predicting the effect of news framing on satisfaction as emotional response moderated by political ideology
Variable Model I b Model II b
(Constant) 5.32* (0.49) 3.26* (0.49)
Framing -0.90 (0.68) 2.09* (0.73)
Political Ideology -0.16 (0.11) 0.29* (0.12)
Framing x Political Ideology 0.14 (0.16) -0.49* (0.20)
Adjusted R² 0.006 0.080
F for change in R² 1.13 3.00
N 71 70
* p < .050 Note. Unstandardized regression weights. Standard errors between brackets.
News framing, b = -0.90, t = -1.32, p = .190, 95% CI [-2.25, 0.46], and political ideology, b = -0.16, t = -1.41, p = .162, 95% CI [-0.39, 0.07], did not significantly predict satisfaction as emotional response. On average, participants exposed to the AfD framing condition scored 0.90 lower on the satisfaction scale than participants in the control condition. With one unit increase on the left-right ideology scale, the predicted satisfaction decreased by 0.16, indicating the more left-leaning a participant was, the more satisfaction was elicited. The interaction effect between news framing and political ideology was also not significant, b = 0.14, t = 0.91, p = .366, 95% CI [-0.17, 0.45]. These results are not supportive of H3a.
To test 3b, another linear regression analysis was conducted.14 News framing, b = 2.09, b* = 0.80, t = 2.87, p = .006, 95% CI [0.64, 3.55], and political ideology, b = 0.29, b* = 0.36, t = 2.40, p = .019, 95% CI [0.05, 0.54], both significantly predicted satisfaction as
emotional response. On average, participants exposed to the Die Linke framing condition scored 2.09 higher on the satisfaction scale than participants in the control condition. With one unit increase on the left-right ideology scale, the predicted satisfaction increased by 0.29, indicating the more right-leaning a participant was, the more satisfaction was elicited. The interaction effect between news framing and political ideology was also significant, b = -0.49, t = -2.53, p = .014, 95% CI [-0.88, -0.10]. The effect of news framing on satisfaction as emotional response was hence more negative for more right-leaning participants. These results are supportive of H3b.
Anger and Its Effect on Implicit Attitudes Towards a Populist Party
To test whether participants’ implicit attitudes towards a populist party were affected by anger elicited by the news coverage or the party itself, two independent sample t-tests were performed, with Anger as independent and Implicit Attitudes as dependent variable (Table 3). For the right-wing populism study15, the findings displayed that there was no significant difference in implicit attitudes in favour of the AfD between the news coverage (M = -0.27, SD = 0.89) and the party condition (M = -0.59, SD = 0.40), t (4.16) = 0.82, p = .455, 95% CI [-0.76, 1.42]. Thus, H4a and H4b are rejected.
Table 3
T-tests displaying the effect of anger elicited either by the news coverage or the party itself on implicit attitudes towards a populist party
Anger: News coverage Anger: Party
N M SD N M SD Voting Intention AfD 5 -0.27 0.89 50 -0.59 0.40 Voting Intention Die Linke 8 0.18 0.28 5 0.06 0.63
*p < .050 Note. Implicit attitudes towards a populist party were measured on a scale between -2 and +2, with more negative numbers indicating lower implicit attitudes and more positive numbers indicating higher implicit attitudes.
For the left-wing populism study16, the findings displayed that there was no significant difference in implicit attitudes towards Die Linke between the news coverage (M = 0.18, SD = 0.28) and the party condition (M = 0.06, SD = 0.63), t (11) = 0.50, p = .629, 95% CI [-0.43, 0.68]. However, what needs to be taken into account is that the findings are of very low power due to small sample size. Nonetheless, the results are not supportive of H5a and H5b.
Discussion and Conclusion
Ever since the rise of populism, researchers have argued that the media does indeed play a substantial role in fostering the success of populist parties (e.g. Boomgaarden & Vliegenthart, 2007; Krämer, 2014; Mudde, 2004; Rooduijn, 2014). Despite a few examples supporting the assumption that media exposure to populist issues can affect populist voting behaviour (Bos, van der Brug and de Vreese, 2011; Hameleers, Bos and de Vreese, 2018), we still know little about which elements of news media coverage can have a drastic impact on citizens’ attitudes towards populist parties.
The aim of this study was thus to investigative media framing’s role in affecting the electoral success of populist parties and how this effect might be influenced by the emotional responses news framing can elicit. This research chose a slightly different angle than former studies, taking a step forward in focussing upon the media’s representation of populist parties. Whilst previously established research primarily aimed at media populism (Hameleers, Bos and de Vreese, 2018), party cues (Bos, Lefevere, Thuijssen, & Sheets, 2014) or issue coverage closely intertwined with populist parties from the right side of the political spectrum (Sheets, Bos, & Boomgaarden, 2015), the main purpose of this study was to examine how news framing, which explicitly blames a populist party from both ends of the political spectrum to be a threat to democracy, can have an immediate effect on one’s emotional response and subsequently on people’s implicit populist voting intention.
The results display that the exposure to negative news framing of right- and left-wing populist parties does not elicit more emotional response than the exposure to no frame. As a matter of fact, despite the findings being insignificant, they clearly depict that the exposure to
no news framing evokes a higher degree of emotional response than the framing condition for both populist parties. This is not in accordance with previous research by Wirz (2018), stating that discrete emotions are elicited by news coverage about populist parties in the absence of populist style. A potential explanation of the outcome is, however, that media coverage of populist parties in Germany has always been emotionally charged (Gäbler, 2018) and that a sole party cue with or without a negative depiction of the party is already capable of evoking emotions, given the topic’s intrinsic emotionality (Wirz et al., 2018).
Secondly, this research expected some citizens to be more likely to be affected by news framing than others. Hence, the study investigated whether political ideology had a moderating effect on the participants’ emotional response to the news framing they were exposed to, with more right-leaning individuals (left-leaning individuals, respectively) being more angry with biased news coverage of the right-wing populist party (the left-wing populist party, respectively). Despite the fact that previously established research has already
identified political ideology as an indicator for a stronger effect of news framing on emotional response (Bos, Lefevere, Thijssen, & Sheets, 2014; Sheets, Bos, & Boomgaarden, 2015; Wirz, 2018), these findings showed that political ideology does not moderate the effect of exposure to news framing on anger as emotional response. Nevertheless, I also tested for satisfaction as emotional response. The results displayed no evidence for more left-leaning citizens to be more satisfied with negative news coverage of a right-wing populist party. The findings indicate, however, a negative effect for more right-leaning citizens who are exposed to a news article about a left-wing populist party. For these citizens, the negative news framing of Die Linke confirmed their prior attitudes, a result of attitudinal congruent persuasion (Stroud, 2008), implying that a sole party cue has a stronger effect than the framing itself.
Thirdly, this research analysed whether anger as emotional response to news framing can influence populist voting intention, with anger elicited by the news coverage having a greater effect on implicit voting intention than anger elicited by the party itself. The findings suggest that anger elicited by negative news coverage of a left-wing (or right-wing) populist
party cannot increase support for the very same. This finding is not in accordance with previous research (Lecheler, Schuck, & de Vreese, 2015; Rico, Guinjoan, & Anduiza, 2017; Kieslich & Marcinkowski, 2019), stating anger enables greater willingness to support populist parties and demands. Following up on this, what needs to be taken into consideration is that the party selected as a representative of right-wing populism in Germany is controversial in its core: The Alternative for Germany is no longer only considered to be a representative of right-wing populism, but rather regarded as a far-right, anti-immigrant party increasingly melding with extremists and further spreading its tentacles into the centre of society. Against this background, anger elicited by the party itself in both the framed and no frame condition comes as no surprise. It is no longer the framing of the party that elicits discrete emotions; it’s rather the party’s presence itself. The Left, on the other hand, is a case of its own in Germany: Despite its obvious use of populist rhetoric and clear-cut, moderately successful application of the “us against them” formula, it is not considered to be a populist party by the majority of German citizens. Future research is to tackle this issue by asking participants questions specifically tailored to their attitudes towards the news coverage of populist parties in order to identify whether the exposure to framed news coverage is the actual trigger for their emotional state.
There are further limitations which might have imperilled the results of this research. What needs to be taken into account is the high proportion of highly educated, slightly left-leaning participants compared to only a few lower educated or right-left-leaning experimental subjects. Whilst sample size was large enough, with participants equally distributed across all four conditions, the convenience sample was highly skewed in favour of one specific milieu, implying that a larger sample size might have uncovered different results. Further research is hence needed to investigate whether the same or different effects occur when analysing a sample representative of the entire German society.
Furthermore, the choice of country poses a threat to the study’s external validity as an analysis of only one country-specific case study constraints generalisability. Germany is a very specific case, given the country’s constant process of accounting for its history of
National Socialism and the German Democratic Republic. Further cross-country research across a broader range of countries will be of the essence to concretise our understanding of the media’s role in strengthening the success of populist parties – that is, the relationship between news media coverage of populist parties, citizen attitudes and voting intentions is very likely to vary by context.
Against this background, the operationalisation of implicit populist attitudes needs to be regarded as a potential gap in this study, given that the results are highly dependent on case selection – and a multiple target measurement. Both populist parties were only, for obvious reasons (see Notes 3), compared with the liberal FDP – a smaller party which is not currently running the country. As previous research with a single target measurement has already yielded significant results (Bos, Sheets, & Boomgaarden, 2018), it would be interesting to either again implement a single target measurement for the German case in future research or to analyse whether stronger or weaker effects could be found when comparing both populist parties with a governing party that can actually be held accountable – and blamed – for current policies and legislation.
What needs to be further taken into account is that the term populism has lately been excessively used in news coverage, despite being mostly devoid of any broadly known, precise definition. Therefore, a news article refraining from the use of populist terminology might elicit a stronger emotional response because of the unusual fact that it is written in a neutral and objective manner. Future research is hence to analyse whether another manipulative feature, e.g. incorporating an issue cue alongside the party cue, would yield different results, as a broad study by Kieslich and Marcinkowski (2019) and research by Sheets, Bos and Boomgaarden (2015) already suggest.
Lastly, a clear drawback of the study is the exposure to one article at exclusively one single point in time – a long-term effect of news framing on populist voting intention is hence not assessable. Consequently, longitudinal measurement of the exposure to news framing on populist voting intentions, e.g. in the form of a combination of content analysis and
experiment, will allow us to better understand the impact of news media’s coverage of populist parties on citizens’ attitude towards populist parties in the future.
Despite these constraints, this research does not only provide a conceptual framework to better grasp the role news coverage of populist parties can play in evoking positive and negative emotional responses amongst its readers, thereby, shaping citizens’ attitudes towards populist parties for the better or the worse; it further serves as a
substantive argument for media organisations to reconsider the effect their words can have on their readers in order not to foster more societal polarisation – a cornerstone future research is to build onto.
Notes
1. Levene’s test for equality was significant, F = 26.83, p < .001. Therefore, equal variances cannot be assumed, with the two groups being of roughly equal size.
2. Levene’s test for equality was significant, F = 16.78, p < .001. As a consequence, equal variances cannot be assumed, with the two groups being of roughly equal size.
3. Implicit attitudes towards the AfD (Die Linke, respectively) are measured by the D-scores produced by the MT-IAT. The MT-IAT measures individual differences in mental
associations between multiple “target” concepts, in this case the AfD (Die Linke,
respectively), the Free Democratic Party (FDP) and two opposing evaluations (good or
bad). The FDP was selected as opposing target, given it is the only party currently present in the German Bundestag which cannot be explicitly assigned to either the left or the right side of the political spectrum. Participants were asked to classify words that represent the target concepts and evaluations (AfD/Die Linke, FDP, good, and bad) into two categories, using distinct keystrokes and responding as quickly as possible. In one round, participants were told to hit one key (e.g. the “I” key) if either “AfD” or “good”/ “Die Linke” or “good” came up on the screen, and a different key (e.g. “E” key) if either “FDP” or “bad” came up. In another scenario, participants hit the “I” key if either “AfD” or “bad”/ “Die Linke” or “bad” came up and “E” if “FDP” or “good” came up. Relative association strengths were measured by examining the speed with which participants responded in the contrasting conditions (AfD/ Die Linke and good with the same key versus AfD/ Die
Linke and bad with the same key). Longer response times and more errors indicate
weaker associations between concepts and evaluations, while shorter response times with fewer errors indicate stronger associations. In order to counterbalance targets and categories, participants were randomly assigned to one out of four possible IAT
configurations automatically provided by IATgen. IAT scores can generally run from -2 to +2, where negative numbers in this case mean negative attitudes towards the AfD/ Die
of 0 indicates no relative association of the AfD/ Die Linke as “good” versus “bad”— both concepts are associated equally strong.
4. Cronbach’s α was considered reliable enough when item 1 (“content”) was excluded. 5. The KMO measure of sampling adequacy showed a result of 0.80, which is above the
required 0.50 score. Bartlett’s test of sphericity further proved to be significant (p < .001). As a consequence, both requirements to continue with the factor analysis were met. The Eigenvalue criterion table displayed there are two Eigenvalues above the 1.0 threshold (Component 1: 4.31, Component 2: 1.78). The result is supported by the scree plot, depicting two factors before the point of inflexion. These two factors explain 67.63 % of variance in the original items (Factor 1: 47.90%, Factor 2: 19.74%). After an orthogonal rotation was performed, the pattern matrix displayed that there are two clusters of items corresponding to the two factors when taking the factor loading threshold of 0.40 into account, which can be labelled as follows: (1) Negative Emotions and (2) Positive
Emotions. The rotated component matrix further depicted that Item 4 (“encouraged”) had the strongest (0.84) and Item 1 (“content”) the weakest (0.58) relationship with Factor 2, whereas Item 6 (“angry”) had the strongest (0.91) and Item 3 (“afraid”) the weakest (0.79) relationship with Factor 1. For the purpose of this analysis, Principal Component Analysis was, however, neglected, given that anger and fear needed to be treated as different discrete emotions. A reliability analysis using Cronbach’s alpha was conducted to test how reliable the scores of the items for “anger”, “fear” and “satisfaction” are. For further analysis, three new variables – Anger, Fear and Satisfaction – were computed by taking the joint means of the above-mentioned correlating items. The higher one scores on each of these scales, the more one has felt the emotion while reading the article.
6. To test whether the randomisation of participants across conditions in terms of gender was successful, a randomisation check using a contingency table was performed, with
Conditions as independent variable and Gender as dependent variable. The findings
experimental conditions, χ² (3, N = 141) = 1.22, p = .747. As a consequence, the randomisation of male and female participants across conditions was successful. 7. To examine whether the randomisation of participants across conditions in terms of age
was successful, a one-way ANOVA was conducted, with Conditions as independent variable and Age as dependent variable. Levene’s test of equality was not significant, F = 2.18, p = .093. As a consequence, equal variances can be assumed and the
randomisation check is considered to be reliable. The results display that there is no significant difference between the mean age of all four experimental subjects, F (3, 150) = 0.72, p = .540, η² = .99. Thus, participants were indeed randomly assigned to the
conditions in terms of age.
8. A last randomisation check was performed using a contingency table, with Conditions as independent variable and Level of Education as dependent variable. The findings
revealed that all educational levels were equally distributed across all four experimental conditions, χ² (24, N = 141) = 19.34, p = .734.
9. Levene’s test of equality was not significant, F = 1.31, p = .256. Consequently, equal variances can be assumed and the t-test is deemed to be reliable.
10. Levene’s test of equality was not significant, F = 0.17, p = .389. Thus, equal variances can be assumed and t-test is considered to be reliable.
11. Given that the condition variables Condition AfD and Condition Die Linke are
dichotomous ones, two dummy variables needed to be computed before the analysis, denoting 0 for control AfD (or Die Linke) and 1 for frame AfD (or Die Linke). The
regression model with Anger as dependent, Condition AfD as independent and Political
Ideology as moderator variable was not significant, F (3, 67) = 0.86, p = .473. Residuals
were normally distributed and the assumption of homoscedasticity was met.
12. The regression model with Condition Die Linke as independent variable was not proven to be significant either, F (3, 66) = 1.70, p = .175. Residuals were normally distributed and the assumption of homoscedasticity was met.
13. The regression model with Satisfaction as dependent, Condition AfD as independent and
Political Ideology as moderator variable was not significant, F (3, 67) = 1.13, p = .343,
despite residuals being normally distributed. The assumption of homoscedasticity was not met. The regression model therefore needs to be treated with caution.
14. The regression model with Condition Die Linke as independent variable was significant, F (3, 66) = 3.00, p = .037. The assumptions of homoscedasticity and distribution of
residuals were met. The regression model therefore deemed to be reliable.
15. Prior to analysis, anger was split into two groups: Anger elicited by the news coverage and anger elicited by the party itself. For the right-wing populism study, Levene’s test of equality was significant, F = 11.30, p < .001. Consequently, equal variances cannot be assumed, with the two groups not being of equal size.
16. For the left-wing populism study, Levene’s test of equality was not significant, F = 2.17, p = .168. Consequently, equal variances can be assumed, despite the two groups not being of equal size and n < 30.
References
Adriaansen, M., & van Praag, P. (2010). Nieuwe scheidslijnen en de turbulente relatie tussen politiek, media en burgers [New strip lines and the turbulent relationship between politics, media and citizens]. In H. Dijstelbloem, P. den Hoed, J. W. Holtslag, & S. Schouten (Eds.), Het gezicht van de publiekezaak: openbaar bestuur onder ogen (pp. 241-272). Amsterdam: Amsterdam University Press.
Albertazzi, D., & McDonnell, D. (2008). Twenty-first century populism: The spectre of
Western European democracy. Basingstoke, England: Palgrave Macmillan.
Arzheimer, K. (2009). Contextual factors and the extreme right vote in Western Europe, 1980–2002. American Journal of Political Science, 53(2), 259–275.
Boomgaarden, H.G. & Vliegenthart, R. (2007). Explaining the rise of anti-immigrant parties: The role of news media content. Electoral Studies, 26(2007), 404-417.
Bos, L., & Brants, K. (2014). Populist rhetoric in politics and media: A longitudinal study of the Netherlands. European Journal of Communication, 29(6), 703-719.
Bos, L., & van der Brug, W. (2010). Public images of leaders of anti-immigration parties: Perceptions of legitimacy and effectiveness. Party Politics, 16, 777–799.
Bos, L., van der Brug, W., & de Vreese, C. (2011). How the Media Shape Perceptions of Right-Wing Populist Leaders. Political Communication, 28(2), 182–206. doi: 10.1080/10584609.2011.564605
Bos, L., Lefevere, J., Thijssen, R., & Sheets, P. (2017). The impact of mediated party issue strategies on electoral support. Party Politics, 23(69), 760-771.
Bos, L., Sheets, P., & Boomgaarden, H. G. (2018). The Role of Implicit Attitudes in Populist Radical-Right Support. Political Psychology, 39(1), 69-87.
Brader, T (2006). Campaigning for Hearts and Minds: How Emotional Appeals in Political
Ads Work. Chicago: University of Chicago Press.
Brader, T., & Marcus, G. (2013). Emotion and political psychology. In L. Huddy, D.O. Sears, & J.S. Lev (Eds.), The Oxford Handbook of Political Psychology (pp. 165-204). Oxford: OUP.
Brants, K., & Van Praag, P. (2017). Beyond media logic. Journalism Studies, 18(4), 395-408. de Vreese, C. (2005). News Framing: Theory and Typology. Information Design Journal,
13(1), 51-62.
Entman, R. M. (1993). Framing: Toward Clarification of a Fractured Paradigm. Journal of
Communication, 43 (4), 51–58.
Esser, F., Stépinska, A., & Hopmann, D. N. (2016). Populism and the media. Cross-national findings and research perspectives. In T. Aalberg, F. Esser, C. Reinemann, J. Strömbäck & C. d. Vreese (Eds.), Populist political communication in Europe (pp. 365-380). London: Routledge.
Gäbler, B. (2018). AfD und die Medien: Erfahrungen und Lehren für die Praxis [The AfD and the media: Experiences and lessons for the field]. Otto-Brenner-Stiftung Arbeitsheft, 95, 1-124.
Greenwald, A. G., McGhee, D. E., & Schwartz, J. K. L. (1998). Measuring individual differences in implicit cognition: The implicit association test. Journal of Personality and
Social Psychology, 74, 1464-1480.
Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding and using the Implicit Association Test: I. An improved scoring algorithm. Journal of Personality and Social
Psychology, 85, 197–216.
Gross, K. (2008). Framing persuasive appeals: Episodic and thematic framing, emotional response, and policy opinion. Political Psychology, 29(2), 169–192.
Hameleers, M., Bos, L., & de Vreese, C. H. (2016). “They did it”: The effects of emotionalized blame attribution in populist communication. Communication Research, 44(6), 870–900. doi:10.1177/0093650216644026
Hameleers, M., Bos, L., & de Vreese, C. H. (2017). Shoot the messenger? The media’s role in framing populist attributions of blame. Journalism, 1-20.
Hameleers, M., Bos, L., & de Vreese, C. H. (2018). Framing Blame: towards a better understanding of the effects of populist communication on populist party preferences.
Journal of Elections, Public Opinion and Parties, 23(3), 380-398.
Harcup, T. & O’Neill, D. (2017). What is News? News values revisited again. Journalism
Studies, 18(12), 1470-1488.
Holm, E. M. (2012). Emotions as mediators of framing effects. Aarhus, Denmark: Politica. Kieslich, K., & Marcinkowski, F. (2019). Involuntary Media Populism – How German
Mainstream Media Inadvertently Facilitate Populist Movements by Evoking Fear and Anger (Paper prepared for The European Consortium for Political Research General Conference 2019). Heinrich-Heine Universität Düsseldorf, Düsseldorf.
Krämer, B. (2014). Media Populism: A Conceptual Clarification and Some Theses on its Effects. Communication Theory, 24, 42–60.
Kroh, M. (2007). Measuring Left-Right Political Orientation: The Choice of Response Format.
The Public Opinion Quarterly, 71(2), 204-220.
Kühne, R. (2012). Media-Induced Affects and Opinion Formation: How Related And Unrelated Affects Influence Political Opinions. Living Reviews in Democracy, 3, 1–20. Kühne, R. & Schemer, C. (2013). The Emotional Effects of News Frames on Information
Processing and Opinion Formation. Communication Research, 42(3), 387–407.
Ladd, J. M., & Lenz, G. S. (2008). Reassessing the role of anxiety in vote choice. Political
