Tainted Love: How Stigmatization of a Political Party in News Media Reduces Its Electoral Support

Tainted Love: How Stigmatization of a Political Party in News Media Reduces Its Electoral Support Abstract In contemporary democracies, a political party typically needs good press to attract voters. A glum scenario for a party would be that news media systematically stigmatize it. To what extent does stigmatization lower its electoral support? This article examines voters’ reactions to news media coverage of the Dutch party PVV. Using a media content analysis linked to a 2014 panel survey of a sample representative of the Dutch electorate, we find that, among voters who held anti-immigrant attitudes, exposure to stigmatization lowered these voters’ perceived legitimacy of the PVV. This, in turn, decreased their propensity to vote for that party. This suggests that stigmatization can be a strong tool in the hands of those who intend to damage a party. News media can be biased in various ways. What are the electoral effects of media bias? This important question has hardly ever been satisfactorily addressed. This is largely because media bias effects research suffers from the well-known subjectivity problem (Groeling, 2013). This article investigates effects of media bias while minimizing problems related to subjectivity. The subjectivity problem refers to the notion that each voter will have their own perception of a news item being biased or not. Partisans may “sincerely perceive news as being biased against their preferred stance, even when it is actually unbiased” (Groeling, 2013, p. 139) because of well-known cognitive biases—or because of the reputation of a particular media source. In this article, we solve this problem by examining indisputable stigmatization. We thus look for terms that voters should consider extremely negative connotations no matter their background. We find a Dutch political party that in 2014 was suddenly consistently linked to terms closely associated with the embodiment of pure evil. Or, we should say, what in that particular context is universally considered as such, Hitler’s Third Reich (van Heerden and van der Brug, 2017). We focus on the Freedom Party (PVV) in the 2014 European Parliamentary elections in The Netherlands. This provides an excellent case for studying electoral effects of stigmatizing a party, as the PVV suddenly changed from principled shunning of National Front (FN) in France, Flemish Interest (VB) in Belgium, and Freedom Party of Austria (FPÖ) to close cooperation with these parties. In response, various news media repeatedly associated the PVV with alleged characteristics of FN, VB, and FPÖ—including fascism and neo-Nazism. To what extent, how, and among which voters did being systematically associated with fascism and neo-Nazism affect decisions to vote for the PVV? We exploit variation in exposure of individual voters to such associations to address these questions. The 2014 European Parliamentary election campaign was chosen as the time under study, as that was the first time that voters massively tuned in after the party’s change of course. In doing so, this study contributes to our understanding of the electoral effects of news media bias in established democracies. Furthermore, we go beyond many previous media influence studies by taking into account voter heterogeneity and the mechanism underlying the electoral effect under study by modeling moderation and mediation effects. Also, we improve on much earlier research on political parties by bringing in concepts from psychology/sociology (stigmatization) and communication science (attribute framing; protest paradigm), and linking them to concepts used in political science (perceived party legitimacy; anti-immigrant attitude) using a communication science model (the Differential Susceptibility to Media Effects Model [DSMM] model). We explain our comprehensive approach in the next paragraph. Theoretical Model Media bias has predominantly been studied in the U.S. context (D’Alessio & Allen, 2000; Groeling, 2013). No research exists on media bias vis-à-vis anti-immigration parties—parties that are absent from the U.S. context. Although it is always difficult to identify a relevant benchmark against which to measure bias (Entman, 2007; Hopmann, van Aelst, & Legnante, 2012), anti-immigration parties such as the PVV, FN, VB, and FPÖ seem likely candidates to find bias against. This is because in many ways, these parties challenge the political status quo, and several scholars have argued that political challengers of the status quo generally fight an uphill battle. “Effective political action is likely when it does not disturb power, income, or status hierarchies” (Edelman, 1977, p. 141). In political discourse, “[p]erspectives that challenge the status quo are not accorded the legitimacy that would make them subjects of serious discussion” (Michael Lipsky, in Edelman, 1977, p. xvii). Indeed, the news media are often said to be “agents of social control” (Gans, 2004 [1979], p. 295). Two main criticisms are fielded here. First, the news media are said to silence challengers by hardly paying any attention to them. Second, they are accused of ridiculing and stigmatizing challengers. Concerning silence, Bennett (1990, 2016) argues that the news media engage in “indexing.” Indexing is “the tendency of mainstream news organizations to index or adjust the range of viewpoints in a story to the dominant positions of those whom journalists perceive to have enough power to affect the outcome of a situation” (Bennett, 2016, p.16). As long as challengers such as anti-immigration parties are small and not in power, their viewpoints will hardly be taken into account—if at all. Analogous to observations by Molotch (1979) about social movements, we can state that challengers of the political status quo are nothing like standard official sources from which anything is by default fit to print. In contrast with news from established actors, news from social movements is not “prima facie interesting, important, and defensible to work supervisors as worthy of publication” (Molotch, 1979, p. 77). Cohen (1972), in contrast, states that “(t)he mass media, in fact, devote a great deal of space to deviance” (p. 17). Similarly, Paletz and Entman (1981, pp. 125–126) claim that “marginal groups do not go entirely uncovered by the mass media.” This said, “media coverage of various groups is drastically and dramatically different,” some groups being “scorned as pariahs” (p. 124). Similarly, Paletz and Entman (1981, p. 127) contend that “dissident or unconventional voices are sometimes heard. But they are usually treated in a way that deprives them of their eloquence and force; their threat to the larger distribution of power is muted.” Shoemaker (1984) also claims that deviant groups are not less prominent in the news—but they are treated less favorably (p. 66). This brings us to ridicule and stigma. If not by way of silencing them, new political groups may be marginalized by way of ridiculing or stigmatizing them (Ferree, 2005; Linden & Klandermans, 2006; van Zoonen, 1992). Indeed, anti-immigration parties may be reported on through the “protest paradigm” (Chan and Lee, 1984). The protest paradigm “is one framework that media scholars have used to systematically understand the specific type of frames that news media often use to weaken legitimacy, obscure a protest’s social/political concerns, or both” (Weaver & Scacco, 2013, p. 64). When an event is covered through the protest paradigm, the organizing political actor’s central concerns are ignored. Instead of on substantive matters underlying the event, the coverage focuses on particularities of the event, its participants, or their actions (Boykoff, 2006; Gitlin, 1980), for instance, a participant’s funny banner, a clumsy comment, or a peculiar clothing style, or, as another example, the one violent action in a generally peaceful and quiet demonstration, or the reactions of (often uninformed) bystanders to disruptive aspects of the event. This, the argument goes, leads to trivialization, marginalization, and often criminalization of the political actor. The paradigm commonly pertains to social movements. However, it may well be applicable to political parties. This said, the paradigm may not translate automatically or without systematic differences to coverage of social movements. An indication for this is that Lee (2014) finds fewer references to violence and disruption in news media coverage, and more quotes of protester sources when political matters are addressed. When applying the protest paradigm to anti-immigration parties, one would expect not only an emphasis on the deviant appearances of anti-immigration activists and politicians (e.g., PVV leader Geert Wilders’s bleached hair), but also (in an analogy to the criminalization of social movements) the portraying of these actors as evil. In other words, one would expect the stigmatization of these parties. Coverage of Anti-Immigration Parties Only few previous studies exist on the news media coverage of anti-immigration parties. They ascribe some importance to the role of the mass media in the emergence of these parties. Mazzoleni (2003), for example, writes that “(t)he mass media themselves may also be “players” in the political game, endorsing or opposing populist stances and policies, both intentionally and unintentionally” (p. 2). Some studies suggest that these parties benefit electorally from the media visibility of their leaders (Vliegenthart, Boomgaarden, & van Spanje, 2012) and from the visibility of their policy issues in the media (Boomgaarden & Vliegenthart, 2007; Walgrave & De Swert, 2004). Other studies have investigated how the parties themselves were covered by the media (Art, 2006; Ellinas, 2010; Mazzoleni, 2003; Schafraad, d’Haenens, Scheepers, & Wester, 2012). Among the few anti-immigration parties that have been studied are the FPÖ, the Northern League (LN) in Italy, the Front National (FN) in France, the Republicans (REP) in Germany, and the Flemish Interest (VB) in Belgium. In all five cases, media actors reacted strongly to the emergence of these parties. In Austria (Art, 2006, pp. 189–191; Plasser & Ulram, 2003, pp. 38–39) and Italy (Biorcio, 2003, pp. 78–88), some media outlets reacted with stigmatization, whereas others supported the anti-immigration parties. In contrast, anti-immigration parties in France (Birenbaum & Villa, 2003, pp. 51–52), Germany (Art, 2006, pp. 165–166), and Belgium (Schafraad et al., 2012, p. 373–374) were often stigmatized by all major media outlets. These patterns of reactions persisted for decades on end, suggesting that “culturally rooted frames and editorial routines prevent” media from changing their approach to new political movements such as anti-immigration parties (Schafraad et al., 2012, p. 373). This is in line with the literature on new left movements (Gamson & Modigliani, 1989; Gitlin, 1980; van Zoonen, 1992). Coverage of Anti-Immigration Parties and Its Effects Effects of election news coverage on party choice have rarely been studied. Findings relate to both visibility and tone of news coverage of parties affecting voters’ decision at the ballot box (Hopmann, Vliegenthart, De Vreese, & Albaek, 2010; Kleinnijenhuis, Hoof, Oegema, & De Ridder, 2007; Norris, Curtis, Sanders, Scammell, & Semetko, 1999; Oegema & Kleinnijenhuis, 2000; van Spanje & De Vreese, 2014). Effects of media bias are found in the U.S. context (Druckman and Parkin, 2005). None of these are studies of anti-immigration parties, however. Another study examined individual-level effects of coverage of an anti-immigration party leader (Bos, van der Brug, & De Vreese, 2011). In that study, the dependent variable is not party choice but the perceived effectiveness and legitimacy of the leader. To our knowledge, only one study has tested individual-level effects of news media coverage on voting for anti-immigration parties. That study, conducted by Burscher, van Spanje, & De Vreese (2015), included 17,014 respondents, 20,084 news items, and 13 anti-immigration parties in 11 countries. It showed that exposure to immigration and crime news enhanced voters’ propensity to vote for an anti-immigration party. Just as that article, the present study examines media effects. Concerning media effects, perhaps the most elaborate analytical model is Valkenburg and Peter’s (2013) DSMM. We link this model to several strands of the literature in this article. We then apply it to the case of the PVV in 2014, following various other applications (Beyens, Vandenbosch, & Eggermont, 2015; Beullens & van den Bulck, 2013). The DSMM consists of three elements relevant here. These are media use, media effects, and susceptibility factors (cf. Valkenburg & Peter, 2013). Media use is the “use of media types, content and technologies” (Valkenburg & Peter, 2013, p. 222). Media effects are “within-person changes” in, for example, attitudes and behavior “that result from media use” (p. 222). Media effects last beyond the media use situation (p. 224). Susceptibility factors include “all person dimensions that predispose the selection of and responsiveness to media” such as political attitudes (Valkenburg & Peter, 2013, p. 226). In this article we argue, and demonstrate empirically, that particular media use leads to particular media effects, given a particular susceptibility factor. More specifically, we show that a voter’s exposure to stigmatization of the PVV (media use) lowers her propensity to vote for that party (media effect) if she holds a negative attitude toward immigrants (susceptibility factor). Below, we elaborate on each of these three elements in turn. Media use Stigmatization of a party starts with attributing a label to that party (just as, for instance, the “protest paradigm” can be seen as a set of frames that can be applied to a new social movement—Weaver & Scacco, 2013). “Stigma exists when elements of labelling, stereotyping, separation, status loss, and discrimination occur together in a power situation that allows them” (Link & Phelan, 2001, p. 377). Seen in this light, stigmatization of a party involves—among other things—making particular compromising aspects of that party more salient than others. What does this mean for the case of the PVV? Justified or not, the PVV is lumped into the same category as FN, VB, and FPÖ (labeling), a category that is referred to as “fascist,” “neo-Nazi,” or the like (stereotyped), which sets the PVV apart from “normal” or “democratic” parties (separation), leading to lower utility for voters of a vote for the PVV (status loss) and thus a lower propensity to vote for the PVV (discrimination). The news media operate in a contextual “power situation” that allows them to categorize the PVV in this way. Media effects Here, we have entered the realm of media effects. The literature on voting for anti-immigration parties is relevant here. Anti-immigration party voting has been researched from various angles. Factors that play a role in voting for an anti-immigration party (see van der Brug & Fennema, 2007, for an overview) include factors related to news media (Boomgaarden & Vliegenthart, 2007; Walgrave & De Swert, 2004). Mediating factors include voters’ perception of how “legitimate” an anti-immigration party is—that is, whether the party is nonviolent and democratic (Bos & van der Brug, 2010). This links to the argument that many voters are “of two minds about immigration and minority politics and that this internal conflict affects their expressed policy preferences and vote choices in predictable ways” (Blinder, Ford, & Ivarsflaten, 2013, pp. 841–842). On the one hand, they oppose immigration. On the other, they subscribe to anti-prejudice, pro-democracy, and anti-violence norms, deeming, e.g., neo-Nazism and fascism unacceptable. Hence, associating an anti-immigration party with neo-Nazism or fascism may reduce its support—through decreasing its legitimacy. Susceptibility factor Obviously, we do not expect all voters to be equally affected. Electoral decisions of voters who agree with the PVV will clearly be more likely to be influenced than voters whose views are radically different from the party’s views. Not surprisingly, a particularly strong predictor of voting for this anti-immigration party is attitudes toward immigrants (Lubbers, Gijsberts, & Scheepers, 2002). We hypothesize that the more a voter holds anti-immigrant attitudes, the greater the media impact on her voting behavior will be. This is because voters who disagree with the PVV tend to be stable in their rejection of the party regardless of how it is portrayed or treated (van Spanje & Weber, 2017). Among anti-immigrant voters, in turn, one might suspect that moderately sophisticated voters are most susceptible to political persuasion, as Converse (1962) argues and Zaller (1989) demonstrates is often the case. This latter step, however, we cannot adequately test given the limits to our data. Thus, we have only anti-immigrant attitude as a “differential-susceptibility variable” that acts as a moderator of the effect of media use on the mediating media effect and as a moderator of the second-order effect of media use (Valkenburg & Peter, 2013, p. 231). One may argue that the DSMM has a broader scope, also looking at selective exposure, for example. In this article, we hold within-person time-invariant factors constant during a 2-month period. This means that the characteristics of the voter and her environment are kept constant, as they are not expected to change within that window. At the same time, this enforces a focus on the three factors that are expected to change in the short run: exposure to stigmatization, perceived legitimacy, and voting behavior. Figure 1 reflects the analytical framework used in this article. Figure 1 View largeDownload slide The analytical model Figure 1 View largeDownload slide The analytical model As shown in Figure 1, we hypothesize that individual voters’ exposure to stigmatization of the PVV in the news media negatively affects that party’s perceived legitimacy among these voters, which in turn should decrease their propensity to vote for the party. The stronger their anti-immigrant attitude is, the larger these effects are expected to be. Stigmatization and Demonization Each political party arguably has an infinite number of characteristics that could be made salient. Earlier research concentrated on various kinds of aspects that media may make salient such as party cohesion versus party discord (Groeling, 2010) or a party's political leader and his character (Iyengar and Kinder, 1987). In our particular example, the aspect of the PVV that is made salient is its cooperation with FN, VB, and FPÖ. This involves associating the PVV with these three parties, labeling the category of these parties, e.g., “extreme right,” stereotyping them, and setting them apart from other parties. Among consumers of that media content, this may lead to status loss of the PVV and discrimination against it. “Stigma involves status loss—a downward placement in the status hierarchy” (Link & Phelan, 2001, p. 379). The PVV drops in the hierarchy of parties in a voter’s mind, leading to a lower utility associated with a vote for that party and therefore a lower propensity to vote for it. Concerning the labeling, van Heerden (2014) proposes the concept of “demonization.” She defines demonization as “portraying an actor as the embodiment of absolute evil” (2014, p. 10). Demonization is a more specific concept than the labeling element of stigmatization (p. 16). The implicit consequence of demonization is, as van Heerden points out, more far-reaching than any implicit consequence of stigmatization: after all, anyone would agree that any “absolute evil” should be eradicated (p. 16). This sheer unanimity of perception gets us, for all intents and purposes, around the problem of subjectivity in media bias research. In the remainder of this article, we thus use demonization as a concept in the labeling of a party. In addition, we use the broader concept of stigmatization in this article, referring to not only the labeling of a party but also the stereotyping, separation, status loss, and discrimination. One might claim that in the case under investigation, the labeling is only indirect. On this view, it is not the PVV that is labeled. Instead, the PVV is associated with parties that are labeled. Yet, we hold that even in that case of such “labelling by proxy” there is still labeling, albeit indirect. The labeling of parties associated with the party arguably leads to stigmatization of that party itself, using the analogy of “guilt by association.” It is the PVV that faces the consequences of status loss and discrimination—as we will see. In fact, the indirectness of the labeling is a reason to consider the PVV in 2014 a “least likely case” in terms of electoral effects of stigmatization. After all, if even indirect labeling damages a party, the effect of such labeling should be considerable.1 Another reason is that PVV’s only member Wilders had had his own successful political movement for 10 years at the time of our study, which arguably made it difficult for others to have voters profoundly rethink their image of the party. That we find that this party image was nonetheless tarnished is yet another indication of the strength of the forces that stigmatization may unleash. Hypotheses On the basis of the theoretical considerations stated above, we formulate four hypotheses. Our first hypothesis is that exposure to news media content demonizing the PVV reduces a voter’s propensity to vote for that party (H1). We base this expectation on literatures on stigmatization and on the protest paradigm. Effects of media coverage of parties on voting are largely unknown. However, it is found in other contexts that exposure to news framing can affect voting behavior (Shah, Domke, & Wackman, 1996; van Spanje & De Vreese, 2014). Another indication is that visibility and tone of news media coverage of a party have been demonstrated empirically to influence voting for that party, a negative tone reducing individuals’ inclination to vote for the party (Hopmann, Vliegenthart, De Vreese, & Albaek, 2010). More specifically about stigmatization, the status loss and discrimination that demonization should bring along manifest themselves in lower propensity to vote for the party. We argue that this negative effect is because of voters “learning” about parties through exposure to attribute framing. Through exposure to information about attributes of a subject (e.g., an association of a political party to known extremist parties), this attribute will become more salient in the voter’s mind, promoting an evaluation of the subject using this particular attribute (see Levin, Schneider, & Gaeth, 1998). Several studies found evidence that voters use the information from the framed attribute to form political opinions (Nelson, Clawson, & Oxley, 1997, Nelson & Oxley, 1999) and in making (voting) decisions (Hardisty, Johnson, & Weber, 2010; Shah et al., 1996, van Spanje & De Vreese, 2014). Given that fascism and neo-Nazism are perceived as evil, the association with these terms resonates with the idea to reject such parties, which consecutively leads to a decrease in the propensity to vote for the associated party (cf. Schemer, Wirth, & Matthes, 2012). Our second hypothesis is that a voter’s perceived legitimacy of the PVV mediates the effect of exposure to demonization on her propensity to vote for that party (H2). More specifically, exposure is expected to negatively affect PVV legitimacy, which should in turn positively influence PVV vote propensity (cf. Bos & van der Brug, 2010). Exposure is expected to negatively affect PVV legitimacy because of the particular stigma involved. If a party is labeled “neo-Nazi” or “fascist,” its viability as an option in a democracy is clearly in question. Given the overwhelming support for democracy as a political regime in contemporary established democracies, this in turn decreases the party’s electoral attractiveness. This reasoning is in accordance with the empirical findings by Bos and van der Brug (2010) suggesting a considerable positive effect of a party’s perceived legitimacy on propensity to vote for that party. Our third hypothesis is that a voter’s attitude toward immigrants strengthens the effect of media use on PVV legitimacy (H3). PVV leader Wilders often targets immigrants using strongly negative words, for instance, when he proposed a “head rag tax” on headscarves, following the principle of “the polluter should pay” (Korteweg & Yurdakul, 2014, p. 120). Voters who have a positive attitude toward immigrants would not find the PVV legitimate to begin with—its association with FN, VB, or FPÖ not mattering much. Voters who hold a negative attitude toward immigrants may find the PVV legitimate yet deem neo-Nazism and fascism beyond the pale. Among these anti-immigrant voters, associations of the PVV with neo-Nazism and fascism are thus expected to reduce the party’s legitimacy. Our fourth hypothesis is that a voter’s attitude toward immigrants also strengthens the effect of PVV legitimacy on PVV vote propensity (H4). A negative attitude toward immigrants is a strong predictor of voting for a party that campaigns against immigration (van der Brug & Fennema, 2007), as the PVV does. Voters with positive attitudes toward immigrants are unlikely to vote for the PVV, no matter how legitimate they think the PVV is. Among voters with negative attitudes toward immigrants, in contrast, their likeliness to vote for the PVV should decrease when their perception is that the party is not legitimate anymore. Note that we do not in any way imply that the PVV in 2014 was a passive victim of stigmatization. In fact, party leader Wilders must have known well what he called onto himself and his party when he decided to start close cooperation with FN, VB, and FPÖ. The common assumption in the literature that the stigmatized is a passive victim is unsubstantiated (Link & Phelan, 2001, p. 365; Major & O’Brien, 2005, p. 411). That also holds in the case of the PVV—as we will see in the next section. The Case MP Wilders founded the PVV in 2006. Campaigning mainly on an anti-immigration and anti-Islam platform, the PVV has received between 5.9 and 15.5% of the vote in general elections. PVV MPs’ controversial public statements triggered accusations of fascism and also led to prosecution for hate speech (see van Spanje and De Vreese, 2015). Wilders has always tried to steer clear of fascism, neo-Nazism, or any ideologies that are clearly beyond the pale in contemporary Western Europe. An important element in this strategy is his distancing himself from other anti-immigration politicians. In 2009, he refused to join FN, VB, FPÖ, and a few other parties in talks about cooperation in the European parliament. However, Wilders suddenly and radically changed his stance toward these parties, trying to form a European parliamentary group with them in 2014. From that moment onward, the PVV has been repeatedly associated with these parties. Willingly or unwillingly, this has led to demonizing associations of the PVV. The election campaign in the run-up to the 2014 European Parliamentary election was the first campaign after the party’s change of course, and thus the first time that voters en masse followed politics again. In this article we use a panel survey during that election campaign, studying variation in voters’ exposure to demonizing associations of the PVV and its effects on their propensity to vote for that party. Methods To address our research question, we assess the four hypotheses using media content data linked to panel survey data (cf. Dilliplane, Goldman, & Mutz, 2013). A four-wave panel survey was collected within the framework of the 2014 European Election Campaign Study (De Vreese, Azrout, & Möller, 2014), of which the last three waves contained measures of variables of interest to this study (the moderator, anti-immigrant attitude, was measured in Wave 2; all other variables were measured in Waves 3 and 4). Field work took place from March 20 to 30, 2014 (Wave 2), April 17 to 28, 2014 (Wave 3), and from May 26, 2014 to June 2, 2014 (Wave 4). The time that the survey was fielded was short, as 90% of responses were in within 4–6 days in each wave. Fieldwork was conducted by TNS NIPO, a research institute that complies with ESOMAR guidelines for survey research. A random sample was drawn from the TNS NIPO database (which consists of 200,000 nationally representative individuals recruited through multiple recruitment strategies, including telephone, face-to-face, and online recruitment), with quotas enforced on age, gender, and education. This led to a net sample of 1,267 voters, from whom we have full data on all relevant variables.2 The media content data were constructed using all the media coverage in which the PVV was mentioned in the eight main national newspapers3 and the main national news site (www.nu.nl) between the second and the fourth wave. The articles were collected through the LexisNexis Academic data set, including all articles that mention either the party or its leader (Nprior Wave 3 = 286, Nprior Wave 4 = 560).4 Operationalizations Demonization of the PVV in the media is measured by coding per news item the extent to which a party that the PVV cooperates with is portrayed as “fascist,” “racist,” “(neo-)Nazi,” “extreme right,” “radical,” “discriminating,” or “anti-Semite.” We use these seven particular labels because we aim to capture the specific “demonization” labeling. In the context of The Netherlands in 2014, the moral scheme of WWII is still dominant (van Heerden & van der Brug, 2017, p. 37). As a consequence, Nazism and fascism are widely considered political regimes that represent absolute evil (cf. van Heerden, 2014, p. 14). Van Heerden thus operationalizes demonization within that context as “portraying a political actor as the embodiment of Nazism/fascism” (p. 14). We broaden this category from two to seven labels to also include references to “antisemitism,” “racism,” “discrimination,” “radical right,” and “extreme right,” which should be placed in the same narrow category of connotations with the Third Reich, responsible for the culmination of absolute evil, the Holocaust. In the contemporary Dutch context, there is a clear distinction between the seven labels mentioned and other labels. As van Heerden and van der Brug (2017, p. 37) note “While the atrocities of the Nazi’s are in terms of the number of victims comparable to those of other authoritarian regimes…, adherents of those ideologies are not prosecuted in the same way as fascists.” Other labels, such as “populist” or “eurosceptic,” may not necessarily be demonizing in voters’ eyes. The same goes for “anti-establishment” or “anti-Islam.” In the eyes of many voters, these labels have nothing negative or delegitimizing. Indeed, many Dutch voters may describe themselves as anti-establishment or anti-Islam at the moment. However, no Dutch voter in her right mind would openly call herself “fascist,” “(neo-)Nazi,” “racist,” “radical right,” “discriminating,” “extreme right,” or “anti-Semite” because if she did, she would immediately become an outcast (cf. Tillie, 2008; van Heerden, 2014). These “demonizing” labels can be seen as clearly delegitimizing, and therefore potentially damaging a party, across the board. From the articles mentioning the PVV and/or Wilders, we automatically selected those articles that mentioned a party that the PVV cooperates with5 and contained one of the demonizing words.6 To make sure that the co-occurrence of these mentionings indeed mean that the article links the PVV to this demonized cooperating party, we also manually coded these articles.7 In 13 of the 77 articles in which the combination of the PVV, a cooperating party, and a demonizing word was made, the author did either not link the PVV to the other party or did not use the demonizing word to describe the cooperating party. Only the remaining 64 news articles are therefore coded as demonizing the PVV. We divide the number of articles in which the PVV was demonized by the total number of articles in which the PVV is mentioned to create a (relative) media demonization scale. To correct for the different length of the time (i.e., the different numbers of editions of the newspapers), we also divided the relative score by the number of newspaper editions in each period. See Table 1 for the frequency of the occurrence of the PVV and demonization of that party, split by the periods before survey Wave 3 and between Waves 3 and 4. Table 1 Frequency of the Occurrence of PVV and of Demonization of the Party Before survey Wave 3 Between survey Waves 3 and 4 N PVV N demon -ization Relative demon -ization N PVV N demon -ization Relative demon -ization 1 De Telegraaf 35 2 0.06 93 17 0.18 2 NRC Handelsblad 43 1 0.02 82 7 0.09 3 NRC.NEXT 18 0 0.00 27 3 0.11 4 Trouw 53 0 0.00 96 11 0.11 5 AD 36 0 0.00 72 4 0.06 6 De Volkskrant 51 1 0.02 117 6 0.05 7 Metro 24 0 0.00 25 3 0.12 8 Spits 13 1 0.08 11 1 0.09 9 www.nu.nl 13 1 0.08 37 6 0.16 Total 286 6 0.02 560 58 0.10 Before survey Wave 3 Between survey Waves 3 and 4 N PVV N demon -ization Relative demon -ization N PVV N demon -ization Relative demon -ization 1 De Telegraaf 35 2 0.06 93 17 0.18 2 NRC Handelsblad 43 1 0.02 82 7 0.09 3 NRC.NEXT 18 0 0.00 27 3 0.11 4 Trouw 53 0 0.00 96 11 0.11 5 AD 36 0 0.00 72 4 0.06 6 De Volkskrant 51 1 0.02 117 6 0.05 7 Metro 24 0 0.00 25 3 0.12 8 Spits 13 1 0.08 11 1 0.09 9 www.nu.nl 13 1 0.08 37 6 0.16 Total 286 6 0.02 560 58 0.10 Table 1 Frequency of the Occurrence of PVV and of Demonization of the Party Before survey Wave 3 Between survey Waves 3 and 4 N PVV N demon -ization Relative demon -ization N PVV N demon -ization Relative demon -ization 1 De Telegraaf 35 2 0.06 93 17 0.18 2 NRC Handelsblad 43 1 0.02 82 7 0.09 3 NRC.NEXT 18 0 0.00 27 3 0.11 4 Trouw 53 0 0.00 96 11 0.11 5 AD 36 0 0.00 72 4 0.06 6 De Volkskrant 51 1 0.02 117 6 0.05 7 Metro 24 0 0.00 25 3 0.12 8 Spits 13 1 0.08 11 1 0.09 9 www.nu.nl 13 1 0.08 37 6 0.16 Total 286 6 0.02 560 58 0.10 Before survey Wave 3 Between survey Waves 3 and 4 N PVV N demon -ization Relative demon -ization N PVV N demon -ization Relative demon -ization 1 De Telegraaf 35 2 0.06 93 17 0.18 2 NRC Handelsblad 43 1 0.02 82 7 0.09 3 NRC.NEXT 18 0 0.00 27 3 0.11 4 Trouw 53 0 0.00 96 11 0.11 5 AD 36 0 0.00 72 4 0.06 6 De Volkskrant 51 1 0.02 117 6 0.05 7 Metro 24 0 0.00 25 3 0.12 8 Spits 13 1 0.08 11 1 0.09 9 www.nu.nl 13 1 0.08 37 6 0.16 Total 286 6 0.02 560 58 0.10 We link the demonization of the PVV in the media to the survey data by a self-reported media exposure measure. Media exposure was gauged in both Waves 3 and 4 by asking how many days in a normal week the respondent reads “the following” newspapers, followed by a list of eight national newspapers in randomized order. For each of the newspapers, a respondent could indicate that she read it anywhere from 0 to 7 days a week. In addition, the respondent was asked to indicate how many days in a normal week she read about politics on the by far most often used Dutch news website, www.nu.nl, of which the content was coded. We consecutively construct a demonization exposure measure by multiplying the reported frequency (of the respondent’s use of a particular outlet) with the relative demonization (in that outlet) in the period preceding the wave in which the question was asked. Per respondent and per wave we add this for all newspapers together. This leads to a demonization exposure variable in which 0 means a respondent is not exposed to any demonization of the PVV (i.e., either the respondent does not use any of these media outlets or the respondent uses merely outlets that do not contain any demonization). The value of this demonization exposure variable increases as the respondent increases her use of an outlet that contains demonization, and also as the outlet the respondent uses increases demonization. Voters’ perception of legitimacy of the PVV is measured by reactions to four statements. These are “To what extent does the PVV comply with the laws, you think?”; “To what extent do you think the PVV has the right to get access to power?”; “To what extent does the PVV leadership respect the rules of our democracy, you think?”; and “To what extent do you think the PVV abides by the prevailing social norms in our society?” Answering options ranged from “not at all” (0) to “to a high degree” (6). This leads to a single scale (eigen value 3.31 and 83% explained variance) that is highly reliable (Cronbach’s α = .92). Voters’ propensity to ever vote for the PVV is gauged by the standard question common to many election surveys since van der Eijk and Niemöller (1984). In response to this question, voters can indicate the propensity that they “will ever vote for” a political party. They can do so on a scale varying from “very unlikely” (1) to “very likely”(10), with an explicit “don’t know” option. The question is designed to get the utility a voter would derive from voting for a party (van der Eijk, van der Brug, Kroh, & Franklin, 2006). Propensities to vote were asked for each of 11 relevant Dutch parties in random order. Obviously, the relevant one here is the propensity to vote for the PVV. Attitudes toward immigrants are measured using five items: “Immigrants abuse the Dutch social welfare system, as they take more out of it than they put in”; “Immigrants are a threat to the Dutch population’s security”: “The religious practices of immigrants enrich the Dutch way of life and traditions” (reversely coded); “Immigrants are an important cause of crime in the Netherlands”; and “Immigration is good for the Dutch labor market” (reversely coded). Answering options to these items, ranging from 0 (“strongly disagree”) to 6 (“strongly agree”), form a scale (1 factor with eigen value of 3.10, explained variance 62%) that is highly reliable (Cronbach’s α = .85). Data Analysis To test our hypotheses, we combine fixed-effects regression modeling with mediation (Baron & Kenny, 1986; Preacher, Rucker, & Hayes, 2007). Using fixed effects, we focus not on the between-subject variance to estimate effects but on the within-subject change across the waves (Allison, 2009). As a result, in a fixed-effects model, all time-invariant factors are implicitly controlled for (thus, explicitly controlling for them in the model is redundant). We combine the fixed effects with mediation to be able to test the indirect effect through legitimacy perceptions.8 We start with a baseline model, assessing the unmediated and unmoderated fixed effect of exposure to demonization on the propensity to vote (H1). Second, we assess the mediation by perceived legitimacy (H2) by testing the fixed effect of the demonization exposure on legitimacy perceptions (the independent variable should predict the mediator) and predicting propensity to vote with both demonization exposure and perceived legitimacy (the mediator should have a significant effect, while the effect of the independent variable should decrease compared with the initial model). To assess the moderation (H3 and H4), we add anti-immigrant attitude as well as its interaction with exposure to demonization and with perceived legitimacy. We focus on the direction and the significance (based on the Sobel test) of the coefficients to test our hypotheses, as well as on the marginal effects. Results We first turn to the unmediated and unmoderated effect of exposure to demonization on the propensity to vote for the PVV (H1). Model 1 in Table 2 shows that the coefficient (b = −1.82) is in the expected direction (increased exposure to demonization decreases the propensity to vote), significant at the p < .10 level (p = .094). To interpret the real-world implications of this coefficient, we assess what change on the propensity to vote this coefficient implies for the average observed change in exposure to demonization. On average, exposure to demonization increases with a score of 0.02, implying an average decrease of 0.03 on the (1–10) propensity to vote scale. In light of this marginal support and the modest real-world implication, we conclude that our first hypothesis is only partially supported. This is underwhelming evidence, which is partly because of the (also hypothesized) moderation effect of anti-immigrant attitude, as we will see. Table 2 Fixed-Effects Predicting Probability to Vote for the PVV 1 2 3 4 Exposure to demonization −1.82* −1.29 3.36 1.66 (1.38) (1.36) (3.38) (2.44) Legitimacy 0.28**** −0.13* (0.03) (0.06) Anti-immigrant attitude −0.00 −0.00 (0.01) (0.01) Stigma ** AIA −1.44** −0.78 (0.86) (0.85) Legitimacy ** AIA 0.11**** (0.02) R2 .004 .032 .006 .041 1 2 3 4 Exposure to demonization −1.82* −1.29 3.36 1.66 (1.38) (1.36) (3.38) (2.44) Legitimacy 0.28**** −0.13* (0.03) (0.06) Anti-immigrant attitude −0.00 −0.00 (0.01) (0.01) Stigma ** AIA −1.44** −0.78 (0.86) (0.85) Legitimacy ** AIA 0.11**** (0.02) R2 .004 .032 .006 .041 Note. Entries are unstandardized fixed-effects coefficients, with standard errors in parentheses. ****p < .001; *** p < .01; ** p < .05; * p < .1 (one-tailed). Table 2 Fixed-Effects Predicting Probability to Vote for the PVV 1 2 3 4 Exposure to demonization −1.82* −1.29 3.36 1.66 (1.38) (1.36) (3.38) (2.44) Legitimacy 0.28**** −0.13* (0.03) (0.06) Anti-immigrant attitude −0.00 −0.00 (0.01) (0.01) Stigma ** AIA −1.44** −0.78 (0.86) (0.85) Legitimacy ** AIA 0.11**** (0.02) R2 .004 .032 .006 .041 1 2 3 4 Exposure to demonization −1.82* −1.29 3.36 1.66 (1.38) (1.36) (3.38) (2.44) Legitimacy 0.28**** −0.13* (0.03) (0.06) Anti-immigrant attitude −0.00 −0.00 (0.01) (0.01) Stigma ** AIA −1.44** −0.78 (0.86) (0.85) Legitimacy ** AIA 0.11**** (0.02) R2 .004 .032 .006 .041 Note. Entries are unstandardized fixed-effects coefficients, with standard errors in parentheses. ****p < .001; *** p < .01; ** p < .05; * p < .1 (one-tailed). Before turning to moderation, we first assess the unmoderated indirect effect though perceived legitimacy. In the final table, Table 3, Model 5 shows that an increase in exposure to demonization decreases perceived legitimacy (b = −1.87, p = .011). Again, we interpret the real-world implication of this coefficient, and estimate that the average change in the exposure to demonization leads to a decrease of 0.03 on the (1–7) perceived legitimacy scale. Turning back to predicting the propensity to vote (back to Table 2), but now with both the independent variable and the mediator (see Model 2 in Table 2), we find that higher perceived legitimacy leads to higher propensity to vote (b = 0.28, p < .001), while the coefficient for exposure to demonization decreases compared with Model 1 and is not significant anymore (b = −1.29, p = .171). With an indirect effect estimated at −0.52 (se = 0.24, p = .014), these findings imply that we indeed have an indirect effect through perceived legitimacy, which supports H2. Table 3 Fixed-Effects Predicting Perceived Legitimacy of the PVV 5 6 Exposure to demonization −1.87** 6.37**** (0.81) (0.01) Anti-immigrant attitude −0.00 (0.01) Stigma ** AIA −2.29**** (1.45) R2 .004 .012 5 6 Exposure to demonization −1.87** 6.37**** (0.81) (0.01) Anti-immigrant attitude −0.00 (0.01) Stigma ** AIA −2.29**** (1.45) R2 .004 .012 Note. Entries are unstandardized fixed-effects coefficients, with standard errors in parentheses. ****p < .001; *** p < .01; ** p < .05; * p < .1 (one-tailed). Table 3 Fixed-Effects Predicting Perceived Legitimacy of the PVV 5 6 Exposure to demonization −1.87** 6.37**** (0.81) (0.01) Anti-immigrant attitude −0.00 (0.01) Stigma ** AIA −2.29**** (1.45) R2 .004 .012 5 6 Exposure to demonization −1.87** 6.37**** (0.81) (0.01) Anti-immigrant attitude −0.00 (0.01) Stigma ** AIA −2.29**** (1.45) R2 .004 .012 Note. Entries are unstandardized fixed-effects coefficients, with standard errors in parentheses. ****p < .001; *** p < .01; ** p < .05; * p < .1 (one-tailed). We now turn to the moderation by anti-immigrant attitude. We first hypothesized that this moderator would affect the effect of exposure to demonization on perceived legitimacy. As Model 6 in Table 3 shows, we find a significant interaction (p < .001) in the expected direction (b = −2.29). The negative sign implies that the more anti-immigrant a respondent is, the more negative the effect of exposure to demonization on exposure. To further explore the interaction, we plotted the marginal effect of exposure to demonization on perceived legitimacy for different values of anti-immigrant attitude (see Figure 2). We see a negative slope (as indicated by the negative coefficient of the interaction). The confidence interval shows that the marginal effect becomes significant for values of anti-immigrant attitude above 3.5, with an estimated coefficient of −7.36 (se = 1.45, p < .001) at the maximum value of anti-immigrant attitude. To interpret these effect sizes, we estimate the change in perceived legitimacy by the average change in exposure to demonization by different values of anti-immigrant attitude. We find an expected decrease in perceived legitimacy of 0.05 when anti-immigrant attitude is at 4, a decrease of 0.09 when anti-immigrant attitude is at 5, and a decrease of 0.13 when anti-immigrant attitude is at its maximum 6. This shows that individuals who are negative toward immigrants are affected by demonization in a way that is consistent with H3. Figure 2 View largeDownload slide Marginal effect of exposure to stigmatization on perceived legitimacy (x axis) for different values of anti-immigrant attitudes (y axis). The dashed line represents the 90% confidence interval Figure 2 View largeDownload slide Marginal effect of exposure to stigmatization on perceived legitimacy (x axis) for different values of anti-immigrant attitudes (y axis). The dashed line represents the 90% confidence interval H4 predicts that the effect of perceived legitimacy on propensity to vote is moderated by anti-immigrant attitude. As shown in Model 4 of Table 2, we find a significant positive coefficient (b = 0.11, p < .001). The sign indicates that the more anti-immigrant a respondent is, the more positive the effect of perceived legitimacy on the propensity to vote. Looking at the confidence interval of the marginal effects (Figure 3), we see that for anti-immigrant attitude values above 2, the legitimacy effect on propensity to vote is significant, and the marginal effect is increasing further as anti-immigrant attitude values increase further. H4 is supported. Figure 3 View largeDownload slide Marginal effect of perceived legitimacy on propensity to vote (x axis) for different values of anti-immigrant attitudes (y axis). The dashed line represents the 90% confidence interval Figure 3 View largeDownload slide Marginal effect of perceived legitimacy on propensity to vote (x axis) for different values of anti-immigrant attitudes (y axis). The dashed line represents the 90% confidence interval Given the support for H3 and H4, we can also assess to which degree the total effect and the indirect effects are moderated by anti-immigrant attitude. With regard to the moderation of the total effect, we find that this is significantly moderated (b = −1.44, p = .047, see Model 3 of Table 2). The marginal effect in Figure 4 shows that we find significant negative effects of exposure to demonization on propensity to vote only as anti-immigrant attitude values are above 4, with expected change in propensity to vote by the mean change in demonization exposure of −0.042 when anti-immigrant attitude is at 4, −0.068 when anti-immigrant attitude is at 5, and −0.093 when anti-immigrant attitude is at 6. This also explains why we find an only weak main total effect when testing H1. Figure 4 View largeDownload slide Marginal total effect of exposure to stigmatization on propensity to vote (x axis) for different values of anti-immigrant attitudes (y axis). The dashed line represents the 90% confidence interval Figure 4 View largeDownload slide Marginal total effect of exposure to stigmatization on propensity to vote (x axis) for different values of anti-immigrant attitudes (y axis). The dashed line represents the 90% confidence interval When adding the interaction between anti-immigrant attitude and perceived legitimacy to the model predicting propensity to vote (see Model 4 of Table 2), both the main and moderated effects of exposure to demonization fail to reach conventional levels of statistical significance. Furthermore, there is a significant marginal effect of the indirect effect through perceived legitimacy starting at anti-immigrant values of about 3.5, i.e., when we find both significant marginal effects for exposure to demonization on perceived legitimacy (anti-immigrant attitude values above 3.5) and significant marginal effects for perceived legitimacy on propensity to vote (anti-immigrant attitude values above 2), and the stronger the anti-immigrant attitude, the stronger the indirect path through perceived legitimacy.9 Conclusion To what extent, and how, does media bias have electoral impact? In this article, we have examined the systematic stigmatization of a party as a result of being demonized in various main news outlets—arguably an extreme case of news media reaction to a party. Examining such stigmatization of a political party as the embodiment of absolute evil circumvents most problems related to subjectivity that commonly plague media bias research. Although not all subjectivity problems are ruled out by this (e.g., PVV supporters may systematically more often miss the stigmatizing part of the message), the chance of diverging perceptions of a particular demonizing message is minimized. After all, the absolute evil is not subject to subjectivity. As it turns out, this stigmatization can manifest itself by negatively affecting voters’ perceptions of that party, and by reducing its electoral support. In the case under investigation, the Dutch PVV in the 2014 European election campaign, exposure to demonization has been demonstrated to lower the party’s perceived legitimacy. This reduced legitimacy in turn decreased the party’s support. These effects only occurred in the electoral pool of voters with strong anti-immigrant attitudes in which the PVV has always been fishing. We use a strong research design of a media content analysis linked to a voter panel survey, which is cutting edge in political communication research (Dilliplane, Goldman, & Mutz, 2013). Admittedly, the effect sizes are modest. But for practical reasons, we focused on newspapers, which enabled us to use automated content analysis, and newspapers are only a small part of voters’ news consumption. Given that we estimated the effect sizes from average newspaper usage, we expect that if, for instance, television news were included, implications would be more impressive. With regard to the average news consumption, we do need to note that we make use of a self-reported exposure measure, which is known to be associated with problems related to overreporting (cf. Prior, 2007). This said, we have been able to find consistent effects even in the presence of such measurement error. Furthermore, we find these effects in what we consider a least likely case. First, it is important to keep in mind that the demonization under investigation is an indirect one, as the PVV’s collaborators are labeled “Nazi” or “fascist”—not the party itself. Second, by 2014 the PVV was an established party that had been successful for years on end and about which most voters had made up their mind. The fact that we nonetheless find the predicted effects is telling. We speculate that this means that stigmatization of a party can be powerful when a new party is targeted that has little access to mass media. This is because a new party’s public image can still be shaped to a large extent by news media, and because a party without much mass media access cannot counteract the stigmatization by communicating its own side of the story to voters. However, let us emphasize once more that we have only investigated one case—and that, moreover, that case concerns a relatively new party. Other cases may yield effects that are different, or none at all. Furthermore, not only other cases but also other methods should be used, such as experimentation. This is because given the (survey) method used our possibilities to adequately control for rival explanations of the findings are limited. With this in mind, we turn to the implications of our study. Our research demonstrates the usefulness of modeling mediation and moderation effects, perhaps doing more justice to the complex realities than several previous studies. The DSMM model proves to be a particularly useful framework for doing so. By looking at a mechanism through which media affect vote choice (mediation), and by explicitly taking into account voter heterogeneity in our models (moderation), we go beyond the state of the art in studies of media coverage of anti-immigration parties, and in media effects. We integrate concepts used in political science (perceived party legitimacy; anti-immigrant attitude) into the DSMM framework. Similarly, the sociological trichotomy of silence, ridicule, and stigma (Ferree, 2005) seem to travel well to the realm of political parties, just as the protest paradigm concept (Chan & Lee, 1984)—at least, concerning the focus on the marginalization of a relatively new, outsider party by making stigmatizing aspects salient, and drawing attention away from its demands. Turning to our findings, they sit well with research on many voters being internally conflicted about anti-immigration parties (Blinder et al., 2013) and add to the existing knowledge on effects of media bias in measuring effects of undisputed negative media content about a political actor. The results suggest that effect sizes are modest. Yet, cumulative bias against a party may have a substantial negative impact—which brings us to the societal implications of our study. Our results may speak to societal debates on reactions to political extremism. The finding that systematically demonizing the PVV lowers its electoral appeal suggests that the party can be damaged electorally. Previous research has found that the PVV was hurt by such demonization, or, more precisely, its predecessor was (van Heerden & van der Brug, 2017). Together with our results, this suggests that media reactions are a potentially important weapon to fight the electoral rise of this party, and most likely its many counterparts across Europe as well. Of course, we do need to be careful here, as our study is limited to just one case (the PVV) at just one time point (in 2014). Furthermore, the stigmatization was indirect. Direct stigmatization may have a different effect because it may to a greater extent stick to the PVV. To be able to generalize these findings to direct stigmatization, or to other times and places, further research is encouraged. At the same time, not only anti-immigration parties can be targeted. All kinds of stigma are possible of all kinds of party. A striking empirical observation to illustrate this notion is that almost all Dutch parties have ever been called either “Nazi” or “fascist” in the news media, many of which regularly (van Heerden, 2014, p. 31). In most cases, such statements lack credibility and may therefore prove impotent in electoral terms. Yet, it is plausible to argue that many political ideas, or political parties carrying these ideas, could fall prey to a successful attempt to tarnish their reputation though news media. Joost van Spanje is an Associate Professor of Political Communication and Journalism as well as a member of ASCoR, University of Amsterdam. Rachid Azrout is a Lecturer of Political Communication and Journalism as well as a member of ASCoR, University of Amsterdam. Footnotes 1On the other hand, direct stigmatization may have a smaller effect, as an indirect message is more difficult to resist than direct stigmatization, which may be more easily dismissed by blaming it on political enemies or the news media being biased against a party. 2In the first wave, 2,189 respondents participated (AAPOR RR1 78.1%), 1,819 in the second wave (recontact rate 83.1%), 1,537 in the third wave (recontact rate 84.5%), and 1,379 in the fourth wave (recontact rate 89.7%). Because 112 respondents answered “don’t know” to the propensity to vote question in the third wave, in the fourth wave, or in both these waves, we are left with 1,267 respondents for whom we have full data on two time points. The samples show appropriate distributions in terms of gender, age, and education compared with census data. Panel attrition did not lead to a significant difference in the composition of the panel with regard to age and gender. The average level of education has slightly decreased between Wave 1 and Wave 4. As we are interested in underlying relationships and using fixed effects, we consider the deviations in the sample vis-à-vis the adult population less problematic. 3Telegraaf, NRC, NRC Next, Trouw, Algemeen Dagblad, De Volkskrant, Metro, and Spits. 4The LexisNexis Academic search string lists: “pvv” OR “partij voor de vrijheid” OR “wilders.” 5The LexisNexis Academic search string lists: “vlaams belang” OR “dewinter” OR “front national” OR “le pen” OR “lega nord” OR “bossi” OR “fpö” OR “fpo” OR “Strache” OR “Haider” OR “Zweden Democraten” OR “akesson.” 6The LexisNexis Academic search string lists: “extreemrechts!” OR “extreem-rechts!” OR “extreem rechts!” OR “discrimin!” OR “racis!” OR “radica!” OR “extremis!” OR “facis!” OR “neonazi” OR “neo-nazi” OR “anti-semi!” OR “antisemi!”. 7One coder coded the material and another a subsample (only between Waves 3 and 4, n = 69). Intercoder reliability was sufficient (91.3% agreement; Krippendorff’s α = .73). 8Although fixed effects are arguably the closest test to causality apart from experimental designs, we do need to note that in the absence of random assignment to the independent variable, our models are under the assumption that changes in political preferences (i.e., propensity to vote PVV) do not change media consumption behavior. One could, of course, argue that supporters of the PVV would choose to turn away from outlets that are critical of (i.e., demonize) the PVV. To address this concern, we tested whether propensity to vote for the PVV predicted change in media consumption, and found that this was not the case. Also, we ran our models not using our exposure to demonization measure but a raw exposure measure, and found that raw exposure has neither a main effect on propensity to vote for the PVV nor a moderated one. This shows that it is not just the outlet that voters use that matters, but the actual content voters are exposed to. Given that individual voters do not have any direct influence on news media content, these findings support our assumptions underlying our models (results available on request from the authors). 9To check the possibility that the effect is a knowledge effect, we rerun the analyses using knowledge (five multiple choice items) instead of anti-immigrant attitudes. With propensity to vote as the dependent variable and not controlling for legitimacy, the interaction between exposure and knowledge is not significant (b = 0.03, se = 0.91, p = .487). Having both anti-immigrant attitudes and knowledge as moderators leads to an insignificant interaction of knowledge (b = −0.13, se = 0.91, p = .445), while the interaction of anti-immigrant attitudes remains significant (b = −1.45, se = 0.86, p = .046). Although we cannot conclusively test this given data limitations, these findings imply that our results are not just the results of a knowledge effect. References Allison P. D. ( 2009 ). Fixed-effects regression models . Thousand Oaks, CA : Sage . Google Scholar Crossref Search ADS Art D. ( 2006 ). The politics of the Nazi past in Germany and Austria . New York, NY : Cambridge University Press . Baron R. M. , Kenny D. A. ( 1986 ). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations . Journal of Personality and Social Psychology , 51 , 1173 – 1182 . Google Scholar Crossref Search ADS Bennett W. L. ( 1990 ). Toward a theory of press-state relations in the United States . Journal of Communication , 40 , 103 – 127 . Google Scholar Crossref Search ADS Bennett W. L. ( 2016 ). News: The politics of illusion ( 10th ed. ). Chicago : University of Chicago Press . Google Scholar Crossref Search ADS Beullens K. , van den Bulck J. ( 2013 ). Predicting young drivers' car crashes: The role of music video viewing and the playing of driving games. Results from a prospective cohort study . Media Psychology , 16 , 88 – 114 . doi: 10.1080/15213269.2012.754582. Google Scholar Crossref Search ADS Beyens I. , Vandenbosch L. , Eggermont S. ( 2015 ). Early adolescent boys’ exposure to internet pornography. Relationships to pubertal timing, sensation seeking, and academic performance . The Journal of Early Adolescence , 35 , 1045 – 1068 . doi: 10.1177/0272431614548069. Google Scholar Crossref Search ADS Biorcio R. ( 2003 ). The Lega Nord and the Italian media system. In Mazzoleni G. , Stewart J. , Horsfield B. (Eds.), The media and neo-populism: A contemporary comparative analysis (pp. 71 – 94 ). Westport, CT : Praeger . Birenbaum G. , Villa M. ( 2003 ). The media and neo-populism in France. In Mazzoleni G. , Stewart J. , Horsfield B. (Eds.), The media and neo-populism: A contemporary comparative analysis (pp. 45 – 70 ). Westport, CT : Praeger . Blinder S. , Ford R. , Ivarsflaten E. ( 2013 ). The better angels of our nature: How the antiprejudice norm affects policy and party preferences in Great Britain and Germany . American Journal of Political Science , 57 , 841 – 857 . Boomgaarden H. , Vliegenthart R. ( 2007 ). Explaining the rise of anti-immigrant parties: The role of news media content in The Netherlands . Electoral Studies , 26 , 404 – 417 . Google Scholar Crossref Search ADS 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 . Google Scholar Crossref Search ADS Bos L. , van der Brug W. , De Vreese C. H. ( 2011 ). How the media shape perceptions of right-wing populist leaders . Political Communication , 28 , 182 – 206 . Google Scholar Crossref Search ADS Boykoff J. ( 2006 ). The repression of dissent . New York, NY : Routledge . Burscher B. , van Spanje J. H. P. , De Vreese C. H. ( 2015 ). Owning the issues of crime and immigration: The relation between immigration and crime news and anti-immigrant voting in 11 countries . Electoral Studies , 38 , 59 – 69 . Google Scholar Crossref Search ADS Chan J. M. , Lee F. L. F. ( 1984 ). Journalistic paradigms on civil protest: A case study of Hong Kong. In Arno A. , Dissanayake W. (Eds.), The news media in national and international conflict (pp. 183 – 202 ). Boulder, CO : Westview Press . Chong D. , Druckman J. ( 2007 ). Framing theory . Annual Review of Political Science , 10 , 103 – 126 . Google Scholar Crossref Search ADS Cohen S. ( 1972 ). Folk Devils and Moral Panics. The Creation of the Mods and Rockers . London : MacGibbon & Kee . Converse P. E. ( 1962 ). Information flow and the stability of partisan attitudes . Public Opinion Quarterly , 26 , 578 – 599 . Google Scholar Crossref Search ADS D’Alessio D. , Allen M. ( 2000 ). Media bias in presidential elections: A meta-analysis . Journal of Communication , 50 , 133 – 156 . Google Scholar Crossref Search ADS De Vreese C. H. ( 2012 ). New avenues for framing research . American Behavioral Scientist , 56 , 365 – 375 . Google Scholar Crossref Search ADS De Vreese C. H. , Azrout R. , Möller J. E. ( 2014 ). 2014 European Parliament Election Campaign Study: Data and Documentation . Amsterdam : University of Amsterdam . Della Vigna S. , Kaplan E. ( 2007 ). The Fox News effect: Media bias and voting . The Quarterly Journal of Economics , 122 , 1187 – 1234 . Google Scholar Crossref Search ADS Dilliplane S. , Goldman S. K. , Mutz D. C. ( 2013 ). Televised exposure to politics: New measures for a fragmented media environment . American Journal of Political Science , 57 , 236 – 248 . Google Scholar Crossref Search ADS Druckman J. N. , Parkin M. ( 2005 ). The impact of media bias: How editorial slant affects voters . The Journal of Politics , 67 , 1030 – 1049 . Google Scholar Crossref Search ADS Edelman M. ( 1977 ). Political language: Words that succeed and policies that fail . New York, NY : Academic Press . Ellinas A. A. ( 2010 ). The media and the far right in Western Europe. Playing the nationalist card . New York, NY : Cambridge University Press . Google Scholar Crossref Search ADS Entman R. M. ( 1993 ). Framing: Toward clarification of a fractured paradigm . Journal of Communication , 45 , 51 – 58 . Google Scholar Crossref Search ADS Entman R. M. ( 2007 ). Framing bias: Media in the distribution of power . Journal of Communication , 57 , 163 – 173 . Google Scholar Crossref Search ADS Ferree M. M. ( 2005 ). Soft repression: Ridicule, stigma, and silencing in gender-based movements. In Davenport C. , Johnston H. , Mueller C. (Eds.), Repression and mobilization (pp. 138 – 156 ). Minneapolis, MN : University of Minnesota Press . Gamson W. A. , Modigliani A. ( 1989 ). Media discourse and public opinion on nuclear power: A constructionist approach . American Journal of Sociology , 95 ( 1 ), 1 – 37 . Google Scholar Crossref Search ADS Gans H. J. (2004) [1979] . Deciding what’s news. A study of CBS evening news, NBC nightly news, Newsweek, and time . Evanston, IL : Northwestern University Press . Gitlin T. ( 1980 ). The whole world is watching: Mass media in the making & unmaking of the new left . Berkeley, CA; Los Angeles, CA; London : University of California Press . Groeling T. ( 2010 ). When politicians attack: Party cohesion ni the media . New York, NY : Cambridge University Press . Google Scholar Crossref Search ADS Groeling T. ( 2013 ). Media bias by the numbers . Annual Review of Political Science , 16 , 129 – 151 . Google Scholar Crossref Search ADS Hardisty D. J. , Johnson E. J. , Weber E. U. ( 2010 ). A dirty word or a dirty world? Attribute framing, political affiliation and query theory . Psychological Science , 21 , 86 – 92 . DOI: 10.1177/0956797609355572. Google Scholar Crossref Search ADS Hopmann D. N. , van Aelst P. , Legnante G. ( 2012 ). Political balance in the news: A review of concepts, operationalizations and key findings . Journalism , 13 ( 2 ), 240 – 257 . Google Scholar Crossref Search ADS Hopmann D. N. , Vliegenthart R. , De Vreese C. H. , Albaek E. ( 2010 ). Effects of election news coverage: How visibility and tone influence party choice . Political Communication , 27 , 389 – 405 . DOI: 10.1080/10584609.2010.516798. Google Scholar Crossref Search ADS Iyengar S. , Kinder D. ( 1987 ). News that matters: Television and American opinion . Chicago : The University of Chicago Press . Kleinnijenhuis J. , Hoof A. M. J., van , Oegema D. , De Ridder J. A. ( 2007 ). A test of rivaling approaches to explain news effects: News on issue positions of parties, real-world developments, support and criticism, and success and failure . Journal of Communication , 57 , 366 – 384 . Google Scholar Crossref Search ADS Korteweg A. , Yurdakul G. ( 2014 ). The headscarf debates: Conflicts of national belonging . Stanford, CA : Stanford University Press . Lee F. L. F. ( 2014 ). Triggering the protest paradigm: Examining factors affecting news coverage of protests . International Journal of Communication , 8 , 2725 – 2746 . Levin I. P. , Schneider S. L. , Gaeth G. J. ( 1998 ). All frames are not created equal: A typology and critical analysis of framing effects . Organizational Behavior and Human Decision Processes , 76 , 149 – 188 . Google Scholar Crossref Search ADS Linden A. , Klandermans B. ( 2006 ). Stigmatization and repression of extreme-right activism in The Netherlands . Mobilization , 11 , 213 – 228 . Link B. G. , Phelan J. C. ( 2001 ). Conceptualizing stigma . Annual Review of Sociology , 27 , 363 – 385 . Google Scholar Crossref Search ADS Lubbers M. , Gijsberts M. , Scheepers P. ( 2002 ). Extreme right-wing voting in Western Europe . European Journal of Political Research , 41 , 345 – 378 . Google Scholar Crossref Search ADS Major B. , O’Brien L. T. ( 2005 ). The social psychology of stigma . Annual Review of Psychology , 56 , 393 – 421 . Google Scholar Crossref Search ADS Mazzoleni G. ( 2003 ). The media and the growth of neo-populism in contemporary democracies. In Mazzoleni G. , Stewart J. , Horsfield B. (Eds.), The media and neo-populism: A contemporary comparative analysis (pp. 1 – 20 ). Westport, CT : Praeger . Molotch H. ( 1979 ). Media and movements. In Zald M. N. , McCarthy J. D. (Eds.), The dynamics of social movements: Resource mobilization, social control, and tactics (pp. 71 – 93 ). Cambridge, MA : Winthrop Publishers . Nelson T. E. , Oxley Z. M. ( 1999 ). Issue framing effects on belief importance and opinion . Journal of Politics , 61 , 1040 – 1067 . Google Scholar Crossref Search ADS Nelson T. E. , Clawson R. A. , Oxley Z. M. ( 1997 ). Media framing of civil liberties conflict and its effect on tolerance . American Political Science Review , 91 , 567 – 583 . Google Scholar Crossref Search ADS Norris P. , Curtice J. , Sanders D. , Scammell M. , Semetko H. A. ( 1999 ). On message: Communicating the campaign . London : Sage . Oegema D. , Kleinnijenhuis J. ( 2000 ). Personalization in political television news: A 13-wave survey study to assess effects of text and footage . Communications , 25 , 43 – 60 . Google Scholar Crossref Search ADS Paletz D. L. , Entman R. M. ( 1981 ). Media, power, politics . New York, NY; London : The Free Press . Plasser F. , Ulram P. A. ( 2003 ). Striking a responsive chord: Mass media and right-wing populism in Austria. In Mazzoleni G. , Stewart J. , Horsfield B. (Eds.), The media and neo-populism: A contemporary comparative analysis (pp. 21 – 43 ). Westport, CT : Praeger . Preacher K. J. , Rucker D. D. , Hayes A. F. ( 2007 ). Assessing moderated mediation hypotheses: Theory, methods, and prescriptions . Multivariate Behavioral Research , 42 , 185 – 227 . Google Scholar Crossref Search ADS Prior M. ( 2007 ). The immensely inflated news audience: Assessing bias in self-reported news exposure . Public Opinion Quarterly , 73 , 130 – 143 . Google Scholar Crossref Search ADS Schafraad P. , d’Haenens L. , Scheepers P. , Wester F. ( 2012 ). The evolution of a media image: Newspaper attention to the Flemish Far Right 1987-2004 . Acta Politica , 47 , 356 – 377 . Google Scholar Crossref Search ADS Schafraad P. , Scheepers P. , Wester F. ( 2009 ). Media attention to Pim Fortuyn and LPF during the 2002 elections. Negative bias or following news routines? In Konig R. P , Huysmans F. J. M. , Nelissen P. W. M. (Eds.), Meaningful media. Communication research on the social construction of reality (pp. 188 – 200 ). Nijmegen : Tandem Felix . Schemer C. , Wirth W. , Matthes J. ( 2012 ). Value resonance and value framing effect on voting intentions in direct-democratic campaigns . American Behavioral Scientist , 56 , 334 – 352 . DOI: 10.1177/000276421142632. Google Scholar Crossref Search ADS Shah D. V. , Domke D. , Wackman D. B. ( 1996 ). “To thine own self be true.” Values, framing, and voter decision-making strategies . Communication Research , 23 , 509 – 560 . doi: 10.1177/009365096023005001. Google Scholar Crossref Search ADS Shoemaker P. J. ( 1984 ). Media treatment of deviant political groups . Journalism Quarterly , 61 , 66 – 75 . Google Scholar Crossref Search ADS Tillie J. ( 2008 ). Gedeeld land. Het multiculturele ongemak van Nederland. (Divided country. The discomfort of the Dutch multicultural society) . Amsterdam : Meulenhoff . Valkenburg P. M. , Peter J. ( 2013 ). The differential susceptibility to media effects model . Journal of Communication , 63 , 221 – 243 . DOI: 10.1111/jcom.12024. Google Scholar Crossref Search ADS van der Brug W. , Fennema M. ( 2007 ). What causes people to vote for a radical right party? A review of recent work . International Journal of Public Opinion Research , 19 , 474 – 487 . Google Scholar Crossref Search ADS van der Eijk C. , Niemöller B. ( 1984 ). Het Potentiele Electoraat van de Nederlandse Politieke Partijen . Beleid en Maatschappij , 11 , 192 – 204 . van der Eijk C. , van der Brug W. , Kroh M. , Franklin M. N. ( 2006 ). Rethinking the dependent variable in voting behavior: On the measurement and analysis of electoral utilities . Electoral Studies , 25 , 424 – 447 . Google Scholar Crossref Search ADS van Heerden S. C. ( 2014 ). What did you just call me? A study on the demonization of Dutch political parties between 1995 and 2011 . Amsterdam : University of Amsterdam . van Heerden S. C. , van der Brug W. ( 2017 ). Demonisation and electoral support for populist radical right parties: A temporary effect . Electoral Studies , 47 , 36 – 45 . Google Scholar Crossref Search ADS van Spanje J. H. P. , De Vreese C. H. ( 2014 ). Europhile media and eurosceptic voting: Effects of news media coverage on eurosceptic voting in the 2009 European parliamentary elections . Political Communication , 31 , 325 – 354 . Google Scholar Crossref Search ADS van Spanje J. H. P. , De Vreese C. H. ( 2015 ). The good, the bad, and the voter. The impact of hate speech prosecution of a politician on the electoral support for his party . Party Politics , 21 , 115 – 130 . Google Scholar Crossref Search ADS van Spanje J. H. P. , Weber T. ( 2017 ). Does Ostracism Affect Party Support? Comparative Lessons and Experimental Evidence . Party Politics (Advance Online Publication September 12, 2017). van Zoonen E. A. ( 1992 ). The women’s movement and the media: Constructing a public identity . European Journal of Communication , 7 , 453 – 476 . Google Scholar Crossref Search ADS Vliegenthart R. , Boomgaarden H. G. , van Spanje J. H. P. ( 2012 ). Anti-immigrant party support and media visibility: A cross-party over-time perspective . Journal of Elections, Public Opinion and Parties , 22 , 315 – 358 . Google Scholar Crossref Search ADS Walgrave S. , De Swert K. ( 2004 ). The making of the (issues of the) Vlaams Blok: On the contribution of the news media to the electoral success of the Belgian extreme-right party . Political Communication , 21 , 479 – 500 . Google Scholar Crossref Search ADS Weaver D. A. , Scacco J. M. ( 2013 ). Revisiting the protest paradigm: The tea party as filtered through prime-time cable news . The International Journal of Press/Politics , 18 , 61 – 84 . Google Scholar Crossref Search ADS Zaller J. ( 1989 ). Bringing converse back in: Modeling information flow in political campaigns . Political Analysis , 1 , 181 – 234 . Google Scholar Crossref Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of The World Association for Public Opinion Research. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png International Journal of Public Opinion Research Oxford University Press

Tainted Love: How Stigmatization of a Political Party in News Media Reduces Its Electoral Support

Loading next page...
 
/lp/ou_press/tainted-love-how-stigmatization-of-a-political-party-in-news-media-0VvE055WIJ
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of The World Association for Public Opinion Research. All rights reserved.
ISSN
0954-2892
eISSN
1471-6909
D.O.I.
10.1093/ijpor/edy009
Publisher site
See Article on Publisher Site

Abstract

Abstract In contemporary democracies, a political party typically needs good press to attract voters. A glum scenario for a party would be that news media systematically stigmatize it. To what extent does stigmatization lower its electoral support? This article examines voters’ reactions to news media coverage of the Dutch party PVV. Using a media content analysis linked to a 2014 panel survey of a sample representative of the Dutch electorate, we find that, among voters who held anti-immigrant attitudes, exposure to stigmatization lowered these voters’ perceived legitimacy of the PVV. This, in turn, decreased their propensity to vote for that party. This suggests that stigmatization can be a strong tool in the hands of those who intend to damage a party. News media can be biased in various ways. What are the electoral effects of media bias? This important question has hardly ever been satisfactorily addressed. This is largely because media bias effects research suffers from the well-known subjectivity problem (Groeling, 2013). This article investigates effects of media bias while minimizing problems related to subjectivity. The subjectivity problem refers to the notion that each voter will have their own perception of a news item being biased or not. Partisans may “sincerely perceive news as being biased against their preferred stance, even when it is actually unbiased” (Groeling, 2013, p. 139) because of well-known cognitive biases—or because of the reputation of a particular media source. In this article, we solve this problem by examining indisputable stigmatization. We thus look for terms that voters should consider extremely negative connotations no matter their background. We find a Dutch political party that in 2014 was suddenly consistently linked to terms closely associated with the embodiment of pure evil. Or, we should say, what in that particular context is universally considered as such, Hitler’s Third Reich (van Heerden and van der Brug, 2017). We focus on the Freedom Party (PVV) in the 2014 European Parliamentary elections in The Netherlands. This provides an excellent case for studying electoral effects of stigmatizing a party, as the PVV suddenly changed from principled shunning of National Front (FN) in France, Flemish Interest (VB) in Belgium, and Freedom Party of Austria (FPÖ) to close cooperation with these parties. In response, various news media repeatedly associated the PVV with alleged characteristics of FN, VB, and FPÖ—including fascism and neo-Nazism. To what extent, how, and among which voters did being systematically associated with fascism and neo-Nazism affect decisions to vote for the PVV? We exploit variation in exposure of individual voters to such associations to address these questions. The 2014 European Parliamentary election campaign was chosen as the time under study, as that was the first time that voters massively tuned in after the party’s change of course. In doing so, this study contributes to our understanding of the electoral effects of news media bias in established democracies. Furthermore, we go beyond many previous media influence studies by taking into account voter heterogeneity and the mechanism underlying the electoral effect under study by modeling moderation and mediation effects. Also, we improve on much earlier research on political parties by bringing in concepts from psychology/sociology (stigmatization) and communication science (attribute framing; protest paradigm), and linking them to concepts used in political science (perceived party legitimacy; anti-immigrant attitude) using a communication science model (the Differential Susceptibility to Media Effects Model [DSMM] model). We explain our comprehensive approach in the next paragraph. Theoretical Model Media bias has predominantly been studied in the U.S. context (D’Alessio & Allen, 2000; Groeling, 2013). No research exists on media bias vis-à-vis anti-immigration parties—parties that are absent from the U.S. context. Although it is always difficult to identify a relevant benchmark against which to measure bias (Entman, 2007; Hopmann, van Aelst, & Legnante, 2012), anti-immigration parties such as the PVV, FN, VB, and FPÖ seem likely candidates to find bias against. This is because in many ways, these parties challenge the political status quo, and several scholars have argued that political challengers of the status quo generally fight an uphill battle. “Effective political action is likely when it does not disturb power, income, or status hierarchies” (Edelman, 1977, p. 141). In political discourse, “[p]erspectives that challenge the status quo are not accorded the legitimacy that would make them subjects of serious discussion” (Michael Lipsky, in Edelman, 1977, p. xvii). Indeed, the news media are often said to be “agents of social control” (Gans, 2004 [1979], p. 295). Two main criticisms are fielded here. First, the news media are said to silence challengers by hardly paying any attention to them. Second, they are accused of ridiculing and stigmatizing challengers. Concerning silence, Bennett (1990, 2016) argues that the news media engage in “indexing.” Indexing is “the tendency of mainstream news organizations to index or adjust the range of viewpoints in a story to the dominant positions of those whom journalists perceive to have enough power to affect the outcome of a situation” (Bennett, 2016, p.16). As long as challengers such as anti-immigration parties are small and not in power, their viewpoints will hardly be taken into account—if at all. Analogous to observations by Molotch (1979) about social movements, we can state that challengers of the political status quo are nothing like standard official sources from which anything is by default fit to print. In contrast with news from established actors, news from social movements is not “prima facie interesting, important, and defensible to work supervisors as worthy of publication” (Molotch, 1979, p. 77). Cohen (1972), in contrast, states that “(t)he mass media, in fact, devote a great deal of space to deviance” (p. 17). Similarly, Paletz and Entman (1981, pp. 125–126) claim that “marginal groups do not go entirely uncovered by the mass media.” This said, “media coverage of various groups is drastically and dramatically different,” some groups being “scorned as pariahs” (p. 124). Similarly, Paletz and Entman (1981, p. 127) contend that “dissident or unconventional voices are sometimes heard. But they are usually treated in a way that deprives them of their eloquence and force; their threat to the larger distribution of power is muted.” Shoemaker (1984) also claims that deviant groups are not less prominent in the news—but they are treated less favorably (p. 66). This brings us to ridicule and stigma. If not by way of silencing them, new political groups may be marginalized by way of ridiculing or stigmatizing them (Ferree, 2005; Linden & Klandermans, 2006; van Zoonen, 1992). Indeed, anti-immigration parties may be reported on through the “protest paradigm” (Chan and Lee, 1984). The protest paradigm “is one framework that media scholars have used to systematically understand the specific type of frames that news media often use to weaken legitimacy, obscure a protest’s social/political concerns, or both” (Weaver & Scacco, 2013, p. 64). When an event is covered through the protest paradigm, the organizing political actor’s central concerns are ignored. Instead of on substantive matters underlying the event, the coverage focuses on particularities of the event, its participants, or their actions (Boykoff, 2006; Gitlin, 1980), for instance, a participant’s funny banner, a clumsy comment, or a peculiar clothing style, or, as another example, the one violent action in a generally peaceful and quiet demonstration, or the reactions of (often uninformed) bystanders to disruptive aspects of the event. This, the argument goes, leads to trivialization, marginalization, and often criminalization of the political actor. The paradigm commonly pertains to social movements. However, it may well be applicable to political parties. This said, the paradigm may not translate automatically or without systematic differences to coverage of social movements. An indication for this is that Lee (2014) finds fewer references to violence and disruption in news media coverage, and more quotes of protester sources when political matters are addressed. When applying the protest paradigm to anti-immigration parties, one would expect not only an emphasis on the deviant appearances of anti-immigration activists and politicians (e.g., PVV leader Geert Wilders’s bleached hair), but also (in an analogy to the criminalization of social movements) the portraying of these actors as evil. In other words, one would expect the stigmatization of these parties. Coverage of Anti-Immigration Parties Only few previous studies exist on the news media coverage of anti-immigration parties. They ascribe some importance to the role of the mass media in the emergence of these parties. Mazzoleni (2003), for example, writes that “(t)he mass media themselves may also be “players” in the political game, endorsing or opposing populist stances and policies, both intentionally and unintentionally” (p. 2). Some studies suggest that these parties benefit electorally from the media visibility of their leaders (Vliegenthart, Boomgaarden, & van Spanje, 2012) and from the visibility of their policy issues in the media (Boomgaarden & Vliegenthart, 2007; Walgrave & De Swert, 2004). Other studies have investigated how the parties themselves were covered by the media (Art, 2006; Ellinas, 2010; Mazzoleni, 2003; Schafraad, d’Haenens, Scheepers, & Wester, 2012). Among the few anti-immigration parties that have been studied are the FPÖ, the Northern League (LN) in Italy, the Front National (FN) in France, the Republicans (REP) in Germany, and the Flemish Interest (VB) in Belgium. In all five cases, media actors reacted strongly to the emergence of these parties. In Austria (Art, 2006, pp. 189–191; Plasser & Ulram, 2003, pp. 38–39) and Italy (Biorcio, 2003, pp. 78–88), some media outlets reacted with stigmatization, whereas others supported the anti-immigration parties. In contrast, anti-immigration parties in France (Birenbaum & Villa, 2003, pp. 51–52), Germany (Art, 2006, pp. 165–166), and Belgium (Schafraad et al., 2012, p. 373–374) were often stigmatized by all major media outlets. These patterns of reactions persisted for decades on end, suggesting that “culturally rooted frames and editorial routines prevent” media from changing their approach to new political movements such as anti-immigration parties (Schafraad et al., 2012, p. 373). This is in line with the literature on new left movements (Gamson & Modigliani, 1989; Gitlin, 1980; van Zoonen, 1992). Coverage of Anti-Immigration Parties and Its Effects Effects of election news coverage on party choice have rarely been studied. Findings relate to both visibility and tone of news coverage of parties affecting voters’ decision at the ballot box (Hopmann, Vliegenthart, De Vreese, & Albaek, 2010; Kleinnijenhuis, Hoof, Oegema, & De Ridder, 2007; Norris, Curtis, Sanders, Scammell, & Semetko, 1999; Oegema & Kleinnijenhuis, 2000; van Spanje & De Vreese, 2014). Effects of media bias are found in the U.S. context (Druckman and Parkin, 2005). None of these are studies of anti-immigration parties, however. Another study examined individual-level effects of coverage of an anti-immigration party leader (Bos, van der Brug, & De Vreese, 2011). In that study, the dependent variable is not party choice but the perceived effectiveness and legitimacy of the leader. To our knowledge, only one study has tested individual-level effects of news media coverage on voting for anti-immigration parties. That study, conducted by Burscher, van Spanje, & De Vreese (2015), included 17,014 respondents, 20,084 news items, and 13 anti-immigration parties in 11 countries. It showed that exposure to immigration and crime news enhanced voters’ propensity to vote for an anti-immigration party. Just as that article, the present study examines media effects. Concerning media effects, perhaps the most elaborate analytical model is Valkenburg and Peter’s (2013) DSMM. We link this model to several strands of the literature in this article. We then apply it to the case of the PVV in 2014, following various other applications (Beyens, Vandenbosch, & Eggermont, 2015; Beullens & van den Bulck, 2013). The DSMM consists of three elements relevant here. These are media use, media effects, and susceptibility factors (cf. Valkenburg & Peter, 2013). Media use is the “use of media types, content and technologies” (Valkenburg & Peter, 2013, p. 222). Media effects are “within-person changes” in, for example, attitudes and behavior “that result from media use” (p. 222). Media effects last beyond the media use situation (p. 224). Susceptibility factors include “all person dimensions that predispose the selection of and responsiveness to media” such as political attitudes (Valkenburg & Peter, 2013, p. 226). In this article we argue, and demonstrate empirically, that particular media use leads to particular media effects, given a particular susceptibility factor. More specifically, we show that a voter’s exposure to stigmatization of the PVV (media use) lowers her propensity to vote for that party (media effect) if she holds a negative attitude toward immigrants (susceptibility factor). Below, we elaborate on each of these three elements in turn. Media use Stigmatization of a party starts with attributing a label to that party (just as, for instance, the “protest paradigm” can be seen as a set of frames that can be applied to a new social movement—Weaver & Scacco, 2013). “Stigma exists when elements of labelling, stereotyping, separation, status loss, and discrimination occur together in a power situation that allows them” (Link & Phelan, 2001, p. 377). Seen in this light, stigmatization of a party involves—among other things—making particular compromising aspects of that party more salient than others. What does this mean for the case of the PVV? Justified or not, the PVV is lumped into the same category as FN, VB, and FPÖ (labeling), a category that is referred to as “fascist,” “neo-Nazi,” or the like (stereotyped), which sets the PVV apart from “normal” or “democratic” parties (separation), leading to lower utility for voters of a vote for the PVV (status loss) and thus a lower propensity to vote for the PVV (discrimination). The news media operate in a contextual “power situation” that allows them to categorize the PVV in this way. Media effects Here, we have entered the realm of media effects. The literature on voting for anti-immigration parties is relevant here. Anti-immigration party voting has been researched from various angles. Factors that play a role in voting for an anti-immigration party (see van der Brug & Fennema, 2007, for an overview) include factors related to news media (Boomgaarden & Vliegenthart, 2007; Walgrave & De Swert, 2004). Mediating factors include voters’ perception of how “legitimate” an anti-immigration party is—that is, whether the party is nonviolent and democratic (Bos & van der Brug, 2010). This links to the argument that many voters are “of two minds about immigration and minority politics and that this internal conflict affects their expressed policy preferences and vote choices in predictable ways” (Blinder, Ford, & Ivarsflaten, 2013, pp. 841–842). On the one hand, they oppose immigration. On the other, they subscribe to anti-prejudice, pro-democracy, and anti-violence norms, deeming, e.g., neo-Nazism and fascism unacceptable. Hence, associating an anti-immigration party with neo-Nazism or fascism may reduce its support—through decreasing its legitimacy. Susceptibility factor Obviously, we do not expect all voters to be equally affected. Electoral decisions of voters who agree with the PVV will clearly be more likely to be influenced than voters whose views are radically different from the party’s views. Not surprisingly, a particularly strong predictor of voting for this anti-immigration party is attitudes toward immigrants (Lubbers, Gijsberts, & Scheepers, 2002). We hypothesize that the more a voter holds anti-immigrant attitudes, the greater the media impact on her voting behavior will be. This is because voters who disagree with the PVV tend to be stable in their rejection of the party regardless of how it is portrayed or treated (van Spanje & Weber, 2017). Among anti-immigrant voters, in turn, one might suspect that moderately sophisticated voters are most susceptible to political persuasion, as Converse (1962) argues and Zaller (1989) demonstrates is often the case. This latter step, however, we cannot adequately test given the limits to our data. Thus, we have only anti-immigrant attitude as a “differential-susceptibility variable” that acts as a moderator of the effect of media use on the mediating media effect and as a moderator of the second-order effect of media use (Valkenburg & Peter, 2013, p. 231). One may argue that the DSMM has a broader scope, also looking at selective exposure, for example. In this article, we hold within-person time-invariant factors constant during a 2-month period. This means that the characteristics of the voter and her environment are kept constant, as they are not expected to change within that window. At the same time, this enforces a focus on the three factors that are expected to change in the short run: exposure to stigmatization, perceived legitimacy, and voting behavior. Figure 1 reflects the analytical framework used in this article. Figure 1 View largeDownload slide The analytical model Figure 1 View largeDownload slide The analytical model As shown in Figure 1, we hypothesize that individual voters’ exposure to stigmatization of the PVV in the news media negatively affects that party’s perceived legitimacy among these voters, which in turn should decrease their propensity to vote for the party. The stronger their anti-immigrant attitude is, the larger these effects are expected to be. Stigmatization and Demonization Each political party arguably has an infinite number of characteristics that could be made salient. Earlier research concentrated on various kinds of aspects that media may make salient such as party cohesion versus party discord (Groeling, 2010) or a party's political leader and his character (Iyengar and Kinder, 1987). In our particular example, the aspect of the PVV that is made salient is its cooperation with FN, VB, and FPÖ. This involves associating the PVV with these three parties, labeling the category of these parties, e.g., “extreme right,” stereotyping them, and setting them apart from other parties. Among consumers of that media content, this may lead to status loss of the PVV and discrimination against it. “Stigma involves status loss—a downward placement in the status hierarchy” (Link & Phelan, 2001, p. 379). The PVV drops in the hierarchy of parties in a voter’s mind, leading to a lower utility associated with a vote for that party and therefore a lower propensity to vote for it. Concerning the labeling, van Heerden (2014) proposes the concept of “demonization.” She defines demonization as “portraying an actor as the embodiment of absolute evil” (2014, p. 10). Demonization is a more specific concept than the labeling element of stigmatization (p. 16). The implicit consequence of demonization is, as van Heerden points out, more far-reaching than any implicit consequence of stigmatization: after all, anyone would agree that any “absolute evil” should be eradicated (p. 16). This sheer unanimity of perception gets us, for all intents and purposes, around the problem of subjectivity in media bias research. In the remainder of this article, we thus use demonization as a concept in the labeling of a party. In addition, we use the broader concept of stigmatization in this article, referring to not only the labeling of a party but also the stereotyping, separation, status loss, and discrimination. One might claim that in the case under investigation, the labeling is only indirect. On this view, it is not the PVV that is labeled. Instead, the PVV is associated with parties that are labeled. Yet, we hold that even in that case of such “labelling by proxy” there is still labeling, albeit indirect. The labeling of parties associated with the party arguably leads to stigmatization of that party itself, using the analogy of “guilt by association.” It is the PVV that faces the consequences of status loss and discrimination—as we will see. In fact, the indirectness of the labeling is a reason to consider the PVV in 2014 a “least likely case” in terms of electoral effects of stigmatization. After all, if even indirect labeling damages a party, the effect of such labeling should be considerable.1 Another reason is that PVV’s only member Wilders had had his own successful political movement for 10 years at the time of our study, which arguably made it difficult for others to have voters profoundly rethink their image of the party. That we find that this party image was nonetheless tarnished is yet another indication of the strength of the forces that stigmatization may unleash. Hypotheses On the basis of the theoretical considerations stated above, we formulate four hypotheses. Our first hypothesis is that exposure to news media content demonizing the PVV reduces a voter’s propensity to vote for that party (H1). We base this expectation on literatures on stigmatization and on the protest paradigm. Effects of media coverage of parties on voting are largely unknown. However, it is found in other contexts that exposure to news framing can affect voting behavior (Shah, Domke, & Wackman, 1996; van Spanje & De Vreese, 2014). Another indication is that visibility and tone of news media coverage of a party have been demonstrated empirically to influence voting for that party, a negative tone reducing individuals’ inclination to vote for the party (Hopmann, Vliegenthart, De Vreese, & Albaek, 2010). More specifically about stigmatization, the status loss and discrimination that demonization should bring along manifest themselves in lower propensity to vote for the party. We argue that this negative effect is because of voters “learning” about parties through exposure to attribute framing. Through exposure to information about attributes of a subject (e.g., an association of a political party to known extremist parties), this attribute will become more salient in the voter’s mind, promoting an evaluation of the subject using this particular attribute (see Levin, Schneider, & Gaeth, 1998). Several studies found evidence that voters use the information from the framed attribute to form political opinions (Nelson, Clawson, & Oxley, 1997, Nelson & Oxley, 1999) and in making (voting) decisions (Hardisty, Johnson, & Weber, 2010; Shah et al., 1996, van Spanje & De Vreese, 2014). Given that fascism and neo-Nazism are perceived as evil, the association with these terms resonates with the idea to reject such parties, which consecutively leads to a decrease in the propensity to vote for the associated party (cf. Schemer, Wirth, & Matthes, 2012). Our second hypothesis is that a voter’s perceived legitimacy of the PVV mediates the effect of exposure to demonization on her propensity to vote for that party (H2). More specifically, exposure is expected to negatively affect PVV legitimacy, which should in turn positively influence PVV vote propensity (cf. Bos & van der Brug, 2010). Exposure is expected to negatively affect PVV legitimacy because of the particular stigma involved. If a party is labeled “neo-Nazi” or “fascist,” its viability as an option in a democracy is clearly in question. Given the overwhelming support for democracy as a political regime in contemporary established democracies, this in turn decreases the party’s electoral attractiveness. This reasoning is in accordance with the empirical findings by Bos and van der Brug (2010) suggesting a considerable positive effect of a party’s perceived legitimacy on propensity to vote for that party. Our third hypothesis is that a voter’s attitude toward immigrants strengthens the effect of media use on PVV legitimacy (H3). PVV leader Wilders often targets immigrants using strongly negative words, for instance, when he proposed a “head rag tax” on headscarves, following the principle of “the polluter should pay” (Korteweg & Yurdakul, 2014, p. 120). Voters who have a positive attitude toward immigrants would not find the PVV legitimate to begin with—its association with FN, VB, or FPÖ not mattering much. Voters who hold a negative attitude toward immigrants may find the PVV legitimate yet deem neo-Nazism and fascism beyond the pale. Among these anti-immigrant voters, associations of the PVV with neo-Nazism and fascism are thus expected to reduce the party’s legitimacy. Our fourth hypothesis is that a voter’s attitude toward immigrants also strengthens the effect of PVV legitimacy on PVV vote propensity (H4). A negative attitude toward immigrants is a strong predictor of voting for a party that campaigns against immigration (van der Brug & Fennema, 2007), as the PVV does. Voters with positive attitudes toward immigrants are unlikely to vote for the PVV, no matter how legitimate they think the PVV is. Among voters with negative attitudes toward immigrants, in contrast, their likeliness to vote for the PVV should decrease when their perception is that the party is not legitimate anymore. Note that we do not in any way imply that the PVV in 2014 was a passive victim of stigmatization. In fact, party leader Wilders must have known well what he called onto himself and his party when he decided to start close cooperation with FN, VB, and FPÖ. The common assumption in the literature that the stigmatized is a passive victim is unsubstantiated (Link & Phelan, 2001, p. 365; Major & O’Brien, 2005, p. 411). That also holds in the case of the PVV—as we will see in the next section. The Case MP Wilders founded the PVV in 2006. Campaigning mainly on an anti-immigration and anti-Islam platform, the PVV has received between 5.9 and 15.5% of the vote in general elections. PVV MPs’ controversial public statements triggered accusations of fascism and also led to prosecution for hate speech (see van Spanje and De Vreese, 2015). Wilders has always tried to steer clear of fascism, neo-Nazism, or any ideologies that are clearly beyond the pale in contemporary Western Europe. An important element in this strategy is his distancing himself from other anti-immigration politicians. In 2009, he refused to join FN, VB, FPÖ, and a few other parties in talks about cooperation in the European parliament. However, Wilders suddenly and radically changed his stance toward these parties, trying to form a European parliamentary group with them in 2014. From that moment onward, the PVV has been repeatedly associated with these parties. Willingly or unwillingly, this has led to demonizing associations of the PVV. The election campaign in the run-up to the 2014 European Parliamentary election was the first campaign after the party’s change of course, and thus the first time that voters en masse followed politics again. In this article we use a panel survey during that election campaign, studying variation in voters’ exposure to demonizing associations of the PVV and its effects on their propensity to vote for that party. Methods To address our research question, we assess the four hypotheses using media content data linked to panel survey data (cf. Dilliplane, Goldman, & Mutz, 2013). A four-wave panel survey was collected within the framework of the 2014 European Election Campaign Study (De Vreese, Azrout, & Möller, 2014), of which the last three waves contained measures of variables of interest to this study (the moderator, anti-immigrant attitude, was measured in Wave 2; all other variables were measured in Waves 3 and 4). Field work took place from March 20 to 30, 2014 (Wave 2), April 17 to 28, 2014 (Wave 3), and from May 26, 2014 to June 2, 2014 (Wave 4). The time that the survey was fielded was short, as 90% of responses were in within 4–6 days in each wave. Fieldwork was conducted by TNS NIPO, a research institute that complies with ESOMAR guidelines for survey research. A random sample was drawn from the TNS NIPO database (which consists of 200,000 nationally representative individuals recruited through multiple recruitment strategies, including telephone, face-to-face, and online recruitment), with quotas enforced on age, gender, and education. This led to a net sample of 1,267 voters, from whom we have full data on all relevant variables.2 The media content data were constructed using all the media coverage in which the PVV was mentioned in the eight main national newspapers3 and the main national news site (www.nu.nl) between the second and the fourth wave. The articles were collected through the LexisNexis Academic data set, including all articles that mention either the party or its leader (Nprior Wave 3 = 286, Nprior Wave 4 = 560).4 Operationalizations Demonization of the PVV in the media is measured by coding per news item the extent to which a party that the PVV cooperates with is portrayed as “fascist,” “racist,” “(neo-)Nazi,” “extreme right,” “radical,” “discriminating,” or “anti-Semite.” We use these seven particular labels because we aim to capture the specific “demonization” labeling. In the context of The Netherlands in 2014, the moral scheme of WWII is still dominant (van Heerden & van der Brug, 2017, p. 37). As a consequence, Nazism and fascism are widely considered political regimes that represent absolute evil (cf. van Heerden, 2014, p. 14). Van Heerden thus operationalizes demonization within that context as “portraying a political actor as the embodiment of Nazism/fascism” (p. 14). We broaden this category from two to seven labels to also include references to “antisemitism,” “racism,” “discrimination,” “radical right,” and “extreme right,” which should be placed in the same narrow category of connotations with the Third Reich, responsible for the culmination of absolute evil, the Holocaust. In the contemporary Dutch context, there is a clear distinction between the seven labels mentioned and other labels. As van Heerden and van der Brug (2017, p. 37) note “While the atrocities of the Nazi’s are in terms of the number of victims comparable to those of other authoritarian regimes…, adherents of those ideologies are not prosecuted in the same way as fascists.” Other labels, such as “populist” or “eurosceptic,” may not necessarily be demonizing in voters’ eyes. The same goes for “anti-establishment” or “anti-Islam.” In the eyes of many voters, these labels have nothing negative or delegitimizing. Indeed, many Dutch voters may describe themselves as anti-establishment or anti-Islam at the moment. However, no Dutch voter in her right mind would openly call herself “fascist,” “(neo-)Nazi,” “racist,” “radical right,” “discriminating,” “extreme right,” or “anti-Semite” because if she did, she would immediately become an outcast (cf. Tillie, 2008; van Heerden, 2014). These “demonizing” labels can be seen as clearly delegitimizing, and therefore potentially damaging a party, across the board. From the articles mentioning the PVV and/or Wilders, we automatically selected those articles that mentioned a party that the PVV cooperates with5 and contained one of the demonizing words.6 To make sure that the co-occurrence of these mentionings indeed mean that the article links the PVV to this demonized cooperating party, we also manually coded these articles.7 In 13 of the 77 articles in which the combination of the PVV, a cooperating party, and a demonizing word was made, the author did either not link the PVV to the other party or did not use the demonizing word to describe the cooperating party. Only the remaining 64 news articles are therefore coded as demonizing the PVV. We divide the number of articles in which the PVV was demonized by the total number of articles in which the PVV is mentioned to create a (relative) media demonization scale. To correct for the different length of the time (i.e., the different numbers of editions of the newspapers), we also divided the relative score by the number of newspaper editions in each period. See Table 1 for the frequency of the occurrence of the PVV and demonization of that party, split by the periods before survey Wave 3 and between Waves 3 and 4. Table 1 Frequency of the Occurrence of PVV and of Demonization of the Party Before survey Wave 3 Between survey Waves 3 and 4 N PVV N demon -ization Relative demon -ization N PVV N demon -ization Relative demon -ization 1 De Telegraaf 35 2 0.06 93 17 0.18 2 NRC Handelsblad 43 1 0.02 82 7 0.09 3 NRC.NEXT 18 0 0.00 27 3 0.11 4 Trouw 53 0 0.00 96 11 0.11 5 AD 36 0 0.00 72 4 0.06 6 De Volkskrant 51 1 0.02 117 6 0.05 7 Metro 24 0 0.00 25 3 0.12 8 Spits 13 1 0.08 11 1 0.09 9 www.nu.nl 13 1 0.08 37 6 0.16 Total 286 6 0.02 560 58 0.10 Before survey Wave 3 Between survey Waves 3 and 4 N PVV N demon -ization Relative demon -ization N PVV N demon -ization Relative demon -ization 1 De Telegraaf 35 2 0.06 93 17 0.18 2 NRC Handelsblad 43 1 0.02 82 7 0.09 3 NRC.NEXT 18 0 0.00 27 3 0.11 4 Trouw 53 0 0.00 96 11 0.11 5 AD 36 0 0.00 72 4 0.06 6 De Volkskrant 51 1 0.02 117 6 0.05 7 Metro 24 0 0.00 25 3 0.12 8 Spits 13 1 0.08 11 1 0.09 9 www.nu.nl 13 1 0.08 37 6 0.16 Total 286 6 0.02 560 58 0.10 Table 1 Frequency of the Occurrence of PVV and of Demonization of the Party Before survey Wave 3 Between survey Waves 3 and 4 N PVV N demon -ization Relative demon -ization N PVV N demon -ization Relative demon -ization 1 De Telegraaf 35 2 0.06 93 17 0.18 2 NRC Handelsblad 43 1 0.02 82 7 0.09 3 NRC.NEXT 18 0 0.00 27 3 0.11 4 Trouw 53 0 0.00 96 11 0.11 5 AD 36 0 0.00 72 4 0.06 6 De Volkskrant 51 1 0.02 117 6 0.05 7 Metro 24 0 0.00 25 3 0.12 8 Spits 13 1 0.08 11 1 0.09 9 www.nu.nl 13 1 0.08 37 6 0.16 Total 286 6 0.02 560 58 0.10 Before survey Wave 3 Between survey Waves 3 and 4 N PVV N demon -ization Relative demon -ization N PVV N demon -ization Relative demon -ization 1 De Telegraaf 35 2 0.06 93 17 0.18 2 NRC Handelsblad 43 1 0.02 82 7 0.09 3 NRC.NEXT 18 0 0.00 27 3 0.11 4 Trouw 53 0 0.00 96 11 0.11 5 AD 36 0 0.00 72 4 0.06 6 De Volkskrant 51 1 0.02 117 6 0.05 7 Metro 24 0 0.00 25 3 0.12 8 Spits 13 1 0.08 11 1 0.09 9 www.nu.nl 13 1 0.08 37 6 0.16 Total 286 6 0.02 560 58 0.10 We link the demonization of the PVV in the media to the survey data by a self-reported media exposure measure. Media exposure was gauged in both Waves 3 and 4 by asking how many days in a normal week the respondent reads “the following” newspapers, followed by a list of eight national newspapers in randomized order. For each of the newspapers, a respondent could indicate that she read it anywhere from 0 to 7 days a week. In addition, the respondent was asked to indicate how many days in a normal week she read about politics on the by far most often used Dutch news website, www.nu.nl, of which the content was coded. We consecutively construct a demonization exposure measure by multiplying the reported frequency (of the respondent’s use of a particular outlet) with the relative demonization (in that outlet) in the period preceding the wave in which the question was asked. Per respondent and per wave we add this for all newspapers together. This leads to a demonization exposure variable in which 0 means a respondent is not exposed to any demonization of the PVV (i.e., either the respondent does not use any of these media outlets or the respondent uses merely outlets that do not contain any demonization). The value of this demonization exposure variable increases as the respondent increases her use of an outlet that contains demonization, and also as the outlet the respondent uses increases demonization. Voters’ perception of legitimacy of the PVV is measured by reactions to four statements. These are “To what extent does the PVV comply with the laws, you think?”; “To what extent do you think the PVV has the right to get access to power?”; “To what extent does the PVV leadership respect the rules of our democracy, you think?”; and “To what extent do you think the PVV abides by the prevailing social norms in our society?” Answering options ranged from “not at all” (0) to “to a high degree” (6). This leads to a single scale (eigen value 3.31 and 83% explained variance) that is highly reliable (Cronbach’s α = .92). Voters’ propensity to ever vote for the PVV is gauged by the standard question common to many election surveys since van der Eijk and Niemöller (1984). In response to this question, voters can indicate the propensity that they “will ever vote for” a political party. They can do so on a scale varying from “very unlikely” (1) to “very likely”(10), with an explicit “don’t know” option. The question is designed to get the utility a voter would derive from voting for a party (van der Eijk, van der Brug, Kroh, & Franklin, 2006). Propensities to vote were asked for each of 11 relevant Dutch parties in random order. Obviously, the relevant one here is the propensity to vote for the PVV. Attitudes toward immigrants are measured using five items: “Immigrants abuse the Dutch social welfare system, as they take more out of it than they put in”; “Immigrants are a threat to the Dutch population’s security”: “The religious practices of immigrants enrich the Dutch way of life and traditions” (reversely coded); “Immigrants are an important cause of crime in the Netherlands”; and “Immigration is good for the Dutch labor market” (reversely coded). Answering options to these items, ranging from 0 (“strongly disagree”) to 6 (“strongly agree”), form a scale (1 factor with eigen value of 3.10, explained variance 62%) that is highly reliable (Cronbach’s α = .85). Data Analysis To test our hypotheses, we combine fixed-effects regression modeling with mediation (Baron & Kenny, 1986; Preacher, Rucker, & Hayes, 2007). Using fixed effects, we focus not on the between-subject variance to estimate effects but on the within-subject change across the waves (Allison, 2009). As a result, in a fixed-effects model, all time-invariant factors are implicitly controlled for (thus, explicitly controlling for them in the model is redundant). We combine the fixed effects with mediation to be able to test the indirect effect through legitimacy perceptions.8 We start with a baseline model, assessing the unmediated and unmoderated fixed effect of exposure to demonization on the propensity to vote (H1). Second, we assess the mediation by perceived legitimacy (H2) by testing the fixed effect of the demonization exposure on legitimacy perceptions (the independent variable should predict the mediator) and predicting propensity to vote with both demonization exposure and perceived legitimacy (the mediator should have a significant effect, while the effect of the independent variable should decrease compared with the initial model). To assess the moderation (H3 and H4), we add anti-immigrant attitude as well as its interaction with exposure to demonization and with perceived legitimacy. We focus on the direction and the significance (based on the Sobel test) of the coefficients to test our hypotheses, as well as on the marginal effects. Results We first turn to the unmediated and unmoderated effect of exposure to demonization on the propensity to vote for the PVV (H1). Model 1 in Table 2 shows that the coefficient (b = −1.82) is in the expected direction (increased exposure to demonization decreases the propensity to vote), significant at the p < .10 level (p = .094). To interpret the real-world implications of this coefficient, we assess what change on the propensity to vote this coefficient implies for the average observed change in exposure to demonization. On average, exposure to demonization increases with a score of 0.02, implying an average decrease of 0.03 on the (1–10) propensity to vote scale. In light of this marginal support and the modest real-world implication, we conclude that our first hypothesis is only partially supported. This is underwhelming evidence, which is partly because of the (also hypothesized) moderation effect of anti-immigrant attitude, as we will see. Table 2 Fixed-Effects Predicting Probability to Vote for the PVV 1 2 3 4 Exposure to demonization −1.82* −1.29 3.36 1.66 (1.38) (1.36) (3.38) (2.44) Legitimacy 0.28**** −0.13* (0.03) (0.06) Anti-immigrant attitude −0.00 −0.00 (0.01) (0.01) Stigma ** AIA −1.44** −0.78 (0.86) (0.85) Legitimacy ** AIA 0.11**** (0.02) R2 .004 .032 .006 .041 1 2 3 4 Exposure to demonization −1.82* −1.29 3.36 1.66 (1.38) (1.36) (3.38) (2.44) Legitimacy 0.28**** −0.13* (0.03) (0.06) Anti-immigrant attitude −0.00 −0.00 (0.01) (0.01) Stigma ** AIA −1.44** −0.78 (0.86) (0.85) Legitimacy ** AIA 0.11**** (0.02) R2 .004 .032 .006 .041 Note. Entries are unstandardized fixed-effects coefficients, with standard errors in parentheses. ****p < .001; *** p < .01; ** p < .05; * p < .1 (one-tailed). Table 2 Fixed-Effects Predicting Probability to Vote for the PVV 1 2 3 4 Exposure to demonization −1.82* −1.29 3.36 1.66 (1.38) (1.36) (3.38) (2.44) Legitimacy 0.28**** −0.13* (0.03) (0.06) Anti-immigrant attitude −0.00 −0.00 (0.01) (0.01) Stigma ** AIA −1.44** −0.78 (0.86) (0.85) Legitimacy ** AIA 0.11**** (0.02) R2 .004 .032 .006 .041 1 2 3 4 Exposure to demonization −1.82* −1.29 3.36 1.66 (1.38) (1.36) (3.38) (2.44) Legitimacy 0.28**** −0.13* (0.03) (0.06) Anti-immigrant attitude −0.00 −0.00 (0.01) (0.01) Stigma ** AIA −1.44** −0.78 (0.86) (0.85) Legitimacy ** AIA 0.11**** (0.02) R2 .004 .032 .006 .041 Note. Entries are unstandardized fixed-effects coefficients, with standard errors in parentheses. ****p < .001; *** p < .01; ** p < .05; * p < .1 (one-tailed). Before turning to moderation, we first assess the unmoderated indirect effect though perceived legitimacy. In the final table, Table 3, Model 5 shows that an increase in exposure to demonization decreases perceived legitimacy (b = −1.87, p = .011). Again, we interpret the real-world implication of this coefficient, and estimate that the average change in the exposure to demonization leads to a decrease of 0.03 on the (1–7) perceived legitimacy scale. Turning back to predicting the propensity to vote (back to Table 2), but now with both the independent variable and the mediator (see Model 2 in Table 2), we find that higher perceived legitimacy leads to higher propensity to vote (b = 0.28, p < .001), while the coefficient for exposure to demonization decreases compared with Model 1 and is not significant anymore (b = −1.29, p = .171). With an indirect effect estimated at −0.52 (se = 0.24, p = .014), these findings imply that we indeed have an indirect effect through perceived legitimacy, which supports H2. Table 3 Fixed-Effects Predicting Perceived Legitimacy of the PVV 5 6 Exposure to demonization −1.87** 6.37**** (0.81) (0.01) Anti-immigrant attitude −0.00 (0.01) Stigma ** AIA −2.29**** (1.45) R2 .004 .012 5 6 Exposure to demonization −1.87** 6.37**** (0.81) (0.01) Anti-immigrant attitude −0.00 (0.01) Stigma ** AIA −2.29**** (1.45) R2 .004 .012 Note. Entries are unstandardized fixed-effects coefficients, with standard errors in parentheses. ****p < .001; *** p < .01; ** p < .05; * p < .1 (one-tailed). Table 3 Fixed-Effects Predicting Perceived Legitimacy of the PVV 5 6 Exposure to demonization −1.87** 6.37**** (0.81) (0.01) Anti-immigrant attitude −0.00 (0.01) Stigma ** AIA −2.29**** (1.45) R2 .004 .012 5 6 Exposure to demonization −1.87** 6.37**** (0.81) (0.01) Anti-immigrant attitude −0.00 (0.01) Stigma ** AIA −2.29**** (1.45) R2 .004 .012 Note. Entries are unstandardized fixed-effects coefficients, with standard errors in parentheses. ****p < .001; *** p < .01; ** p < .05; * p < .1 (one-tailed). We now turn to the moderation by anti-immigrant attitude. We first hypothesized that this moderator would affect the effect of exposure to demonization on perceived legitimacy. As Model 6 in Table 3 shows, we find a significant interaction (p < .001) in the expected direction (b = −2.29). The negative sign implies that the more anti-immigrant a respondent is, the more negative the effect of exposure to demonization on exposure. To further explore the interaction, we plotted the marginal effect of exposure to demonization on perceived legitimacy for different values of anti-immigrant attitude (see Figure 2). We see a negative slope (as indicated by the negative coefficient of the interaction). The confidence interval shows that the marginal effect becomes significant for values of anti-immigrant attitude above 3.5, with an estimated coefficient of −7.36 (se = 1.45, p < .001) at the maximum value of anti-immigrant attitude. To interpret these effect sizes, we estimate the change in perceived legitimacy by the average change in exposure to demonization by different values of anti-immigrant attitude. We find an expected decrease in perceived legitimacy of 0.05 when anti-immigrant attitude is at 4, a decrease of 0.09 when anti-immigrant attitude is at 5, and a decrease of 0.13 when anti-immigrant attitude is at its maximum 6. This shows that individuals who are negative toward immigrants are affected by demonization in a way that is consistent with H3. Figure 2 View largeDownload slide Marginal effect of exposure to stigmatization on perceived legitimacy (x axis) for different values of anti-immigrant attitudes (y axis). The dashed line represents the 90% confidence interval Figure 2 View largeDownload slide Marginal effect of exposure to stigmatization on perceived legitimacy (x axis) for different values of anti-immigrant attitudes (y axis). The dashed line represents the 90% confidence interval H4 predicts that the effect of perceived legitimacy on propensity to vote is moderated by anti-immigrant attitude. As shown in Model 4 of Table 2, we find a significant positive coefficient (b = 0.11, p < .001). The sign indicates that the more anti-immigrant a respondent is, the more positive the effect of perceived legitimacy on the propensity to vote. Looking at the confidence interval of the marginal effects (Figure 3), we see that for anti-immigrant attitude values above 2, the legitimacy effect on propensity to vote is significant, and the marginal effect is increasing further as anti-immigrant attitude values increase further. H4 is supported. Figure 3 View largeDownload slide Marginal effect of perceived legitimacy on propensity to vote (x axis) for different values of anti-immigrant attitudes (y axis). The dashed line represents the 90% confidence interval Figure 3 View largeDownload slide Marginal effect of perceived legitimacy on propensity to vote (x axis) for different values of anti-immigrant attitudes (y axis). The dashed line represents the 90% confidence interval Given the support for H3 and H4, we can also assess to which degree the total effect and the indirect effects are moderated by anti-immigrant attitude. With regard to the moderation of the total effect, we find that this is significantly moderated (b = −1.44, p = .047, see Model 3 of Table 2). The marginal effect in Figure 4 shows that we find significant negative effects of exposure to demonization on propensity to vote only as anti-immigrant attitude values are above 4, with expected change in propensity to vote by the mean change in demonization exposure of −0.042 when anti-immigrant attitude is at 4, −0.068 when anti-immigrant attitude is at 5, and −0.093 when anti-immigrant attitude is at 6. This also explains why we find an only weak main total effect when testing H1. Figure 4 View largeDownload slide Marginal total effect of exposure to stigmatization on propensity to vote (x axis) for different values of anti-immigrant attitudes (y axis). The dashed line represents the 90% confidence interval Figure 4 View largeDownload slide Marginal total effect of exposure to stigmatization on propensity to vote (x axis) for different values of anti-immigrant attitudes (y axis). The dashed line represents the 90% confidence interval When adding the interaction between anti-immigrant attitude and perceived legitimacy to the model predicting propensity to vote (see Model 4 of Table 2), both the main and moderated effects of exposure to demonization fail to reach conventional levels of statistical significance. Furthermore, there is a significant marginal effect of the indirect effect through perceived legitimacy starting at anti-immigrant values of about 3.5, i.e., when we find both significant marginal effects for exposure to demonization on perceived legitimacy (anti-immigrant attitude values above 3.5) and significant marginal effects for perceived legitimacy on propensity to vote (anti-immigrant attitude values above 2), and the stronger the anti-immigrant attitude, the stronger the indirect path through perceived legitimacy.9 Conclusion To what extent, and how, does media bias have electoral impact? In this article, we have examined the systematic stigmatization of a party as a result of being demonized in various main news outlets—arguably an extreme case of news media reaction to a party. Examining such stigmatization of a political party as the embodiment of absolute evil circumvents most problems related to subjectivity that commonly plague media bias research. Although not all subjectivity problems are ruled out by this (e.g., PVV supporters may systematically more often miss the stigmatizing part of the message), the chance of diverging perceptions of a particular demonizing message is minimized. After all, the absolute evil is not subject to subjectivity. As it turns out, this stigmatization can manifest itself by negatively affecting voters’ perceptions of that party, and by reducing its electoral support. In the case under investigation, the Dutch PVV in the 2014 European election campaign, exposure to demonization has been demonstrated to lower the party’s perceived legitimacy. This reduced legitimacy in turn decreased the party’s support. These effects only occurred in the electoral pool of voters with strong anti-immigrant attitudes in which the PVV has always been fishing. We use a strong research design of a media content analysis linked to a voter panel survey, which is cutting edge in political communication research (Dilliplane, Goldman, & Mutz, 2013). Admittedly, the effect sizes are modest. But for practical reasons, we focused on newspapers, which enabled us to use automated content analysis, and newspapers are only a small part of voters’ news consumption. Given that we estimated the effect sizes from average newspaper usage, we expect that if, for instance, television news were included, implications would be more impressive. With regard to the average news consumption, we do need to note that we make use of a self-reported exposure measure, which is known to be associated with problems related to overreporting (cf. Prior, 2007). This said, we have been able to find consistent effects even in the presence of such measurement error. Furthermore, we find these effects in what we consider a least likely case. First, it is important to keep in mind that the demonization under investigation is an indirect one, as the PVV’s collaborators are labeled “Nazi” or “fascist”—not the party itself. Second, by 2014 the PVV was an established party that had been successful for years on end and about which most voters had made up their mind. The fact that we nonetheless find the predicted effects is telling. We speculate that this means that stigmatization of a party can be powerful when a new party is targeted that has little access to mass media. This is because a new party’s public image can still be shaped to a large extent by news media, and because a party without much mass media access cannot counteract the stigmatization by communicating its own side of the story to voters. However, let us emphasize once more that we have only investigated one case—and that, moreover, that case concerns a relatively new party. Other cases may yield effects that are different, or none at all. Furthermore, not only other cases but also other methods should be used, such as experimentation. This is because given the (survey) method used our possibilities to adequately control for rival explanations of the findings are limited. With this in mind, we turn to the implications of our study. Our research demonstrates the usefulness of modeling mediation and moderation effects, perhaps doing more justice to the complex realities than several previous studies. The DSMM model proves to be a particularly useful framework for doing so. By looking at a mechanism through which media affect vote choice (mediation), and by explicitly taking into account voter heterogeneity in our models (moderation), we go beyond the state of the art in studies of media coverage of anti-immigration parties, and in media effects. We integrate concepts used in political science (perceived party legitimacy; anti-immigrant attitude) into the DSMM framework. Similarly, the sociological trichotomy of silence, ridicule, and stigma (Ferree, 2005) seem to travel well to the realm of political parties, just as the protest paradigm concept (Chan & Lee, 1984)—at least, concerning the focus on the marginalization of a relatively new, outsider party by making stigmatizing aspects salient, and drawing attention away from its demands. Turning to our findings, they sit well with research on many voters being internally conflicted about anti-immigration parties (Blinder et al., 2013) and add to the existing knowledge on effects of media bias in measuring effects of undisputed negative media content about a political actor. The results suggest that effect sizes are modest. Yet, cumulative bias against a party may have a substantial negative impact—which brings us to the societal implications of our study. Our results may speak to societal debates on reactions to political extremism. The finding that systematically demonizing the PVV lowers its electoral appeal suggests that the party can be damaged electorally. Previous research has found that the PVV was hurt by such demonization, or, more precisely, its predecessor was (van Heerden & van der Brug, 2017). Together with our results, this suggests that media reactions are a potentially important weapon to fight the electoral rise of this party, and most likely its many counterparts across Europe as well. Of course, we do need to be careful here, as our study is limited to just one case (the PVV) at just one time point (in 2014). Furthermore, the stigmatization was indirect. Direct stigmatization may have a different effect because it may to a greater extent stick to the PVV. To be able to generalize these findings to direct stigmatization, or to other times and places, further research is encouraged. At the same time, not only anti-immigration parties can be targeted. All kinds of stigma are possible of all kinds of party. A striking empirical observation to illustrate this notion is that almost all Dutch parties have ever been called either “Nazi” or “fascist” in the news media, many of which regularly (van Heerden, 2014, p. 31). In most cases, such statements lack credibility and may therefore prove impotent in electoral terms. Yet, it is plausible to argue that many political ideas, or political parties carrying these ideas, could fall prey to a successful attempt to tarnish their reputation though news media. Joost van Spanje is an Associate Professor of Political Communication and Journalism as well as a member of ASCoR, University of Amsterdam. Rachid Azrout is a Lecturer of Political Communication and Journalism as well as a member of ASCoR, University of Amsterdam. Footnotes 1On the other hand, direct stigmatization may have a smaller effect, as an indirect message is more difficult to resist than direct stigmatization, which may be more easily dismissed by blaming it on political enemies or the news media being biased against a party. 2In the first wave, 2,189 respondents participated (AAPOR RR1 78.1%), 1,819 in the second wave (recontact rate 83.1%), 1,537 in the third wave (recontact rate 84.5%), and 1,379 in the fourth wave (recontact rate 89.7%). Because 112 respondents answered “don’t know” to the propensity to vote question in the third wave, in the fourth wave, or in both these waves, we are left with 1,267 respondents for whom we have full data on two time points. The samples show appropriate distributions in terms of gender, age, and education compared with census data. Panel attrition did not lead to a significant difference in the composition of the panel with regard to age and gender. The average level of education has slightly decreased between Wave 1 and Wave 4. As we are interested in underlying relationships and using fixed effects, we consider the deviations in the sample vis-à-vis the adult population less problematic. 3Telegraaf, NRC, NRC Next, Trouw, Algemeen Dagblad, De Volkskrant, Metro, and Spits. 4The LexisNexis Academic search string lists: “pvv” OR “partij voor de vrijheid” OR “wilders.” 5The LexisNexis Academic search string lists: “vlaams belang” OR “dewinter” OR “front national” OR “le pen” OR “lega nord” OR “bossi” OR “fpö” OR “fpo” OR “Strache” OR “Haider” OR “Zweden Democraten” OR “akesson.” 6The LexisNexis Academic search string lists: “extreemrechts!” OR “extreem-rechts!” OR “extreem rechts!” OR “discrimin!” OR “racis!” OR “radica!” OR “extremis!” OR “facis!” OR “neonazi” OR “neo-nazi” OR “anti-semi!” OR “antisemi!”. 7One coder coded the material and another a subsample (only between Waves 3 and 4, n = 69). Intercoder reliability was sufficient (91.3% agreement; Krippendorff’s α = .73). 8Although fixed effects are arguably the closest test to causality apart from experimental designs, we do need to note that in the absence of random assignment to the independent variable, our models are under the assumption that changes in political preferences (i.e., propensity to vote PVV) do not change media consumption behavior. One could, of course, argue that supporters of the PVV would choose to turn away from outlets that are critical of (i.e., demonize) the PVV. To address this concern, we tested whether propensity to vote for the PVV predicted change in media consumption, and found that this was not the case. Also, we ran our models not using our exposure to demonization measure but a raw exposure measure, and found that raw exposure has neither a main effect on propensity to vote for the PVV nor a moderated one. This shows that it is not just the outlet that voters use that matters, but the actual content voters are exposed to. Given that individual voters do not have any direct influence on news media content, these findings support our assumptions underlying our models (results available on request from the authors). 9To check the possibility that the effect is a knowledge effect, we rerun the analyses using knowledge (five multiple choice items) instead of anti-immigrant attitudes. With propensity to vote as the dependent variable and not controlling for legitimacy, the interaction between exposure and knowledge is not significant (b = 0.03, se = 0.91, p = .487). Having both anti-immigrant attitudes and knowledge as moderators leads to an insignificant interaction of knowledge (b = −0.13, se = 0.91, p = .445), while the interaction of anti-immigrant attitudes remains significant (b = −1.45, se = 0.86, p = .046). Although we cannot conclusively test this given data limitations, these findings imply that our results are not just the results of a knowledge effect. References Allison P. D. ( 2009 ). Fixed-effects regression models . Thousand Oaks, CA : Sage . Google Scholar Crossref Search ADS Art D. ( 2006 ). The politics of the Nazi past in Germany and Austria . New York, NY : Cambridge University Press . Baron R. M. , Kenny D. A. ( 1986 ). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations . Journal of Personality and Social Psychology , 51 , 1173 – 1182 . Google Scholar Crossref Search ADS Bennett W. L. ( 1990 ). Toward a theory of press-state relations in the United States . Journal of Communication , 40 , 103 – 127 . Google Scholar Crossref Search ADS Bennett W. L. ( 2016 ). News: The politics of illusion ( 10th ed. ). Chicago : University of Chicago Press . Google Scholar Crossref Search ADS Beullens K. , van den Bulck J. ( 2013 ). Predicting young drivers' car crashes: The role of music video viewing and the playing of driving games. Results from a prospective cohort study . Media Psychology , 16 , 88 – 114 . doi: 10.1080/15213269.2012.754582. Google Scholar Crossref Search ADS Beyens I. , Vandenbosch L. , Eggermont S. ( 2015 ). Early adolescent boys’ exposure to internet pornography. Relationships to pubertal timing, sensation seeking, and academic performance . The Journal of Early Adolescence , 35 , 1045 – 1068 . doi: 10.1177/0272431614548069. Google Scholar Crossref Search ADS Biorcio R. ( 2003 ). The Lega Nord and the Italian media system. In Mazzoleni G. , Stewart J. , Horsfield B. (Eds.), The media and neo-populism: A contemporary comparative analysis (pp. 71 – 94 ). Westport, CT : Praeger . Birenbaum G. , Villa M. ( 2003 ). The media and neo-populism in France. In Mazzoleni G. , Stewart J. , Horsfield B. (Eds.), The media and neo-populism: A contemporary comparative analysis (pp. 45 – 70 ). Westport, CT : Praeger . Blinder S. , Ford R. , Ivarsflaten E. ( 2013 ). The better angels of our nature: How the antiprejudice norm affects policy and party preferences in Great Britain and Germany . American Journal of Political Science , 57 , 841 – 857 . Boomgaarden H. , Vliegenthart R. ( 2007 ). Explaining the rise of anti-immigrant parties: The role of news media content in The Netherlands . Electoral Studies , 26 , 404 – 417 . Google Scholar Crossref Search ADS 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 . Google Scholar Crossref Search ADS Bos L. , van der Brug W. , De Vreese C. H. ( 2011 ). How the media shape perceptions of right-wing populist leaders . Political Communication , 28 , 182 – 206 . Google Scholar Crossref Search ADS Boykoff J. ( 2006 ). The repression of dissent . New York, NY : Routledge . Burscher B. , van Spanje J. H. P. , De Vreese C. H. ( 2015 ). Owning the issues of crime and immigration: The relation between immigration and crime news and anti-immigrant voting in 11 countries . Electoral Studies , 38 , 59 – 69 . Google Scholar Crossref Search ADS Chan J. M. , Lee F. L. F. ( 1984 ). Journalistic paradigms on civil protest: A case study of Hong Kong. In Arno A. , Dissanayake W. (Eds.), The news media in national and international conflict (pp. 183 – 202 ). Boulder, CO : Westview Press . Chong D. , Druckman J. ( 2007 ). Framing theory . Annual Review of Political Science , 10 , 103 – 126 . Google Scholar Crossref Search ADS Cohen S. ( 1972 ). Folk Devils and Moral Panics. The Creation of the Mods and Rockers . London : MacGibbon & Kee . Converse P. E. ( 1962 ). Information flow and the stability of partisan attitudes . Public Opinion Quarterly , 26 , 578 – 599 . Google Scholar Crossref Search ADS D’Alessio D. , Allen M. ( 2000 ). Media bias in presidential elections: A meta-analysis . Journal of Communication , 50 , 133 – 156 . Google Scholar Crossref Search ADS De Vreese C. H. ( 2012 ). New avenues for framing research . American Behavioral Scientist , 56 , 365 – 375 . Google Scholar Crossref Search ADS De Vreese C. H. , Azrout R. , Möller J. E. ( 2014 ). 2014 European Parliament Election Campaign Study: Data and Documentation . Amsterdam : University of Amsterdam . Della Vigna S. , Kaplan E. ( 2007 ). The Fox News effect: Media bias and voting . The Quarterly Journal of Economics , 122 , 1187 – 1234 . Google Scholar Crossref Search ADS Dilliplane S. , Goldman S. K. , Mutz D. C. ( 2013 ). Televised exposure to politics: New measures for a fragmented media environment . American Journal of Political Science , 57 , 236 – 248 . Google Scholar Crossref Search ADS Druckman J. N. , Parkin M. ( 2005 ). The impact of media bias: How editorial slant affects voters . The Journal of Politics , 67 , 1030 – 1049 . Google Scholar Crossref Search ADS Edelman M. ( 1977 ). Political language: Words that succeed and policies that fail . New York, NY : Academic Press . Ellinas A. A. ( 2010 ). The media and the far right in Western Europe. Playing the nationalist card . New York, NY : Cambridge University Press . Google Scholar Crossref Search ADS Entman R. M. ( 1993 ). Framing: Toward clarification of a fractured paradigm . Journal of Communication , 45 , 51 – 58 . Google Scholar Crossref Search ADS Entman R. M. ( 2007 ). Framing bias: Media in the distribution of power . Journal of Communication , 57 , 163 – 173 . Google Scholar Crossref Search ADS Ferree M. M. ( 2005 ). Soft repression: Ridicule, stigma, and silencing in gender-based movements. In Davenport C. , Johnston H. , Mueller C. (Eds.), Repression and mobilization (pp. 138 – 156 ). Minneapolis, MN : University of Minnesota Press . Gamson W. A. , Modigliani A. ( 1989 ). Media discourse and public opinion on nuclear power: A constructionist approach . American Journal of Sociology , 95 ( 1 ), 1 – 37 . Google Scholar Crossref Search ADS Gans H. J. (2004) [1979] . Deciding what’s news. A study of CBS evening news, NBC nightly news, Newsweek, and time . Evanston, IL : Northwestern University Press . Gitlin T. ( 1980 ). The whole world is watching: Mass media in the making & unmaking of the new left . Berkeley, CA; Los Angeles, CA; London : University of California Press . Groeling T. ( 2010 ). When politicians attack: Party cohesion ni the media . New York, NY : Cambridge University Press . Google Scholar Crossref Search ADS Groeling T. ( 2013 ). Media bias by the numbers . Annual Review of Political Science , 16 , 129 – 151 . Google Scholar Crossref Search ADS Hardisty D. J. , Johnson E. J. , Weber E. U. ( 2010 ). A dirty word or a dirty world? Attribute framing, political affiliation and query theory . Psychological Science , 21 , 86 – 92 . DOI: 10.1177/0956797609355572. Google Scholar Crossref Search ADS Hopmann D. N. , van Aelst P. , Legnante G. ( 2012 ). Political balance in the news: A review of concepts, operationalizations and key findings . Journalism , 13 ( 2 ), 240 – 257 . Google Scholar Crossref Search ADS Hopmann D. N. , Vliegenthart R. , De Vreese C. H. , Albaek E. ( 2010 ). Effects of election news coverage: How visibility and tone influence party choice . Political Communication , 27 , 389 – 405 . DOI: 10.1080/10584609.2010.516798. Google Scholar Crossref Search ADS Iyengar S. , Kinder D. ( 1987 ). News that matters: Television and American opinion . Chicago : The University of Chicago Press . Kleinnijenhuis J. , Hoof A. M. J., van , Oegema D. , De Ridder J. A. ( 2007 ). A test of rivaling approaches to explain news effects: News on issue positions of parties, real-world developments, support and criticism, and success and failure . Journal of Communication , 57 , 366 – 384 . Google Scholar Crossref Search ADS Korteweg A. , Yurdakul G. ( 2014 ). The headscarf debates: Conflicts of national belonging . Stanford, CA : Stanford University Press . Lee F. L. F. ( 2014 ). Triggering the protest paradigm: Examining factors affecting news coverage of protests . International Journal of Communication , 8 , 2725 – 2746 . Levin I. P. , Schneider S. L. , Gaeth G. J. ( 1998 ). All frames are not created equal: A typology and critical analysis of framing effects . Organizational Behavior and Human Decision Processes , 76 , 149 – 188 . Google Scholar Crossref Search ADS Linden A. , Klandermans B. ( 2006 ). Stigmatization and repression of extreme-right activism in The Netherlands . Mobilization , 11 , 213 – 228 . Link B. G. , Phelan J. C. ( 2001 ). Conceptualizing stigma . Annual Review of Sociology , 27 , 363 – 385 . Google Scholar Crossref Search ADS Lubbers M. , Gijsberts M. , Scheepers P. ( 2002 ). Extreme right-wing voting in Western Europe . European Journal of Political Research , 41 , 345 – 378 . Google Scholar Crossref Search ADS Major B. , O’Brien L. T. ( 2005 ). The social psychology of stigma . Annual Review of Psychology , 56 , 393 – 421 . Google Scholar Crossref Search ADS Mazzoleni G. ( 2003 ). The media and the growth of neo-populism in contemporary democracies. In Mazzoleni G. , Stewart J. , Horsfield B. (Eds.), The media and neo-populism: A contemporary comparative analysis (pp. 1 – 20 ). Westport, CT : Praeger . Molotch H. ( 1979 ). Media and movements. In Zald M. N. , McCarthy J. D. (Eds.), The dynamics of social movements: Resource mobilization, social control, and tactics (pp. 71 – 93 ). Cambridge, MA : Winthrop Publishers . Nelson T. E. , Oxley Z. M. ( 1999 ). Issue framing effects on belief importance and opinion . Journal of Politics , 61 , 1040 – 1067 . Google Scholar Crossref Search ADS Nelson T. E. , Clawson R. A. , Oxley Z. M. ( 1997 ). Media framing of civil liberties conflict and its effect on tolerance . American Political Science Review , 91 , 567 – 583 . Google Scholar Crossref Search ADS Norris P. , Curtice J. , Sanders D. , Scammell M. , Semetko H. A. ( 1999 ). On message: Communicating the campaign . London : Sage . Oegema D. , Kleinnijenhuis J. ( 2000 ). Personalization in political television news: A 13-wave survey study to assess effects of text and footage . Communications , 25 , 43 – 60 . Google Scholar Crossref Search ADS Paletz D. L. , Entman R. M. ( 1981 ). Media, power, politics . New York, NY; London : The Free Press . Plasser F. , Ulram P. A. ( 2003 ). Striking a responsive chord: Mass media and right-wing populism in Austria. In Mazzoleni G. , Stewart J. , Horsfield B. (Eds.), The media and neo-populism: A contemporary comparative analysis (pp. 21 – 43 ). Westport, CT : Praeger . Preacher K. J. , Rucker D. D. , Hayes A. F. ( 2007 ). Assessing moderated mediation hypotheses: Theory, methods, and prescriptions . Multivariate Behavioral Research , 42 , 185 – 227 . Google Scholar Crossref Search ADS Prior M. ( 2007 ). The immensely inflated news audience: Assessing bias in self-reported news exposure . Public Opinion Quarterly , 73 , 130 – 143 . Google Scholar Crossref Search ADS Schafraad P. , d’Haenens L. , Scheepers P. , Wester F. ( 2012 ). The evolution of a media image: Newspaper attention to the Flemish Far Right 1987-2004 . Acta Politica , 47 , 356 – 377 . Google Scholar Crossref Search ADS Schafraad P. , Scheepers P. , Wester F. ( 2009 ). Media attention to Pim Fortuyn and LPF during the 2002 elections. Negative bias or following news routines? In Konig R. P , Huysmans F. J. M. , Nelissen P. W. M. (Eds.), Meaningful media. Communication research on the social construction of reality (pp. 188 – 200 ). Nijmegen : Tandem Felix . Schemer C. , Wirth W. , Matthes J. ( 2012 ). Value resonance and value framing effect on voting intentions in direct-democratic campaigns . American Behavioral Scientist , 56 , 334 – 352 . DOI: 10.1177/000276421142632. Google Scholar Crossref Search ADS Shah D. V. , Domke D. , Wackman D. B. ( 1996 ). “To thine own self be true.” Values, framing, and voter decision-making strategies . Communication Research , 23 , 509 – 560 . doi: 10.1177/009365096023005001. Google Scholar Crossref Search ADS Shoemaker P. J. ( 1984 ). Media treatment of deviant political groups . Journalism Quarterly , 61 , 66 – 75 . Google Scholar Crossref Search ADS Tillie J. ( 2008 ). Gedeeld land. Het multiculturele ongemak van Nederland. (Divided country. The discomfort of the Dutch multicultural society) . Amsterdam : Meulenhoff . Valkenburg P. M. , Peter J. ( 2013 ). The differential susceptibility to media effects model . Journal of Communication , 63 , 221 – 243 . DOI: 10.1111/jcom.12024. Google Scholar Crossref Search ADS van der Brug W. , Fennema M. ( 2007 ). What causes people to vote for a radical right party? A review of recent work . International Journal of Public Opinion Research , 19 , 474 – 487 . Google Scholar Crossref Search ADS van der Eijk C. , Niemöller B. ( 1984 ). Het Potentiele Electoraat van de Nederlandse Politieke Partijen . Beleid en Maatschappij , 11 , 192 – 204 . van der Eijk C. , van der Brug W. , Kroh M. , Franklin M. N. ( 2006 ). Rethinking the dependent variable in voting behavior: On the measurement and analysis of electoral utilities . Electoral Studies , 25 , 424 – 447 . Google Scholar Crossref Search ADS van Heerden S. C. ( 2014 ). What did you just call me? A study on the demonization of Dutch political parties between 1995 and 2011 . Amsterdam : University of Amsterdam . van Heerden S. C. , van der Brug W. ( 2017 ). Demonisation and electoral support for populist radical right parties: A temporary effect . Electoral Studies , 47 , 36 – 45 . Google Scholar Crossref Search ADS van Spanje J. H. P. , De Vreese C. H. ( 2014 ). Europhile media and eurosceptic voting: Effects of news media coverage on eurosceptic voting in the 2009 European parliamentary elections . Political Communication , 31 , 325 – 354 . Google Scholar Crossref Search ADS van Spanje J. H. P. , De Vreese C. H. ( 2015 ). The good, the bad, and the voter. The impact of hate speech prosecution of a politician on the electoral support for his party . Party Politics , 21 , 115 – 130 . Google Scholar Crossref Search ADS van Spanje J. H. P. , Weber T. ( 2017 ). Does Ostracism Affect Party Support? Comparative Lessons and Experimental Evidence . Party Politics (Advance Online Publication September 12, 2017). van Zoonen E. A. ( 1992 ). The women’s movement and the media: Constructing a public identity . European Journal of Communication , 7 , 453 – 476 . Google Scholar Crossref Search ADS Vliegenthart R. , Boomgaarden H. G. , van Spanje J. H. P. ( 2012 ). Anti-immigrant party support and media visibility: A cross-party over-time perspective . Journal of Elections, Public Opinion and Parties , 22 , 315 – 358 . Google Scholar Crossref Search ADS Walgrave S. , De Swert K. ( 2004 ). The making of the (issues of the) Vlaams Blok: On the contribution of the news media to the electoral success of the Belgian extreme-right party . Political Communication , 21 , 479 – 500 . Google Scholar Crossref Search ADS Weaver D. A. , Scacco J. M. ( 2013 ). Revisiting the protest paradigm: The tea party as filtered through prime-time cable news . The International Journal of Press/Politics , 18 , 61 – 84 . Google Scholar Crossref Search ADS Zaller J. ( 1989 ). Bringing converse back in: Modeling information flow in political campaigns . Political Analysis , 1 , 181 – 234 . Google Scholar Crossref Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of The World Association for Public Opinion Research. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

International Journal of Public Opinion ResearchOxford University Press

Published: Mar 26, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off