Article The radical right, the labour movement and the competition for the workers' vote MOSIMANN, Nadja, RENNWALD, Line, ZIMMERMANN, Adrian Abstract This article analyses the capacity of radical right parties to attract support from un-ion members in recent elections in Western Europe. It is argued that unionized vot-ers resist the appeals of the radical right better than non-union members. Using data from the European Social Survey 2010–2016, the article shows that union members are overall less likely to vote for the radical right than non-union members. Even though it is found that unionized working-class and middle-class voters are less like-ly to vote radical right than their non-unionized peers in the pooled sample, it is also observed that these subgroups of unionized voters and especially unionized working-class voters are not immune to radical right voting in all the countries analysed. The article thus indicates a growing capacity of the radical right to attract unionized working-class segments of the electorate in some countries and to directly compete with left parties for these voters. Reference MOSIMANN, Nadja, RENNWALD, Line, ZIMMERMANN, Adrian. The radical right, the labour movement and the competition for the workers' vote. Economic and Industrial Democracy, 2018, p. 0143831X1878031 DOI : 10.1177/0143831X18780317 Available at: http://archive-ouverte.unige.ch/unige:108523 Disclaimer: layout of this document may differ from the published version. Mosimann, N., Rennwald, L. and Zimmermann, A., 2018. The Radical Right, the Labour Movement and the Competition for the Workers’ Vote. Economic and Industrial Democracy. Online first August 2018, https://doi.org/10.1177/0143831X18780317 Postprint version The radical right, the labour movement, and the competition for the workers’ vote Nadja Mosimann, Universities of Zurich and Geneva Line Rennwald, University of Geneva Adrian Zimmermann Abstract This article analyses the capacity of radical right parties to attract support from union members in recent elections in Western Europe. It is argued that unionized voters resist the appeals of the radical right better than non-union members. Using data from the European Social Survey 2010–2016, the article shows that union members are overall less likely to vote for the radical right than non-union members. Even though it is found that unionized working-class and middle-class voters are less likely to vote radical right than their non-unionized peers in the pooled sample, it is also observed that these subgroups of unionized voters and especially unionized working-class voters are not immune to radical right voting in all the countries analysed. The article thus indicates a growing capacity of the radical right to attract unionized working-class segments of the electorate in some countries and to directly compete with left parties for these voters. Mosimann, Rennwald, and Zimmermann (2018) 2 Introduction It was the event between the two rounds of the 2017 French presidential election: In a well-orchestrated appearance, Marine Le Pen, presidential candidate from the radical right Front National, went to the Whirlpool plant in Amiens to meet with strike pickets three days after she had been qualified for the second round of the election. In January 2017, the management of Whirlpool had announced the delocalisation of their production of tumble dryers to Poland and had scheduled the closure of the French plant in Amiens for June 2018. The crux of the matter was that Le Pen’s centrist-liberal opponent for the second round, Emmanuel Macron, had held a meeting with trade unions and representatives of workers in the city centre of Amiens that same morning. In front of the plant, Marine Le Pen (unjustly) claimed that the meeting in the city centre was not representative and declared her solidarity with the striking workers who were “resisting against rampant globalization and a shameful economic model.”1 She then went on to criticize her opponent for not actually coming to the plant. Eventually, Macron reacted to Le Pen’s accusations by also visiting the plant and speaking to workers on the picket line. In our view, the highly mediatized event in Amiens perfectly illustrates the attitude of the radical right towards workers and trade unions that is at the centre of our article. Even though Marine Le Pen’s visit to Amiens might seem like random political opportunism at first glance, it is not a coincidence that she labelled trade unions – with whom Macron was meeting – as not representative of workers only to simultaneously claim solidarity with these workers. In this article, we will characterize the attitude of the radical right towards workers and trade unions as consisting exactly of such attacks ________________________ 1 BFMTV, Marine Le Pen sur le site de Whirlpool [Video file] (last accessed, 26th April 2017), available at: http://www.bfmtv.com/mediaplayer/video/marine-le-pen-sur-le-site-de-whirlpool-je-suis-ici-a-maplace-exactement-ou-je-dois-etre-938359.html. 3 Union membership effects on radical right voting on labour movement organisations while pretending to defend workers' interests and appropriating some of the labour movement’s demands. In doing so, we are above all interested in the electoral implications of this strategy of the radical right. More precisely, we focus on the analysis of vote choice among union members and non-union members at the individual level. By exploring whether or not union membership affects citizens’ propensity to vote for radical right-wing parties and how such union membership effects differ across social classes and countries, we contribute to the investigation of the relationship between trade unions and democracy that is at the core of this special issue. Existing literature has led to ambiguous findings on the matter – especially with regard to radical right voting among unionized working-class voters. Moreover, trade unions themselves are worried that the political offer of radical rightwing parties might generate sympathies for them among union members (see Stöss, 2017). Contrary to popular wisdom, we argue that unionized workers resist the radical right’s appeals to at least some degree and in some countries, while simultaneously assuming that the radical right manages to expand its vote shares among non-unionized workers who form a growing constituency in the context of de-unionization. Our contribution to the literature is two-fold: First, we contribute to the literature on the relationship between union membership, political behaviour, and the democratic representation of workers’ interests. Previous research has led to inconclusive findings where the relative ability of the radical right to attract unionized segments of the electorate is concerned. More broadly, we also aim at contributing to the research on trade unions as producers of specific norms that shape political attitudes and political behaviour in democratic societies (e.g., Mosimann and Pontusson, 2017). Second, we contribute to the large literature on radical right parties and their voters’ sociology by focusing our analysis on the demand-side of voters. We also wish to inform the literature on the supply-side of parties’ policy positions and strategies by characterizing the attitude of the radical right towards the labour movement and describing (in a more Mosimann, Rennwald, and Zimmermann (2018) 4 explicit way than previous research) why the political offers of these two movements are in competition.2 By doing so, the article particularly strengthens the literature dealing with the socio-economic agenda of the radical right (e.g., Afonso, 2015; Afonso and Rennwald, 2018; De Lange, 2007; Röth et al., 2017). The next section of this article discusses the attitude of the radical right towards workers and its organisations and explains why the political offers from the radical right and the labour movement are in competition with each other. We then review the literature on the relationship between union membership and radical right voting. In a next section, we explain why we expect union membership to matter to radical right voting and present our hypotheses. After informing about our methodological choices, we analyse the relative electoral success or failure of the radical right’s strategy at the individual level on the basis of European Social Survey (ESS) data for 2010 to 2016. A concluding section summarizes and discusses our findings. Undermining solidarity among workers versus forging solidarity among workers Through their policy positions and rhetoric, radical right parties (explicitly or implicitly) undermine key principles of the labour movement that consist in promoting and defending solidarity among workers irrespective of nationality and origin. Therefore, the political offers of the radical right and the labour movement (that is, trade unions and the political arm of the labour movement) differ radically from each other. At the same time, however, radical right parties also pretend to take care of (certain) workers’ interests. The strategy of the radical right thus consists of attacking labour movement ________________________ 2 We understand all the organisations that were founded by the working class to serve its emancipation as making up the labour movement: Trade unions that defend workers’ rights on the shop floor and in the economic sphere (the economic arm of the labour movement) and socialist parties that promote workers’ interests in the political sphere (the political arm of the labour movement), as well as several structures of cooperatives, self-help organizations, and cultural and sports associations. 5 Union membership effects on radical right voting organisations and undermining their core principles while simultaneously appropriating and re-using some of their demands and struggles. In this section, we will first discuss the radical right’s strategy and why the political offers from the radical right and the labour movement are mutually exclusive. In a second step, we will take a look at history and show the existence of some similarities (but also important differences) between the contemporary radical right’s strategy towards the labour movement and those of fascist movements and regimes in the interwar period.3 Trade unions and the radical right: Two mutually exclusive political offers Radical right parties all over Europe frequently (but to varying degrees) pretend to defend the interests of workers in a way linked to key characteristics of this party family, namely their nativist ideology, their anti-immigration agenda, and their populist rhetoric (e.g., Mudde, 2007). First and foremost, pretending to defend the interests of workers does not take on a universal form, but is restricted to the defence of native workers against migrants. Furthermore, the rhetoric of radical right parties around the pretence to defend workers’ interests is reminiscent of when they talk of themselves as advocates of the “common man” against corrupt political elites. Moreover, the radical right’s discourse on the defence of native workers is also closely linked to their chauvinist welfare ideology, which aims at restricting welfare state benefits to native people (Andersen and Bjørklund, 1990; Kitschelt, 1995). To a varying degree across countries and in combination with elements of economic nationalism, elements of welfare chauvinism have been present in the policy positions of radical right parties for a long time (Mudde, 2007). Over the last years, welfare chauvinism has, however. gained importance in the agenda of the radical right. Radical right parties ________________________ 3 Disassociating themselves from “old” extreme right parties constitutes a key defining characteristic of the new radical right in the scholarly literature (e.g., Ignazi, 1992; Carter, 2005; Mudde, 2007). Mosimann, Rennwald, and Zimmermann (2018) 6 increasingly address welfare state issues in their manifestos and seem to position themselves more and more pro-welfare (chauvinism) in numerous countries (Afonso and Rennwald, 2018; Lefkofridi and Michel, 2017). Even the radical right parties with the clearest neoliberal outlook like the SVP in Switzerland nowadays include elements of welfare chauvinism in their party program (e.g., Mazzoleni, 2008) while also heavily emphasizing the notion of “deservingness” by differentiating between those who deserve social benefits (e.g., pensioners) and those who do not (e.g., the unemployed) (Afonso and Papadopoulos, 2015; see also Ennser, 2016; Van Oorschot, 2006).4 Conversely, trade unions have promoted the idea of solidarity among workers irrespective of nationality, origin, race, or gender since its beginning.5 This core principle is at odds with the exclusionary values and the nativist ideology of the radical right that separates members of an alleged homogeneous national community from all “others” who do not belong to it and supposedly threaten it (e.g., foreigners or minorities). While the labour movement aims at building and reinforcing solidarity among workers, the radical right exploits tensions and creates new divisions among them and thus undermines working-class solidarity.6 Both the radical right and the labour movement hence try to mobilize workers as voters. However, they do so with competing and mutually exclusive offers in terms of policies: Radical right parties, with their anti-immigration agenda, pretend to protect ________________________ In the case of the nationalist New-Flemish Alliance (N-VA) that advocates neoliberal positions, scholars have advanced the concept of “welfare producerism” to capture its socio-economic agenda. It is based on the idea of a hardworking middle class producing the wealth of the nation and thus deserving support from the welfare state (Abts and Kochuyt, 2013; Abts et al., 2016). 5 In their policies and practices, trade unions have sometimes strayed from these principles. It seems, however, difficult to find examples where this core principle has been put into question by the socialist labour movement. In his analysis of the positions of trade unions towards immigration, Donnelly (2016) shows that both ideologically oriented and more pragmatic unions have tended to become more pro-immigration over time. 6 We understand working-class solidarity in a broad sense, that is, as a co-operative form of representing common interests through trade unions and left-wing politics. Besides the working class in a more restricted sense, also segments of what is commonly described as middle class can be involved in building working class solidarity. 4 7 Union membership effects on radical right voting native workers from competition from migrants by restricting migration (and migrants’ rights), whereas the labour movement, that is, trade unions and their allied socialist parties, want to regulate competition among workers by more internal regulations such as minimum wages, working time regulations, and improved labour standards (therefore giving migrants the same rights as natives; see also Donnelly, 2016). “Destroy and co-opt” – Classical fascism and the labour movement Because radical right parties and parties linked to the labour movement compete for workers’ vote, it is not surprising that radical right parties criticize the legitimacy of trade unions. As in the case of Marine Le Pen in France, radical right parties seek to represent workers, but attack the official and recognized organisations of these workers. It is (and often has been) the case that radical right parties accuse labour leaders of betraying the interests of native workers and try to use internal tensions in the labour movement to weaken it. However, the attempts of the contemporary radical right to undermine the legitimacy of labour movement organisations does not normally take on the form of physical attacks. Conversely, the fascist movements of the interwar period – from which contemporary radical right parties generally distance themselves – violently attacked trade unionists, socialists, and communists.7 Moreover, once they had been handed over dictatorial state power, Italian and German fascists violently repressed the labour movement. Independent trade unions, strikes, and free collective bargaining were outlawed, and thousands of trade union leaders, shop-stewards, and activists of the labour parties imprisoned without trial. ________________________ 7 In a notorious terror campaign in Northern and Central Italy in 1921 and 1922, fascist paramilitary groups (black shirts or squadristi) physically attacked trade union members and killed hundreds of them. Moreover, in these “punitive expeditions,” the fascists destroyed offices, printing presses, and cultural centres of the labour organisations and violently forced socialist municipal councillors to step down. Similarly, during the early 1930s, the Sturmabteilung (SA), the paramilitary wing of the Nazi party, terrorised working-class neighbourhoods in Germany (Reichardt, 2002: 125-129). Mosimann, Rennwald, and Zimmermann (2018) 8 Interestingly enough, fascists simultaneously tried to destroy the democratic labour movement in the interwar period and appropriate workers’ demands. As the Italian scholar Brunello Mantelli (2004: 44) has argued, fascists wanted to “destroy and coopt” the labour movement. Fascists thus always explicitly tried to appeal to workers and to win them over with pseudo-socialist elements in their rhetoric and by re-using and diverting symbols and principles of the labour movement. The name chosen by the German Nazi party – Nationalsozialistische Deutsche Arbeiterpartei – indicates that this party claimed to be socialist and a workers’ party. Furthermore, on 1st May 1933, the Nazis had managed to transform the mayday rallies of the labour movement – one of the most important symbols of socialist internationalism – into a “day of national labour” (Wildt, 2014).8 It is also interesting to note the existence of so-called “yellow” workers’ organisations. Active mainly at the beginning of the 20th Century, these organisations propagated a vague programme of class harmony and were involved in systematic and violent strike breaking. They were often funded by large industrial companies (such as Schneider in France, Krupp and Siemens in Germany, and Sulzer in Switzerland) as part of a strategy aimed at defeating trade unions (Gruner, 1988: 829-836). These organisations have been described as one of the fore-runners of fascism in preWorld War I Europe (Sternhell, 2000: 163-166, 304-307) and several personal connections existed between the leaders of “yellow unions” and the extreme right.9 ________________________ The decisive attack of the Nazis against trade unions took place just a day later, on 2nd May 1933. The SA occupied and vandalised trade union buildings, and mistreated and arrested trade union officials. 9 For example, Pierre Biétry, one of the leaders of the French yellow unions, also founded a short lived “national socialist” party and propagated ultra-nationalist and anti-Semitic views (Arnold, 1999; Maillard, 2008). The railway worker Anton Drexler, founder of a small ultra-nationalist “workers party” in Munich that later became Hitler’s NSDAP, also had a background in the yellow union movement (Stegmann, 1971: 410-411). 8 9 Union membership effects on radical right voting Immune or not to the radical right? Class, union membership, and radical right voting Is the strategy of the radical right successful or not at the ballot box? At least workingclass voters have been shown to not have remained immune to the radical right’s strategy of pretending to defend the interests of (native) working-class voters. Research on the class basis of party families indicate that radical right parties have gained traction among working-class voters from the mid-1980s onwards (Afonso and Rennwald, 2018; Arzheimer, 2013; Rydgren, 2013), while social democratic parties have suffered losses among these voters (Arndt, 2013; Gingrich and Häusermann, 2015; Rennwald, 2015). At the beginning of the 21st Century, the working class has become the “contested stronghold” of the left and the radical right, and old and new patterns of class voting co-exist (Oesch and Rennwald, 2017). The capacity of the radical right to attract working-class votes is well-established in the literature. We know, however, less about the penetration of organised segments of the working class. The existing literature is unclear on whether or not union membership makes a difference to radical right voting among the working class, and this article is particularly interested in this question. The inconclusive findings with regard to a union membership effect on radical right voting among the working class is particularly puzzling when we look at history. During the inter-war period, the strength of the socialist labour movement was the main reason for the relative “immunity” of workers against fascist appeals. Classic studies of the electorate of the Nazi party (Geiger, 1930; Lipset, 1960) conclude that it was essentially a party of the old (small and medium business owners) and new (white-collar workers) middle classes.10 In relatively backward and agricultural Italy, the middle- and upper-class bias of the fascist membership and electorate was even more evident than in highly industrialised Germany (Petersen, 1983). ________________________ 10 More recent publications (Hamilton, 1982; Childers, 1983; Falter, 1991) have put this into question and emphasise that the German Nazi party drew support from all social classes. However, nobody seriously contests that blue-collar workers were underrepresented in the fascist electorate (Geary, 2002). Mosimann, Rennwald, and Zimmermann (2018) 10 Today, the literature on the link between union membership and vote choice presents consistent findings with regard to left-wing parties which continue to receive important support from trade union members (Arndt and Rennwald, 2016; Mosimann, 2017; Rennwald, 2015). Findings are less straightforward when it comes to radical right parties and existing research shows some variation in the “union membership effect” on voting for the radical right depending on the countries analysed, the specific social classes focused on, and the analytical strategy chosen. Against this backdrop, we aim to empirically establish if union membership reduces voters’ support for the radical right, and if the effect holds for different social classes and in different countries. Before turning to the hypotheses, let us briefly summarize the existing literature on the link between trade union membership and voting for the radical right. When contrasting support for radical right parties with support for social democratic parties in sixteen West European countries based on European Social Survey (ESS) data for 2002 to 2010, Arndt and Rennwald (2016) show that union members are significantly less likely to support the radical right (see also Arndt and Rennwald, 2017) and that the membership effect is even slightly bigger when contrasting the radical right with the left than when contrasting the centre-right with the left. In a case study of Sweden, Arndt and Rennwald find, however, also signs of support for the Sweden Democrats among members of the blue-collar confederation LO in 2006 and 2010. This hints at some cross-national variation when it comes to union members’ party choice. In an article testing different explanations of radical right voting based on ESS data for 2002, Oesch (2008) finds contradictory results regarding the effect of union membership. Being a union member reduces the likelihood to vote radical right (slightly) in Austria, Norway, and Switzerland, has no effect in France, and even (slightly) increases support for the radical right in Belgium. Finally, Mosimann (2017) analyses union 11 Union membership effects on radical right voting members’ party choice based on data from the Comparative Study of Electoral Systems (CSES) for seven West European countries for 2011 to 2017 and does not find that union membership consistently affects radical right voting. Interestingly, she finds a clear effect among citizens simultaneously supporting redistribution and opposing immigration. Among these voters with leftist redistribution preferences and rightist immigration attitudes, union membership makes an important difference and clearly reduces the likelihood to support the radical right. Existing findings are even more ambiguous if one focuses exclusively on the working class. Using data from the ESS round 4 in 2008 for fourteen West European countries, Bornschier and Kriesi (2013) analyse the determinants of radical right voting among the manual working class. They do not find a general significant effect of union membership on the propensity to support the radical right. When contrasting support for the radical right with support for the mainstream left, they observe, however, that skilled unionized workers are less likely to vote for the radical right than non-unionized ones. This is not the case for unionized routine operatives and the authors suggest that only the core of the manual working class is immunized from radical right voting. Also analysing the determinants of radical right voting among the working class based on ESS data from 2002 to 2008, Rennwald (2015) shows an important effect of union membership in Switzerland. Non-unionized working-class voters are more likely to support the Swiss People’s Party, while unionized working-class voters prefer the Swiss Socialist Party. In Austria, non-unionized working-class voters are also more likely to support the FPÖ than unionized ones, but the effect is less marked than in Switzerland. These results indicate – again – some cross-national variation in the effect of union membership on radical right voting. Mosimann, Rennwald, and Zimmermann (2018) 12 Why union membership matters to radical right voting Why should union members be better equipped to resist the appeals of the radical right than non-union members? It is possible to argue that the politics and strategies of trade unions matter or that union members have some specific characteristics that distinguish them from non-union members. One obvious reason why union members might refrain from voting radical right results from the idea that organisations generate and diffuse specific values, and that members of such organisations are somehow exposed to these values. For this line of reasoning, we draw on the work by Mosimann and Pontusson (2017) who emphasise the role of trade unions in the formation of (redistribution) preferences. They argue that the behaviour and rhetoric of unions create norms that influence the preferences and behaviour of their members. One of the main principles of the labour movement since its beginning is the idea of solidarity among workers irrespective of nationality, origin, race, or gender. By diffusing and reinforcing values of solidarity among their members, trade unions can therefore act as powerful counter-agents to the exclusionary ideology of the radical right. Moreover, union members are not individually exposed to the propaganda of the radical right, because trade unions offer their members a space to collectively discuss these issues and exchange points of views (see Iversen and Soskice, 2015, on unions’ role as political discussion networks). Finally, unions are organisations in which foreign workers can become members and this creates opportunities for exchanges between workers regardless of origin and nationality.11 If one adopts a “bottom-up” perspective, one could argue that workers decision to join a union might not be entirely independent from pre-existing ideological convictions (see Ebbinghaus et al., 2011, for a review of decisions to join unions). It is thus possible that at least some voters who feel close to the goals of the labour movement and oppose those of the radical right self-select into trade unions. Especially in a period ________________________ 11 Yet, migrant workers remain on average less organised than native workers (Kranendonk and de Beer, 2016). 13 Union membership effects on radical right voting in which employers’ attitudes towards trade unions have become more hostile, ideological motivations to join a union may become more important. In the only existing study based on panel data that allows analysing the causal impact of union membership, Hadziabdic and Baccaro (2017) indeed find selection effects in the relationship between union membership and political attitudes. Studying various political attitudes in Switzerland and Great Britain, the authors show that differences between union members and non-union members are to a large extent due to the fact that unions attract workers who are more interested in politics, participate more, and have more sympathies for pro-labour parties. It is therefore not the experience of membership per se that transforms political views. The authors also show that people tend to transform their attitudes and become more similar to union members prior to joining unions. In our view, this effect does not completely rule out some causal impact of union membership – especially in the longer time frame of individuals’ life-trajectories and when it comes to radical right voting. Our cross-sectional design does not allow us to properly test for self-selection effects. We report our results, however, with and without a control for individuals' ideological self-placement (see Mosimann and Pontusson, 2017). If union members are less supportive of radical right parties even when we control for ideology, the idea that union membership is not a mere proxy for having sympathies for pro-labour parties becomes more credible. Since we mainly want to test whether or not there is still congruence between unions and their members in as far as they both reject the political ideas of the radical right, it is furthermore not that important to our argument if we observer union membership effects on radical right voting as a result of the diffusion of values, self-selection, or a combination of both. We therefore posit that: 12 ________________________ 12 Even though we restrict our hypotheses to voting for the radical right in this theoretical section, we will consistently compare this vote choice to voting for mainstream parties of the left – unions’ traditional allies in the political sphere – in our empirical analysis. Additionally, we include the option of non-voting in our empirical analysis. The fact that union members are more likely to turnout is well established in the literature and our findings should not differ from previous literature. Mosimann, Rennwald, and Zimmermann (2018) 14 H1: Union membership decreases citizens’ likelihood to vote for radical right parties. As discussed, previous findings with regard to the effects of union membership on the electoral behaviour of working-class voters are particularly ambiguous. Since the radical right’s strategy that we have described primarily – although not exclusively – targets workers with low and medium skills, they are most likely to be disoriented by the radical right’s promises to defend native workers’ interests. Moreover, differences in political sophistication across classes provide working-class citizens with fewer individual resources to decipher the discourse of the radical right when compared to middle-class citizens. Collective organisation is thus likely to matter more to working-class voters than middle-class ones by compensating the former for a possible lack of individual resources.13 On this basis, we expect that union membership renders working-class and middle-class voters less likely to vote for the radical right as discussed under H 1 but posit that union membership has a stronger negative effect on radical right voting among working-class voters: H2: The negative effect of union membership on the likelihood to vote radical right is stronger among working-class voters than middle-class voters. Data and variables Data The test of our hypotheses is based on rounds 5 to 8 of the European Social Survey (ESS) – or, more precisely, on elections after the start of the Great Recession in 2008 ________________________ 13 With regard to political participation, many authors have highlighted the importance of unions in boosting turnout among groups known for high abstention rates such as working-class citizens (e.g., Mahoney et al., 2008, Putnam, 2000; Verba et al., 1995; Warren, 2001). In a different context, Donnelly (2016) finds that being a union member has a similar effect on pro-immigration preferences as having one additional year of education. 15 Union membership effects on radical right voting covered by these ESS rounds as shown in Table 1. The countries included in our analyses are Austria, Belgium, Denmark, Finland, France, Germany, the Netherlands, Norway, Sweden, Switzerland, and the United Kingdom. In all these countries we not only find parties on the entire left-right spectrum, but also radical right parties with a vote share substantial enough to test our hypotheses. Our final sample contains 16,044 respondents above the age of 18 (16 in the case of Austria) after having excluded non-citizens and other respondents not eligible to vote (further descriptive statistics can be found in the appendix). [Table 1] Dependent variable We rely on a combination of two ESS questions on (a) respondents’ self-reported voting in the last national election in their respective country and (b) respondents’ selfreported party choice in that election to construct our dependent variable. To assess self-reported voting, respondents are presented with the following question: “Some people don’t vote nowadays for one reason or another. Did you vote in the last national election?” The variable takes on the value 1 if the respondent states having voted in her respective country’s last national election and 0 if not.14 We identify party choice with the following ESS question: “Which party did you vote for in that election?” Respondents are presented with a country-specific list of parties that we recode into 1 “social democratic parties” and 2 “radical right parties” as shown in Table 2 (see Oesch and Rennwald, 2017). Based on a recoding of the voting question, we add “abstention” ________________________ 14 Even though self-reporting tends to overstate voting, especially among well-educated respondents (e.g., Ansolabehere and Hersh, 2012), and underreport voting for the radical right (e.g., Ivaldi 2001) we have no reason to think that such misrepresentation is more common among union members than non-union members. Mosimann, Rennwald, and Zimmermann (2018) 16 as third option to these party choices to generate our dependent variable with three vote choices. [Table 2] Independent variables Among the variables included in our analyses, we are most interested in union membership and in how union membership conditions the effect of class on vote choice, especially radical right voting. A dummy variable identifies respondents currently belonging to a trade union based on the ESS question: “Are you or have you ever been a member of a trade union or similar organisation?” We code respondents currently in a trade union with 1, those previously or never belonging to a trade union with 0. As stated above, we expect trade union membership to be negatively correlated with radical right voting. We rely on a simple dichotomy between the working class and the new and old middle class to measure class. Drawing on Oesch’s classification of occupations (Oesch, 2006), we simplify his class schema by conceiving the working class as being composed of service workers, production workers, and office clerks, while the middle class encompasses the other three classes characterized by dependent employment, that is, socio-cultural (semi-)professionals, technical (semi-)professionals, and (junior) managers as well as the two classes characterized by independent employment, that is, small business owners and self-employed professionals as well as large employers. As argued, we anticipate a stronger effect of union membership on radical right voting among working-class voters than middle-class voters and also expect voting for the radical right to be more likely among the working class than the middle class. 17 Union membership effects on radical right voting We include several control variables that matter to party choice and union membership and might also affect the relationship between union membership and radical right voting, namely sex (with men coded as 0 and women coded as 1), age, education (differentiating between 1 “less than upper secondary,” 2 “upper secondary,” and 3 “tertiary”), income (measured in income deciles), and sector of employment (with private sector coded as 0 and public sector coded as 1). Finally, we include ideological selfplacement measured on a scale from zero (left) to 10 (right) in part of our models to address the issue of self-selection. Method We employ multinomial logistic regression models to simultaneously model respondents’ choice between the three unordered categories on our dependent variable namely (1) voting for the social democrats, (2) voting for the radical right, and (3) abstaining with (1) as base category.15 To account for the nested data structure of ESS, our models include country-years fixed effects in the form of country-year dummies. Since the interpretation of multinomial logistic estimates is complex and since we are interested in relationships not directly accessible in the models’ raw estimates, our full models are only shown in the appendix and we present our results as (1) average marginal effects of union membership and other variables of interest and (2) average predicted probabilities and differences in average predicted probabilities. Union membership effects on radical right voting Table 3 shows average marginal effects of union membership and our other independent variables on respondents’ electoral behavior (based on models 1 and 2 in ________________________ 15 Multinomial logistic models rely on the independence of irrelevant alternatives (Long and Freese, 2014: 207). A Hausman-McFadden test (Hausman and McFadden, 1984) of this assumption provides no evidence that the odds of choosing between any two categories on the dependent variable are not independent. Mosimann, Rennwald, and Zimmermann (2018) 18 appendix A3). We calculate contrasts for our categorical variable education, a change from 0 to 1 for our binary variables and a standard deviation change for our continous variables. Effects in the left-hand columns are based on a model not controlling for respondents’ ideological self-placement, while those in the right-hand columns are based on a model that does so. In line with H1, union membership decreases respondents’ likelihood to vote for a radical right party whether we control for respondents’ ideological self-placement or not. 16 As shown in Table 3, it also increases their likelihood to vote for a social democratic party or to abstain from voting irrespective of having a control for ideology in the model or not. Effects of union membership are biggest when it comes to left voting and smallest with regard to radical right voting while effects on non-voting fall somewhere in between. [Table 3] All effects of union membership clear the .001 threshold of statistical significance except the membership effect on radical right voting in a model controlling for ideology (p=.012). Effects are, however, smaller once we account for voters’ ideological selfplacement: The impact of union membership on the probability to vote for the social democrats decreases from 13 percentage points to about 10 percentage points and the effect of union membership on the probability to vote for the radical right shrinks from 4 percentage points to 1.5 percentage points. Its effect on abstaining is almost stable at around 9 percentage points. ________________________ 16 Note that over 12.6 percent of unionized respondents self-identify as extremely rightist in our sample (that is, as having an ideology between a score of 8 and 10 on the left-right axis), while 13.6 percent of non-unionized respondents and thus an only slightly higher share of non-union members self-identifies as extremely rightist too. Note also that unionization rates are pretty constant across the entire ideological spectrum at around 28 to 32 percent – except at the extreme right-pole where it falls to 20 percent. 19 Union membership effects on radical right voting The other variables affect respondents’ vote choice more or less as expected. In a model not controlling for ideology, the probability to vote left increases with income, age, and education and left voting is also more likely among the working class, public sector employees, and women than the middle class, private sector employees, and men. Conversely, non-voting is more likely among the working class, younger, and less educated voters as well as voters with lower income and those working in the private sector. We find, however, no differences between the sexes with regard to non-voting. Finally, the probability to vote radical right increases with income, but decreases with public sector employment and among women. While respondents with an upper secondary education are more likely to vote radical right than those least educated, we find no differences with regard to radical right voting between those with the lowest educational attainment and those with a tertiary education. We also observe that age and class do not alter respondents’ likelihood to vote radical right. Overall, these variables’ effects on vote choice are very robust to the inclusion of a control for ideology. Once we control for ideological differences across respondents, differences in radical right voting across classes become, however, slightly more pronounced and clear the .01 level of statistical significance. Unsurprisingly, self-placing more to the right on the ideological spectrum renders voters significantly less likely to vote left and significantly more likely to vote radical right. Voters with a more rightist ideology are also more likely to abstain. To better assess the effect of union membership on radical right voting, Figure 1 plots the average marginal effect of union membership on respondents’ voting behaviour separately for the eleven countries in our sample. In seven countries, the negative effect of union membership on the probability to vote radical right clears (or comes very close to clearing) the .05 threshold of statistical significance as shown in Figure 1: Aus- Mosimann, Rennwald, and Zimmermann (2018) 20 tria, Denmark, Germany, the Netherlands, Norway, Sweden, and Switzerland. Conversely, union membership has no effect on radical right voting in Belgium, Finland, France, and the United Kingdom. In all countries but France, union members are moreover significantly more likely to vote the social democrats than non-union members. And in all but three countries (Belgium, Denmark, and Switzerland), unionized voters are significantly less likely to abstain from voting than non-unionized ones.17 Figure 1 shows that the negative membership effect on radical right voting is most substantial in Austria, Denmark, and Switzerland. [Figure 1] To explain this cross-national variation, further investigation is needed, and we limit ourselves to some very provisional observations here. In all the countries, in which a negative effect of union membership on radical right voting can be observed, a labour movement of the classic social-democratic type is present. In this type of labour movement, trade unions and social democratic parties are organisationally independent from each other but consider themselves allies that fight for the emancipation of labour in a clear division of work with unions as responsible for the economic and the party as responsible for the political arena. All the countries, in which a union membership effect on radical right voting is absent, have labour movements that differ from this pattern in some respect: The British Labour party was itself set up by unions, which remain collective members of the party, and in France, unions are still heavily influenced by the syndicalist tradition that stresses the need for complete independence from any political party. The Finnish and Belgian labour movements are close to the social democratic model but display some major deviations from this model: In the case of Finland, a very strong trade union movement of the Nordic type co-exists with ________________________ 17 When we employ 90% confidence intervals, we additionally find a negative effect of union membership on radical right voting in Belgium, a positive effect of union membership on left voting in France, and a negative effect of union membership on non-voting in Denmark. 21 Union membership effects on radical right voting a rather weak social democratic party that traditionally faces competition from a quite strong communist party. Belgium is the country with the highest union density in continental Europe with strong Christian and liberal unions besides FGTB-ABVV with strong ties to the socialist parties. There does, however, not seem to be a clear connection between the union membership effect on radical right voting and union density. Such an effect is present and absent in countries with high (Belgium and the Nordic countries), intermediate (Austria and UK), and low (all the other countries in our sample) levels of union density. The particular strength of the negative effect of unionisation on radical right in Denmark, Austria, and Switzerland could be linked to some common feature of the radical right parties in these three countries. All of them have conserved a clearly “bourgeois” outlook with a neoliberal economic policy position. Moreover, all three parties have at some stage openly acted as allies of the more traditional centre-right party in their respective country. The Swiss People’s Party even used to be a traditional centre-right party before its populist turn in the 1990s. Up to a certain degree this is also the case for the Austrian FPÖ – although the fact that it was conceived as a gathering place for former Nazis from the outset complicates the picture. Union membership effects across social classes Table 4 displays average predicted probabilities for respondents’ vote choice and differences in these probabilities for four different groups of respondents: (1) Unionized and (2) non-unionized working-class citizens as well as (3) unionized and (4) non-unionized middle-class citizens. Table 4 shows that all these groups are more likely to vote for the social democrats than the radical right and that the probability to abstain falls somewhere in between. Mosimann, Rennwald, and Zimmermann (2018) 22 Surprisingly, being a union member affects voting patterns in a very similar way among the working class and the middle class: Union members are more likely to vote for the social democrats and less likely to vote for the radical right or to abstain than nonunion members. All these membership effects reach the .001 level of statistical significance.18 As in Table 3, membership has the biggest impact on the probability to vote left (about 13 percentage points) and the smallest impact on the probability to vote radical right (about 4 percentage points). Union membership effects are slightly bigger among working-class respondents than middle-class respondents. Union membership boosts the probability to vote for the social democrats by 1.6 percentage points more among the working class than the middle class and decreases the probability to vote for the radical right or to abstain by about 1 percentage point more among the working class than the middle class. Even though this finding supports H2, differences in the membership effect across classes fail to reach statistical significance.19 [Table 4] As before, we are also interested in whether these union membership effects hold in different countries. Figure 2 thus depicts average marginal effects of union membership on the probability to vote radical right among the working class and the middle class. In Figure 1, we have found a negative union membership effect on the probability to vote radical right in Austria, Denmark, Germany, the Netherlands, Norway, ________________________ All but one of these membership effects are robust to the inclusion of a control for ideology (see Table A5 in the appendix): While union memberhsip continues to decrease the probability to vote radical right among the working class (-.018, p=.033), middle-class voters no longer differ in their likelihood to vote radical right across union membership status once the model controls for ideological differences between union members and non-union members (-.013, p=.112). 19 For the pooled sample, union membership effects are pretty much identical to the effects presented in Table 4 when we code the classes differently or restrict the sample to respondents in dependent employment, that is, when we restrict the middle class to socio-cultural (semi-)professionals, technical (semi-)professionals, and (junior) managers. 18 23 Union membership effects on radical right voting Sweden, and Switzerland. Figure 2 shows that this negative effect holds among working-class voters in Austria, Norway, and Switzerland and is absent in all of the other countries in our sample.20 When it comes to middle-class voters, we find a negative union membership effect on the probability to vote radical right that clears (or comes very close to clearing) the .05 threshold of statistical significance in Austria, Belgium, Denmark, Germany, Norway, Sweden, and Switzerland. Conversely, we see no difference in radical right voting between union members and non-union members among middle-class respondents in Finland, France, the Netherlands, and the UK.21 While membership effects on radical right voting among specific occupational groups in Figure 2 thus correspond to the overall membership effects found in Figure 1, a union membership effect shows up in Belgium once we differentiate between working- and middle-class voters. To explain the specific effects of union membership on political sympathies in different occupational groups across countries, further investigation is needed, especially on the level of union density in particular segments of the salaried middle classes and country-specific traditions in the relationship between white collar unions and political parties. [Figure 2] Conclusion In this article, we have analysed the capacity of radical right parties to attract support from union members in recent elections in Western Europe and if this capacity differs across countries and social classes. We have argued that trade union members, that is, the core of the labour movement, should resist the appeals of the radical right to a ________________________ When employing 90% confidence intervals, we additionally find a negative effect of union membership on radical right voting among the working class in Denmark and the Netherlands, as well as a negative membership effect on radical right voting among the middle class in the Netherlands. 21 In France, we observe that unionized middle-class respondents are more likely to vote for the Front National than their non-unionized peers once we set confidence levels to 90%. 20 Mosimann, Rennwald, and Zimmermann (2018) 24 higher degree than non-union members, that is, the periphery of the labour movement. Such a negative effect of union membership on radical right voting can work through two related mechanisms: On the one hand, trade unions diffuse values of solidarity irrespective of nationality and origin to their members and this core principle of the labour movement is at odds with the authoritarian ultra-nationalism and welfare chauvinism of the radical right. On the other hand, people having pre-existing sympathies for the principles of the labour movement might be more likely to become members of one of their organisations, making union members less likely to support the radical right due to processes of self-selection. The article has underlined how radical right parties seek to compete with the political arm of the labour movement for the votes of workers in general and workers with low to medium skills in particular. We have also discussed how this strategy is reminiscent of the fascist movements’ strategy of co-opt and destroy between World War I and II. Instead of openly attempting to destroy the labour movement as the old extreme right in the interwar years did, the new radical right tries to undermine working-class solidarity with its anti-immigration agenda, by pretending to take care of (native) workers with its welfare chauvinism, and by attacking the legitimacy of trade unions in their discourses. Since the radical right’s strategy is mainly aimed at working-class voters who might partially lack the political sophistication that middle-class voters possess, we have expected trade union membership to have a stronger effect on voting patterns among the working class than the middle class. Our empirical analysis of elections across Western Europe since the start of the Great Recession show that union members display signs of resistance to the strategy of the radical right. Union members continue to mainly support the mainstream left and we find that unionization immunizes voters overall against the radical right – especially where the political and economic arm of the socialist labour movement are strongly 25 Union membership effects on radical right voting linked and where the radical right adheres to a neoliberal ideology and has colluded with the traditional centre-right in the past. Surprisingly, our results show that membership effects on radical right voting are about the same among the working and the middle class in the pooled sample but that effects of union membership among these groups differ across countries and are especially weak when it comes to working-class voters. This suggests that the core of the labour movement is generally less susceptible to voting radical right than its periphery whether or not it is directly targeted by the radical right’s strategy. It also suggests, however, that unionization no longer prevents (working-class) voters from choosing the radical right in certain contexts. Further research should consider how the membership composition of trade unions and the radical right’s strategies towards the labour movement differ across countries to better understand why effects of union membership vary across countries. Mosimann, Rennwald, and Zimmermann (2018) 26 References Abts K, Dalle Mulle E and Laermans R (2016) The Multilayered Appeal of the N-VA's Nationalism. The Communitarisation of Political, Socio-Economic and Cultural Conflicts. Working Paper. Abts K and Kochuyt T (2013) De vreemde bedreiging van de verzorgingstaat. Tijdschrift Voor Sociologie 34(3-4): 227-249. Afonso A (2015) Choosing whom to betray: populist right-wing parties, welfare state reforms and the trade-off between office and votes. European Political Science Review 7(2): 271 - 292. Afonso A and Papadopoulos Y (2015) How the Populist Radical Right Transformed Swiss Welfare Politics: From Compromises to Polarization. Swiss Political Science Review 21(4): 617–635. Afonso A and Rennwald L (2018) Social Class and the Changing Welfare State Agenda of Radical Right Parties in Europe. In: Manow P, Schwander H and Palier B (eds.) Welfare Democracies and Party Politics: Explaining Electoral Dynamics in Times of Changing Welfare Capitalism. Oxford: Oxford University Press. Andersen JG and Bjørklund T (1990) Structural Changes and New Cleavages: the Progress Parties in Denmark and Norway. Acta Sociologica 33(3): 195-217. Ansolabehere S and Hersh E (2012) Validation: What Big Data Reveal About Survey Misreporting and the Real Electorate. Political Analysis 20(4): 437-459. Arndt C (2013) The Electoral Consequences of Third Way Welfare State Reforms. Amsterdam: Amsterdam University Press. Arndt C and Rennwald L (2017) Workplace characteristics and working class vote for the old and new right in Western Europe. British Journal of Industrial Relations 55(1): 137–164. Arndt C and Rennwald L (2016) Union members at the polls in diverse trade union landscapes. European Journal of Political Research 55(4): 702–722. Arnold E (1999) Counter- Revolutionary Themes and the Working Class in France of the Belle Époque: The Case of the "Syndicats Jaunes", 1899-1912. French History 13(2): 99. 27 Union membership effects on radical right voting Arzheimer K (2013) Working-class parties 2.0? Competition between centre-left and extreme right parties. In: Rydgren J (ed.) Class Politics and the Radical Right. London; New York: Routledge, 75-90. Bornschier S and Kriesi H (2013) The Populist Right, the Working Class, and the Changing Face of Class Politics. In: Rydgren J (ed.) Class Politics and the Radical Right. London: Routledge, 10-29. Carter E (2005) The Extreme Right in Western Europe: Success Or Failure? Manchester: Manchester University Press. Childers T (1983) The Nazi Voter: The Social Foundations of Fascism in Germany, 19191933. Chapel Hill, NC: University of North Carolina Press. De Lange SL (2007) A new winning formula? The programmatic appeal of the radical right. Party Politics 13(4): 411-435. Donnelly MJ (2016) Competition and solidarity: union members and immigration in Europe. West European Politics 39(4): 688-709. Ebbinghaus B, Göbel C and Koos S (2011) Social capital, ‘Ghent’ and workplace contexts matter: Comparing union membership in Europe. European Journal of Industrial Relations 17(2) 17(2): 107–124. Ennser L (2016) A Welfare State for Whom? A Group-based Account of the Austrian Freedom Party’s Social Policy Profile. Swiss Political Science Review 22(3): 409–427. Falter J (1991) Hitlers Wähler. München: Beck. Geary D (2002) Nazis and Workers before 1933. Australian Journal of Politics & History 48(1): 40-51. Geiger T (1930) Panik im Mittelstand. Die Arbeit 10: 637-654. Gingrich J and Häusermann S (2015) The decline of the working-class vote, the reconfiguration of the welfare support coalition and consequences for the welfare state. Journal of European Social Policy 25(1): 50-75. Gruner E (1988) Die Arbeitgeberorganisationen - Spiegelbild oder Überbietung der Gewerkschaften? In: Balthasar A, Gruner E and Hirter H (eds.) Gewerkschaften Und Arbeitgeber Auf Dem Arbeitsmarkt: Streiks, Kampf Ums Recht Und Verhältnis Zu Anderen Interessengruppen. Zürich: Chronos, 813-836. Mosimann, Rennwald, and Zimmermann (2018) 28 Hadziabdic S and Baccaro L (2017) Selection, Anticipation, or Molding? Union Effects on Political Attitudes, Working paper. Hamilton RF (1982) Who Voted for Hitler? Princeton, N.J: Princeton University Press. Hausman JA and McFadden D (1984) Specification tests for the multinomial logit model. Econometrica 52: 1219-1240. Ivaldi G (2001) L’analyse comparée des soutiens électoraux du national-populisme en Europe occidentale: Apports et limites des grands programmes d’enquêtes transnationales (1990–1998). In: Perrineau P (ed) Les Croisés De La Société Fermée: L’Europe Des Extrêmes Droites. Paris: Edition de l’Aube, 53-71. Iversen T and Soskice D (2015) Information, Inequality, and Mass Polarization. Comparative Political Studies 48(13): 1781-1813. Kitschelt H (1995) The Radical Right in Western Europe: A Comparative Analysis. Ann Arbor: University of Michigan. Kranendonk M and de Beer P (2016) What Explains the Union Membership Gap between Migrants and Natives? British Journal of Industrial Relations 54(4): 846–869. Lefkofridi Z and Michel E (2017) The Electoral Politics of Solidarity: The Welfare Agendas of Radical Right Parties. In: Banting K and Kymlicka W (eds.) The Strains of Solidarity. Oxford: Oxford University Press, 233-267. Lipset SM (1960) Fascism of the Left, Right and Center. In: Lipset SM (ed.) Political Man. the Social Bases of Politics. Garden City (N.Y.): Doubleday, 127-179. Long SJ and Freese J (2006) Regression Models for Categorical Dependent Variables using Stata. College Station, TX: College Station, TX: Stata Press. Mahoney WA, van Deth J W and Rossteutscher S (2008) Civic Orientations: Does Associational Type Matter? Political Studies 56(2): 261-287. Maillard C (2008) Pierre Biétry (1872- 1918), un parlementaire iconoclaste. Parlement[s], Revue d'Histoire Politique (2): 126. Mantelli B (2004) Kurze Geschichte Des Italienischen Faschismus. Berlin: Wagenbach. Mazzoleni O (2003) Nationalisme Et Populisme En Suisse: La Radicalisation De La "Nouvelle" UDC. Lausanne: Presses polytechniques et universitaires romandes. 29 Union membership effects on radical right voting Mosimann N (2017) Unions as Agents of Electoral Mobilization in Times of Inequality. Trade Union Membership, Redistribution Support and Party Choice in Western Europe. Paper prepared for presentation at the Annual Conference of the European Political Science Association, Milan, 22-24 June, 2017. Mosimann N and Pontusson J (2017) Solidaristic Unionism and Support for Redistribution in Contemporary Europe. World Politics 69(3): 448-492. Mudde C (2007) Populist Radical Right Parties in Europe. Cambridge: Cambridge: Cambridge University Press. Oesch D and Rennwald L (2017) Electoral Competition in Europe’s New Tripolar Political Space. Class Voting for the Left, Centre-Right and Radical Right, EUI Working Paper MWP 2017/02 http://cadmus.eui.eu//handle/1814/45528. Oesch D (2008) Explaining Workers’ Support for Right-Wing Populist Parties in Western Europe: Evidence from Austria, Belgium, France, Norway and Switzerland. International Political Science Review 29(3): 349-373. Oesch D (2006) Redrawing the Class Map. Stratification and Institutions in Britain, Germany, Sweden and Switzerland. London: Palgrave Macmillan. Petersen J (1983) Wählerverhalten und soziale Basis des Faschismus in Italien zwischen 1919 und 1928. In: Schieder W (ed.) Faschismus Als Soziale Bewegung. Deutschland Und Italien Im Vergleich. Göttingen: Vandenhoeck & Ruprecht, 119-156. Putnam RD (2000) Bowling Alone: The Collapse and Revival of American Community. New York: Simon & Schuster. Reichardt S (2002) Faschistische Kampfbünde. Gewalt Und Gemeinschaft Im Italienischen Squadrismus Und in Der Deutschen SA. Köln: Böhlau Verlag. Rennwald L (2015) Partis Socialistes Et Classe Ouvrière. Ruptures Et Continuités Du Lien Électoral En Suisse, En Autriche, En Allemagne, En Grande-Bretagne Et En France (1970-2008). Neuchâtel: Editions Alphil-Presses universitaires suisses. Rennwald L and Pontusson J (2017) Paper Stones Revisited: Class Voting, Unionization and the Electoral Decline of the Mainstream Left. Paper presented at the Annual Conference of the European Political Science Association (Milan, 22-24 June, 2017). Röth L, Spies D and Afonso A (2017) The Impact of Radical Right Parties on Socio-Economic Policies. European Political Science Review. Mosimann, Rennwald, and Zimmermann (2018) 30 Rydgren J (2013) Class Politics and the Radical Right. London & New York: Routledge. Stegmann D (1971) Zwischen Repression und Manipulation. Konservative Machteliten und Arbeiter- und Angestelltenbewegung, 1910-1918. Afs: 351-432. Sternhell Z (2000) Ni Droite, Ni Gauche: L'Idéologie Fasciste En France. Paris: Fayard. Stöss R (2017) Trade Unions and Right-Wing Extremism in Europe. Friedrich-Ebert-Stiftung. Van Oorschot W (2006) Making the Difference in Social Europe: Deservingness Perceptions among Citizens of European Welfare States. Journal of European Social Policy 16(1): 2342. Verba S, Schlozman KL and Brady H (1995) Voice and Equality: Civic Voluntarism in American Politics. Cambridge, MA: Harvard University Press. Warren ME (2001) Democracy and Association. Princeton, N.J.: Princeton University Press. Wildt M (2014) Der Begriff der Arbeit bei Hitler. In: Buggeln M and Wildt M (eds.) Arbeit Im Nationalsozialismus. München: De Gruyter Oldenbour, 3-24. 31 Union membership effects on radical right voting Tables and figures Table 1. Elections included in the sample Country Austria Belgium Denmark Finland France Germany Netherlands Norway Sweden Switzerland United Kingdom ESS 2010 2010 2009 2010 2009 2010 2010 ESS 2012 2010 2011 2011 2012 2009 2012 2009 2010 2011 2010 ESS 2014 2013 2014 2011 2011 2012 2013 2012 2013 2014 2011 2010 ESS 2016 2013 2014 2015 2012 2013 2012 2013 2014 2015 2015 Table 2. Parties included in the sample Country Social Democratic Parties Radical Right Parties Austria Social Democratic Party of Austria, SPÖ Alliance of the Future of Austria, BZÖ Freedom Party of Austria, FPÖ Belgium Socialist Party (Flemish), SP.A Socialist Party (French), PS New-Flemish Alliance, N-VA Flemish Interest, VB National Front, FN Denmark The Social Democrats, SD Danish People’s Party, DF Finland Social Democratic Party of Finland, SDP True Finns, PS France Socialist Party, PS National Front, FN Social Democratic Party, SPD Alternative for Germany, AfD National Democratic Party, NPD Labour Party, PdVA Party for Freedom, PVV/List Wilders Norway Norwegian Labour Party, A Progress Party, FRP Sweden Social Democrats, S/SAP Sweden Democrats, SD Socialist Party, SP Swiss People’s Party, SVP Labour UK Independence Party, UKIP Democratic Unionist Party, DUP Germany Netherlands Switzerland United Kingdom Mosimann, Rennwald, and Zimmermann (2018) 32 Table 3. Average marginal effects of union membership and other variables on vote choice Variables Union membership Class (ref. middle class) Income Sector of employment (ref. private) Age Sex (ref. male) Social Democrats .129 (.000) -.029 (.001) .035 (.000) .099 (.000) .097 (.000) .049 (.000) Radical Right -.039 (.000) .011 (.112) .013 (.000) -.064 (.000) -.003 (.285) -.054 (.000) Abstain .027 (.009) .116 (.000) .024 (.004) -.013 (.143) -.050 (.000) -.103 (.000) -.090 (.000) .018 (.018) -.048 (.000) -.035 (.000) -.094 (.000) .005 (.505) Social Democrats .101 (.000) -.035 (.000) .042 (.000) .062 (.000) .090 (.000) .028 (.001) Radical Right -.015 (.012) .017 (.006) .007 (.024) -.034 (.000) .002 (.402) -.032 (.000) Abstain .028 (.003) .087 (.000) -.163 (.000) .022 (.002) .008 (.316) .152 (.000) -.050 (.000) -.095 (.000) .011 (.001) -.086 (.000) .018 (.018) -.049 (.000) -.027 (.001) -.092 (.000) .004 (.587) Education (ref. less than upper secondary) Upper secondary Tertiary Left-right self-placement P-values in brackets. For continuous variables, the average discrete changes for a standard deviation change is calculated, for dichotomous variables a change from 0 to 1 is calculated, and for categorical variables contrasts are calculated. ESS 2010 to 2016, marginal effects based on models 1 and 2, Appendix A3. Table 4. Predicted probabilities of vote choice conditional on union membership and class Social Democrats Radical Right Abstain .570 .432 .172 .215 .258 .353 .138 (.000) .043 (.000) .095 (.000) .588 .466 .166 .202 .247 .332 diff .122 (.000) .036 (.003) .085 (.000) diff in diff .016 (.287) .007 (.545) .010 (.510) Working Class Union Member Non-Union Member diff Middle Class Union Member Non-Union Member P-values in brackets; t test of equality hypothesis for differences. ESS 2010 to 2016, based on model 3 in Appendix A4. 33 Union membership effects on radical right voting Figure 1. Average marginal effects of union membership on vote choice by country with 95% confidence intervals Austria Belgium Denmark Finland France Germany Netherlands Norway Sweden Switzerland United Kingdom A: Abstention, RR: Radical Right, SD: Social Democrats. ESS 2010 to 2016, marginal effects based on models for single countries identical to model 1, Appendix A3 (available upon request) including time fixed effects instead of country-years fixed effects. Sample sizes are: Austria (817), Belgium (1,830), Denmark (539), Finland (1,250), France (1,657), Germany (2,417), Netherlands (1,573), Norway (1,872), Sweden (1,298), Switzerland (952), United Kingdom (1,839). Mosimann, Rennwald, and Zimmermann (2018) 34 Figure 2. Average marginal effects of union membership on radical right voting among the middle and working class by country with 95% confidence intervals Austria Belgium Denmark Finland France Germany Netherlands Norway Sweden Switzerland United Kingdom ESS 2010 to 2016, marginal effects based on models for single countries identical to model 3, Appendix A4 (available upon request) including time fixed effects instead of country-years fixed effects. 35 Union membership effects on radical right voting Appendix Table A1. Sample size Country Austria Belgium Denmark Finland France Germany Netherlands Norway Sweden Switzerland United Kingdom ESS 2010 604 735 522 699 394 684 ESS 2012 412 284 429 552 545 379 440 302 332 347 ESS 2014 383 421 255 432 574 623 361 338 303 282 443 ESS 2016 434 393 389 531 514 311 395 299 338 365 Total 3,638 4,022 4,415 3,969 Total 817 1,830 539 1,250 1,657 2,417 1,573 1,872 1,298 952 1,839 16,044 Table A2. Descriptive statistics, ESS 2010-16 Variable Country Year Vote choice Union membership Class Income Sector of employment Age Sex Education Left-Right self-placement N 16,044 16,044 16,044 16,044 16,044 16,044 16,044 16,044 16,044 16,044 16,044 MEAN 5 51 5 S.D. 3 17 2 MODUS Germany 2014 Social Democrats Non-union member Middle class 5 Private 55 Male Upper secondary education 5 MEDIAN 2014 5 51 Upper secondary education 5 Mosimann, Rennwald, and Zimmermann (2018) 36 Table A3. Union membership and other variables as determinants of vote choice with vote for social democratic party as base category, fixed effects multinomial logistic regression models Variables Constant Union membership Class (ref. middle class) Income Sector of employment (ref. private) Age Sex (ref. male) Radical Right .573*** (.095) .586*** (.031) 1.145* (.062) .996 (.010) .534*** (.029) .986*** (.001) .645*** (.031) Model 1 Abstention 1.132 (.150) .518*** (.026) 1.153** (.053) .900*** (.008) .695*** (.032) .963*** (.001) .904* (.037) Radical Right .570** (.105) .657*** (.039) 1.262*** (.075) .971** (.011) .640*** (.039) .985*** (.002) .723*** (.038) Model 2 Abstention 1.340* (.184) .537*** (.028) 1.190*** (.056) .889*** (.008) .753*** (.036) .963*** (.001) .927† (.039) Education (ref. less than upper secondary) Upper secondary Tertiary 1.069 (.069) .700*** (.052) .781*** (.043) .513*** (.033) Left-right self-placement Country-years fixed effects Log pseudolikelihood Wald chi2 Pseudo R2 N Yes -13,976 3,711*** .156 16,044 1.071 (.077) .814** (.065) 2.040*** (.034) .776*** (.044) .551*** (.036) 1.361*** (.017) Yes -12,490 4,968*** .246 16,044 Odds ratios based on multinomial logistic regression with robust standard errors, standard errors in brackets - *** significant at .01%, ** significant at 1%, * significant at 5%, † significant at 10% - continuous variables centred at their sample mean and cases with missing values removed from sample. Data from ESS 2010-16. 37 Union membership effects on radical right voting Table A4. Union membership and other variables as determinants of vote choice with vote for social democratic party as base category and interaction between membership and class, fixed effects multinomial logistic regression models Variables Constant Union membership Class (ref. middle class) Income Sector of employment (ref. private) Age Sex (ref. male) Radical Right .566*** (.094) .612*** (.045) 1.179** (.074) .996 (.010) .532*** (.029) .986*** (.001) .644*** (.031) Model 3 Abstention 1.122 (.149) .538*** (.038) 1.177*** (.060) .901*** (.008) .693*** (.032) .963*** (.001) .902* (.037) Radical Right .562** (.104) .690*** (.055) 1.303*** (.091) .972** (.011) .637*** (.039) .985*** (.002) .720*** (.038) Model 4 Abstention 1.324* (.182) .566*** (.040) 1.222*** (.064) .890*** (.008) .750*** (.036) .963*** (.001) .924† (.039) Education (ref. less than upper secondary) Upper secondary Tertiary 1.072 (.069) .700*** (.052) .782*** (.043) .513*** (.033) Left-right self-placement 1.074 (.077) .813** (.065) 2.040*** (.034) .778*** (.045) .550*** (.036) 1.361*** (.017) Interaction Membership * class Country-years fixed effects Log pseudolikelihood Wald chi2 Pseudo R2 N .917 (.090) .927 (.086) Yes -13,976 3,713*** .156 16,044 .907*** (.099) .902*** (.086) Yes -12,489 4,972*** .246 16,044 Odds ratios based on multinomial logistic regression with robust standard errors, standard errors in brackets - *** significant at .01%, ** significant at 1%, * significant at 5%, † significant at 10% - continuous variables centred at their sample mean and cases with missing values removed from sample. Data from ESS 2010-16. Mosimann, Rennwald, and Zimmermann (2018) 38 Table A5. Predicted probabilities of vote choice conditional on union membership and class, controlling for ideology Social Democrats Radical Right Abstain .549 .438 .191 .209 .260 .353 .111 (.000) .018 (.033) .093 (.000) .571 .479 .177 .190 .252 .331 diff .092 (.000) .013 (.112) .079 (.000) diff in diff .019 (.217) .005 (.655) .014 (.366) Working Class Union Member Non-Union Member diff Middle Class Union Member Non-Union Member P-values in brackets; t test of equality hypothesis for differences. ESS 2010 to 2016, based on model 4 in Appendix A4.