Introduction
Photographs of Pegida protests in Germany and Austria, or White supremacists marching in Charlottesville, suggest anecdotally what scholars have demonstrated empirically: supporters of contemporary far-right movements tend to be younger, hypermasculine men drawn from the ‘White working class’ and dissatisfied with ‘politics-as-usual’ (Lubbers et al., Reference Lubbers, Gijsberts and Scheepers2002; Arzheimer and Carter, Reference Arzheimer and Carter2006; Gest et al., Reference Gest, Reny and Mayer2018; Kimmel, Reference Kimmel2018). Indeed, far-right party support exhibits a persistent cross-national ‘gender gap’, whereby men support far-right parties at higher rates than women (Givens, Reference Givens2004; Harteveld and Ivarsflaten, Reference Harteveld and Ivarsflaten2018; Harteveld et al., Reference Harteveld, Dahlberg, Kokkonen and Van Der Brug2019). There is also a stereotype associated with male supporters, that of the White working class, who hold anti-immigrant views and express economic anxiety about being ‘left behind’ (Ford and Goodwin, Reference Ford and Goodwin2014; Gest, Reference Gest2016). Women are rarely featured in these depictions or stereotypes, yet we know sometimes upward of 40% of far-right party support comes from women. Given the sweeping success of far-right parties in Western Europe, this amounts to millions of women’s motivations for far-right support largely left unconsidered and demonstrates the need for further research.
In this paper, we examine women’s support for far-right parties in Western Europe. Women are variously thought to be inoculated against far-right support by different occupational profiles, sociocultural progressivism, greater religiosity, more favorable attitudes toward the welfare state, and an aversion to extremism and social stigma (Funk and Gathmann, Reference Funk and Gathmann2006; Arzheimer and Carter, Reference Arzheimer and Carter2009; Abendschön and Steinmetz, Reference Abendschön and Steinmetz2014; Harteveld and Ivarsflaten, Reference Harteveld and Ivarsflaten2018; Harteveld et al., Reference Harteveld, Dahlberg, Kokkonen and Van Der Brug2019). However, many women do support the far-right, and explanations for the so-called ‘gender gap’ largely sidestep their motivations for doing so, including how their interests and concerns may differ from their male copartisans.
Here, we take an inductive approach to identify motivations for women in supporting the far-right, focusing on attitudinal, occupational, and party-level factors. Beginning at the individual level, we consider a host of sociocultural and economic attitudes and how they relate to far-right support differently for men and women. Anti-immigrant motives for far-right support are well known, but when far-right parties castigate Islam in Europe, they often do so with a rhetorical defense of ‘liberal’ or ‘European’ values, explicitly including attitudes toward women (Zúquete, Reference Zúquete2008). Thus, we examine an expansive suite of attitudinal correlates including and beyond immigration. We also build upon an existing literature that shows the far-right has made inroads among women in nonmanual labor positions and service work (Mayer, Reference Mayer2013), in contrast to the largely blue-collar image of male far-right support (e.g. Betz, Reference Betz1994; Rydgren, Reference Rydgren2013). Then, moving from the individual to the party level, we consider characteristics of far-right parties and their leaders directly. Descriptive representation of women by women may matter for women’s radical right support, given several prominent female party heads. Moreover, party positions on moral and economic issues beyond immigration can affect support from female and male voters, reflecting and reinforcing the role of attitudes at the voter level.
Employing multilevel models using pooled data from the European Social Survey (ESS) and party manifesto data, we find far-right support among women correlates with culturally progressive positions at both the individual and party level. While both men and women hold anti-immigrant views, we can contrast this profile with the other socially conservative positions predictive of far-right support among men, consistent with seminal arguments to ‘working-class authoritarianism’ (e.g. Lipset, Reference Lipset1981). Second, building on existing insights, we find inter alia the blue-collar occupational structure of far-right support is a largely male phenomenon (e.g. Coffé, Reference Coffé and Rydgren2013; Rydgren, Reference Rydgren2013), where women’s far-right support appears to be drawn from those in routine nonmanual (i.e. clerical, service, and sales) work.
Our findings provide important evidence of variation in voter motives by gender as far-right parties continue to gain support across Europe and elsewhere. We discuss the implications of this ranging attitudinal profile in the context of far-right party development from single-issue parties to an ideologically cohesive and sociostructurally rooted party family (Ennser, Reference Ennser2012). We suggest the far-right’s widening appeal is a result of successfully framing xenophobia as a type of progressive chauvinism: rights for me, not for thee. We also point to new avenues for research that explore wider occupational characteristics of far-right supporters, individual attitudes, preferences, as well as party-level studies on issue positioning to attract liberal voters.
Why do women support the far-right?
‘New’ far-right parties emerged in the 1980s, beginning with France’s National Front (FN) European Parliament breakthrough in 1984 and Jörg Haider’s takeover of the Austrian Freedom Party (FPÖ) in 1986 (Rydgren, Reference Rydgren2005).Footnote 1 From diverse origins, the new far-right represents a durable addition to Western European politics and coherent party family (Ennser, Reference Ennser2012). Ideologically, the party family has converged on a mixture of populism, authoritarianism, and nativism (Mudde, Reference Mudde2007). These elements appear in the electorate as anti-immigrant attitudes, distrust of the political establishment, and Euroskepticism (Gomez-Reino and Llamazares, Reference Gómez-Reino and Llamazares2013). They have particular purchase among sociodemographic groups concerned with declining status in postindustrial society, or who are ill equipped to handle the pressures of globalization (Ignazi, Reference Ignazi2003; Gest et al., Reference Gest, Reny and Mayer2018). For this subset of voters, immigration and European integration – not to mention out-of-touch, sclerotic mainstream parties – are threatening. This heightens the appeal of the far-right.
Scholarship on Western Europe’s new far-right identified a gender gap early on, suggesting that ‘as if following some unwritten law, radical right-wing populist parties have consistently attracted a considerably higher number of male than female voters’ (Betz, Reference Betz1994: 142). In traditional accounts of far-right support, disparate educational and occupational profiles, which are correlated with gender, mean that men are more susceptible to both authoritarianism and the insecurities attending globalization and European integration. Men exhibit concerns regarding perceptions of declining status (Rydgren, Reference Rydgren2013; Immerzeel et al., Reference Immerzeel, Coffé and van der Lippe2015; Gest et al., Reference Gest, Reny and Mayer2018), perceived competition with immigrants (Arzheimer, Reference Arzheimer2009), and work situation, where men are more exposed to the globalizing pressures that stimulate support for right-wing parties (Kitschelt and McGann, Reference Kitschelt and McGann1995; Givens, Reference Givens2004; Mayer, Reference Mayer2013).
In this construction, women’s support for the radical right is more commonly accounted for as a ‘control’ dummy variable in regression analyses. But women are not a residual category and exhibit distinct sociodemographic characteristics and attitudes that differentially account for partisanship. To get at an account of why women support far-right parties, we build on a growing literature that examines both the aforementioned gender gap as well as explanations for women’s support directly. We consider each in turn.
First, far-right gender gap accounts locate men as a more available audience to far-right appeals than women. Where younger, educated, professional women gravitated to support parties on the left, given their emphases on gender equality and reproductive rights (Abendschön and Steinmetz, Reference Abendschön and Steinmetz2014), less educated, blue-collar men were susceptible to far-right appeals under these political opportunity structures (Kitschelt and McGann, Reference Kitschelt and McGann1995). The postmaterial ideological package offered by New Left parties, emphasizing gender and reproductive rights, similarly pitted younger, educated women against ‘identity defending’ men available to far-right parties espousing traditional, authoritarian values (Ignazi, Reference Ignazi2003). Far-right parties’ emphasis on national identity also meant that far-right parties had an interest in maintaining the structure of traditional families, including birthrates, that are an anathema to the priorities of young, educated, women (Givens, Reference Givens2004).
Beyond relevant attitudes and sociodemographic structure, women also appear more concerned with the potential social stigma associated with far-right parties (Harteveld et al., Reference Harteveld, Dahlberg, Kokkonen and Van Der Brug2019), as well as an internal motivation to control prejudice (Harteveld and Ivarsflaten, Reference Harteveld and Ivarsflaten2018), resulting in disproportionately male far-right electorates. Scholarship has also suggested that men have a greater sense of self-efficacy than do women and, thus, are more likely to support nontraditionalist parties (Mudde, Reference Mudde2007; Mayer, Reference Mayer2013) and engage in adversarial politics like those of the far-right (Gidengil et al., Reference Gidengil, Hennigar, Blais and Nevitte2005; Harteveld et al., Reference Harteveld, Brug, Dahlberg and Kokkonen2015). The implication, then, is women are more deferential to traditional patterns of authority and therefore less likely to defect from a ‘traditional’ party (Kitschelt and McGann, Reference Kitschelt and McGann1995) Women are also more religious than men on aggregate, and attendance at religious services ‘inoculates’ right-of-center voters against far-right support (Arzheimer and Carter, Reference Arzheimer and Carter2009).
Second, moving from accounts of the gender gap to explanations for women’s outright support requires theorization that considers distinct socioeconomic, demographical, and attitudinal explanations. Here, we build on an important foundation of scholarship that examines why women support right-wing conservative parties, focusing on different values to their male counterparts (de Geus and Shorrocks, Reference de Geus and Shorrocks2020). For example, Coffé (Reference Coffé2019) finds gendered personality traits explain support for the far-right among men, but not women. Women also exhibit different values from other women who might vote for parties on the left (Shames et al., Reference Shames, Och and Cooperman2020), such as resisting traditional gender roles in child rearing (Celis and Childs, Reference Celis and Childs2014) or needs for descriptive representation (on left preferences, see Kittilson, Reference Kittilson2006).Footnote 2 Finally, there is evidence that highlights occupational differences between men and women to explain far-right support (Rippeyoung, Reference Rippeyoung2007). This groundwork establishes a potentially unique set of attitudinal characteristics of far-right women.
Building on these insights, we explore sociocultural attitudes specifically to develop a richer profile of what issues matter to the female far-right voter and how parties appeal to these issues. That is, this paper offers a theory for female far-right support that considers mass opinion and context, requiring a model that takes into account both attitudinal and context or party-level theorizing.
Attitudinal
Our de facto ‘meta-hypothesis’ suggests that correlates of far-right support are statistically different and substantively distinct between women and men. Although the gender gap itself is at least in part predicated on different socialization experiences for men and for women, rather than obvious attitudinal differences as regards the far-right’s flagship immigration issue (Harteveld et al., Reference Harteveld, Brug, Dahlberg and Kokkonen2015), other attitudes may be relevant for female far-right support. Moreover, characteristics of far-right parties might make them more or less appealing to men and women at different rates. The empirical implication, then, is that the far-right gender gap is not simply the result of composition (e.g. women are less likely to be engaged in blue-collar work and therefore less likely to support the radical right) or aversion to social stigma, but distinct attitudes and sociodemographics.
The most predictive attitude for far-right support generally is on immigration. More broadly, the far-right tends to evoke nostalgia, reacting against progressive trends such as greater gender equality, multiculturalism, and postindustrialism (Ignazi, Reference Ignazi2003; Rydgren, Reference Rydgren2013). Cultural values extend beyond ethnocentrism, and the strength of anti-immigrant views may mask variation along other sociocultural attitudes that may differ across men and women. We allow the possibility here that sociocultural values are configurational.
To wit, there is newer evidence that suggests not all far-right voters are entirely conservative. Erzeel and Celis (Reference Erzeel and Celis2016) find right party positioning on postmaterial issues is a strong predictor for attention to gender issues. That is, women that care about postmaterial issues are not precluded from supporting far-right parties nor are far-right parties precluded from incorporating postmaterial values into their political agenda. As such, one of the fastest growing groups of supporters of far-right parties are what Lancaster (Reference Lancaster2020) labels ‘sexually-modern nativists’ (the other groups being conservative nativists and moderate nativists). The nativist component reflects traditional anti-immigrant and nationalism positions. And recent overtures toward women’s equality, gay rights, and freedom of speech by far-right parties against Islam – for example, framing a burqa ban not as anti-religious but as a stance in support of female emancipation (Scott, Reference Scott2009; Zúquete, Reference Zúquete2008) – confirm that women who support far-right parties might be culturally to the left of men who do the same. Indeed, Dalton (Reference Dalton2017) demonstrates a ‘cultural’ dimension related to immigration, or crime and punishment, that is separate from a ‘moral’ dimension (i.e. same-sex marriage and abortion) operating in many EU member states. This leads us to generate two hypotheses about moral values, distinct from positions on immigration (or Euroskepticism) – one on conservativism and – unique to this study – one on cultural progressivism.
Hypothesis 1A: Cultural conservatism correlates with far-right support for both men and women.
Hypothesis 1B: Cultural progressivism correlates with far-right support among women.
Second, we test a set of attitudinal preferences about the economy and redistribution preferences. Early accounts suggesting that far-right voters held right-wing economic preferences have been widely criticized as either misguided or historically contingent (De Lange, Reference de Lange2007). Moreover, in assuming that women support the far-right at lower rates because of their occupation or economic preferences, the relationship between class profiles and preferences on redistribution with female support assumes that economic drivers of far-right support operate the same for men and for women. This assumption deserves to be problematized given the ambiguous and ambivalent economic platforms of far-right parties (Ivarsflaten, Reference Ivarsflaten2008; Rovny, Reference Rovny2013; Röth et al., Reference Röth, Afonso and Spies2018). Our supposition is that far-right supporting women and men will have different economic preferences related to occupation, reliance on the welfare state, and perceived competition with immigrants for scarce resources. As such, Kitschelt and McGann’s (Reference Kitschelt and McGann1995) seminal but historically contingent right-wing authoritarian ‘winning formula’ might be reassessed through a gendered lens, with right-wing economic preferences among men predicting support at a greater rate than among women. Indeed, it is possible that the ambiguous profile of far-right supporters’ economic preferences uncovered in subsequent analyses stems from differences in economic preferences between male and female far-right voters. Hence, we hypothesize
Hypothesis 2: Right-wing economic preferences toward redistribution will correlate with far-right support among men more so than among women.
Sociodemographics
To appropriately contextualize these sociocultural and economic preferences, we also account for a variety of sociodemographic characteristics. The relationship between the working-class and far-right politics is well documented (Betz, Reference Betz1994; Rydgren, Reference Rydgren2013). Explanations for the relationship tend to evoke the perceptions of declining status (Rydgren, Reference Rydgren2013), hierarchical workplace environments (Kitschelt and McGann, Reference Kitschelt and McGann1995), and perceived competition with immigrants (Arzheimer, Reference Arzheimer2009).
Among these, declining status is of particular importance for working-class men, for whom trends toward greater gender equality as well as occupational insecurity are doubly threatening (Gest, Reference Gest2016). Beyond that, workplace environment and immigrant competition explanations for far-right support predict little difference between men and women engaged in blue-collar work in their propensity to support far-right politics (cf. Rippeyoung, Reference Rippeyoung2007; Coffé, Reference Coffé and Rydgren2013). Indeed, we might expect routine service, sales, or clerical workers of either gender to support far-right parties, if the mechanism is perceived competition with immigrants. Far-right parties have made inroads among women in routine nonmanual occupations where the gender gap has narrowed given the deteriorating situation of women in the service proletariat (Mayer, Reference Mayer2013). Absent the narrative on declining status impacting men in blue-collar and trade professions, there is no obvious reason to believe women in routine manual or nonmanual work should have different propensities to support far-right parties. Thus, with regard to occupation, we hypothesize
Hypothesis 3A: Participation in blue- collar and trade work correlate with far-right support among men more so than among women.
Hypothesis 3B: Women engaged in routine nonmanual work will support far-right parties at the same rates as women in blue-collar work.
There are additional sociodemographic factors for which we do not have ex ante predictions about gender differences in far-right support, or over which the voting literature conflicts. Education is perhaps the most important variable in predicting far-right support (Mudde, Reference Mudde2007; Allen, Reference Allen2017), but no existing literature suggests that the relationship between education and far-right support differs by gender, especially independent of occupation. Similarly, Christianity – and especially church attendance – seems to prevent right-wing voters from supporting far-right parties (Arzheimer and Carter, Reference Arzheimer and Carter2009), and accordingly, the difference between men and women on this score may simply be due to women attending church at higher rates. Age and union membership are also important predictors of political behavior, and while the ‘modern’ gender gap suggests younger women will be on average to the left of both older women and men in general (Inglehart and Norris, Reference Inglehart and Norris2000), youth is also an important predictor of male far-right support in the climate of declined partisanship. Similarly, the changed size and composition of unions paint an ambiguous picture, especially once occupation is included in regression analyses.
Party level
In addition to voter characteristics, our approach also considers contextual factors at the party level. That is, voter choice is not merely a function of individual preferences but party positioning. We consider two types of position taking. First, parties may court female supports by featuring female leadership. Despite the gender gap at the level of voters, there is no conspicuous gender gap among far-right politicians, relative to other party families (other than the Greens; Mudde, Reference Mudde2007).Footnote 3 As such, despite adherence to broadly similar policy programs cross-nationally (Ennser, Reference Ennser2012), it is possible that women are more likely to support female far-right leaders than male far-right leaders. This could be due to descriptive representation (Givens, Reference Givens2004), or if female far-right leaders are perceived as less extreme than their male counterparts, even if their platforms are similar (O’Brien, Reference O’Brien2019).
Hypothesis 4: Far-right parties with female party heads will garner more support from women than far-right parties led by men.
Second, parties may court voters programmatically. While the far-right parties under study are all ‘extreme’ in their anti-immigrant platforms, there is variation with respect to their economic and sociocultural positions. For instance, despite an electorate that looks very much like far-right parties elsewhere, Dutch far-right parties are more liberal on issues of gay equality (Rydgren and Van Holsteyn, Reference Rydgren, van Holsteyn and Rydgren2005). The Danish People’s Party is more committed to the welfare state than are many of its counterparts, so much so that a new ‘far-right party’ – Nye Borgerlige – has emerged to compete for its voters with a more economically conservative platform. Historically, these parties also have different origins as either neoliberal vs. welfare chauvinist, or ethnopluralist vs. traditionalist, that may matter for the gender gap (e.g. Art, Reference Art2011).Footnote 4 Hence, we propose two hypotheses about sociocultural and economic dimensions of party competition:
Hypothesis 5: Women will be more likely than men to support far-right parties that articulate left-wing economic positions.
Hypothesis 6: Women will be more likely than men to support far-right parties with culturally progressive positions.
Data and methods
We test these hypotheses using all available waves of the ESS (1–8), covering years 2002–2016. The ESS includes a variety of questions on attitudes and voting behavior and has been frequently used in studies on the far-right (e.g. Ivarsflaten, Reference Ivarsflaten2008; Lucassen and Lubbers, Reference Lucassen and Lubbers2012). This dataset is also highly comparable across waves, enabling us to combine multiple waves across time and space (Bohman and Hjerm, Reference Bohman and Hjerm2016). We consider only country-years that include at least one far-right party. A list of parties and their leaders, along with elections in the sample, are available in Table 1. These parties are drawn from Mudde’s (Reference Mudde2007) classification scheme, updated to reflect post-publication party formation. Female party heads are shown in boldface.
Far-right parties and leaders in Western Europe
To capture party positions, we use the Comparative Manifesto Project (CMP). The CMP has several advantages for this study over expert surveys or other measures of party position and issue salience.Footnote 5 First, manifestos are a good way to assess issue positions for smaller or newer parties like those on the far-right, as they are less susceptible to isomorphism induced by expert assessments (Bornschier, Reference Bornschier2010). Perhaps more importantly, CMP enables a more granular examination of particular issue dimensions, unlike aggregate measures of cultural, moral, or economic conservatism or liberalism present in expert surveys, which combine attitudes about immigration with other impressionistic assessments of party positions. Moreover, because far-right parties are so hierarchically organized, manifestos are a reliable articulation of party preferences, whereas there may be more variance for larger, older, and more horizontally structured parties. Manifestos also better reflect the content of a party platform during an election, as opposed to expert surveys which are collected at regular intervals, but do not always coincide with an election (or the intervals used by the ESS). As with ESS, CMP data are also used regularly on studies of the radical right (e.g. Alonso and Rovira Kaltwasser, Reference Alonso and Rovira Kaltwasser2015; Bohman and Hjerm, Reference Bohman and Hjerm2016).Footnote 6
We are interested in support for far-right parties. The ESS contains two measures of political party support: recalled vote from the previous national election and reported closeness to a particular political party. Because the ESS is not an election study – and hence recalled vote might refer to elections many months or even years in the past – we build our dependent variable from the ‘closeness’ measure, following the coding scheme of Lucassen and Lubbers (Reference Lucassen and Lubbers2012).Footnote 7 Voters that report feeling ‘close’ to a far-right party are coded as a ‘1’, and all others are coded as ‘0’. Where respondents did not report closeness to any party, recalled vote was used as a proxy (see Lucassen and Lubbers, Reference Lucassen and Lubbers2012: 556).Footnote 8 This measure correlates strongly with recalled vote in the sample (r = 0.74) but more accurately reflects the relationship between contemporaneous attitudes.Footnote 9 Moreover, the closeness measure interrupts some of the epiphenomenal mechanisms blocking far-right support among women, like social stigma described above. The resulting sample is shown in Table 2.
Far-right support by gender in sample
Data from ESS. Social desirability and nonresponse likely result in the undercounting of far-right voters (Oesch, Reference Oesch2008; Allen Reference Allen2017). Data are weighted with ESS design weights.
We measure our individual-level demographic predictors of far-right support using demographic variables in the ESS, including age and education which are continuous variables measured in years. We also include measures of union and church membership, given their historical significance in structuring mass politics. To measure church membership, a dichotomous measure for belonging to a Christian religion was created and interacted with another variable for attendance at religious services. Voters who attend a Christian church at least monthly are scored as ‘1’; all others as ‘0’.Footnote 10 We also include an indicator variable for union membership. To record respondents’ occupations, we adopt the nine major groups of the International Standard Classification of Occupations (ISCO), narrowing them to four, given our interest in a somewhat narrower range of occupations. The first group corresponds to those in professional, technical, or higher administrative work (groups 1–3), the second corresponds to routine nonmanual work (e.g. service, clerical, sales; groups four and five), and the third is reserved for blue-collar and trade occupations (groups 7 and 8). A fourth category is reserved for elementary occupations and part-time work (group 9), which captures economic precariousness but less so the perceptions of declining status (Rydgren, Reference Rydgren2013).
We use ESS attitudinal variables to test our attitudinal hypotheses. We measure economic positions with a single variable asking respondents to gauge whether the government should do more to reduce differences in income levels, where low numbers are the conventionally left-wing position (five-point scale). The question most appropriate for assessing a sociocultural dimension unrelated to immigration, available in all eight rounds, asks whether gays and lesbians should be able to live as they wish. Dalton (Reference Dalton2017 finds that this issue, along with abortion and woman in the labor force, are highly correlated and constitute a distinct issue dimension. Thus, we use attitudes on gay equality to represent cultural values. This variable is also measured on a five-point scale, where higher values are more conservative. Relatedly, far-right parties have been rhetorical defenders of gay rights qua European values in their castigation of Islam (Zúquete, Reference Zúquete2008). As such, we include a variable measuring the perceived cultural impact of immigration. While ESS asks several questions about immigration policy and perceived effects of immigration, the question of cultural impact most directly accesses the noneconomic dimension of interest here. This item is measured on an 11-point scale. We also include a four-point measure of political interest as well. All of the measures are grand mean centered in the models below, so that interpretations are comparable across indicators, and descriptive statistics split by gender are presented in Table 3.
Demographic and attitudinal data for men and women in the sample
Weighted data from ESS.
Table 3 does not reveal stark differences between men and women with respect to political attitudes, consistent with Harteveld et al (Reference Harteveld, Brug, Dahlberg and Kokkonen2015), or many ascriptive characteristics. Table 3 does provide insight into the occupational differences between men and women, where women are clearly overrepresented in clerical work, service, and sales. Men are overrepresented in blue-collar work, including craft and trade professions. It is possible on superficially that far-right parties have greater support among men is simply due to the fact that there are more men in blue-collar work. However, among the occupational categories, 49% of female far-right support comes from nonmanual routine work, despite only 38% of women in the sample being employed in those professions. The 7% of female far-right support coming from blue-collar workers is comparable to the 6% of blue-collar women overall.Footnote 11 As such, we expect occupational composition not to be the gender gap’s primary driver.
Our final three measures relate to party-level characteristics. The presence of a female far-right leader is a dichotomous measure, which takes the value of ‘1’ if a far-right party had a female leader when the ESS questionnaire was given (year and month). Female leaders are listed in boldface in Table 1. We are also interested in assessing – and controlling for – how moral and economic positions inform support. Using CMP data,Footnote 12 a ‘welfare’ position scale is compiled by subtracting leftist positions from rightist ones, whereas the moral dimension is measured by a single variable indicating support for moral progressivism (or negative mentions of traditional morality).Footnote 13 This metric includes support for divorce, abortion, secularism, and ‘general support for modern family composition’ and is the party-level counterpoint to the individual attitudes toward gay equality above. Finally, we control for far-right party vote share, to account for the possibility that women are just supporting more popular parties (Harteveld et al., Reference Harteveld, Dahlberg, Kokkonen and Van Der Brug2019), and that those parties are led by women.
Moving to our analytical strategy, we estimate a series of multilevel logistic regression models as we are interested in the association of a dichotomous outcome – support for a far-right party – with a schedule of predictor variables for both individuals and parties. The sample is divided by gender and we specify three sets of two models. This specification has the appeal of not using male supporters as the default from which female voters differ and otherwise avoids a more cumbersome model loaded with interaction terms.Footnote 14 Model 1 includes sociodemographic predictors only for men and women. Model 2 includes all voter-level sociodemographic and attitudinal variables. Model 3 adds the contextual factors of party leader, party position, and far-right vote share. The utility in ‘restarting’ with the baseline sociodemographic model is that the independent effects of these variables can be examined, as it is expected political attitudes correlate with occupation and other sociodemographic traits. Each model is presented separately for both men and women with differences in coefficient estimates tested using a cross-model Wald test (significant differences appear in boldface and italics). Because the relatively short period of time under study implies a constellation of time-invariant institutional arrangements impacting party preference at the country level,Footnote 15 we also include country-fixed effects. We include a random effect for country*year, which approximates the stochastic events around each relevant election of which there are too many to model and adjusts for disparate sample sizes and interdependence of observations within a given cluster. There are 69 clusters (country-years) in the sample. We use an unstructured covariance structure to allow for correlation between random effects, as elections in a particular country may not be independent over time, or European-wide phenomena like the ‘Great Recession’ may impact multiple elections at once. The results for the three pairs of models are summarized in Table 4 (country effects are located in Appendix A).
Results from multilevel logistic regression models for men and women
Data from ESS and CMP. *P < 0.05, **P < 0.01, ***P < 0.001. Logits presented. The bolded and italicized coefficients differ significantly between men and women from cross-model Wald test for equality of coefficients at P < 0.05. Country-fixed effects omitted and are also included in Appendix A. A full interaction model is specified in Appendix B. Robustness check using North Carolina Chapel Hill Expert Survey available in Appendix C.
Results
Model 1 comports to theoretical expectations of the sociodemographic correlates of far-right support, confirming it as a valid baseline model. We see education showing a negative relationship for both men and women, albeit with a consistently steeper negative slope for women. We also find that church membership reduces far-right support among both men and women, suggesting the ‘inoculation’ effect – whereby church membership entails loyalty to Christian democratic or conservative parties – remains part of the male narrative of far-right support (Arzheimer and Carter, Reference Arzheimer and Carter2009). By contrast, union membership is uncorrelated with far-right support in any model, likely reflecting the specific attributes of the trades currently unionized.
As expected, sociocultural and technical professionals are less likely than blue-collar and service sector workers to support the far-right irrespective of gender. However, although blue-collar men support far-right parties at higher rate than the reference category, the same is not true for blue-collar women. This is compelling evidence that the narrative surrounding the status decline is perceived by men in this sort of work and is supportive of Hypothesis 3A above. In subsequent models, findings supportive of Hypothesis 3B suggest service, clerical, and sales work predicts support for far-right parties among women over technical and professional work and over blue-collar and trade work. This suggests a clear difference in occupational profile among women who support the far-right, altogether different than the blue-collar occupations from which far-right parties derive their male support.
Model 2 introduces attitudinal characteristics in addition to the sociodemographics. Inclusion of these variables does not change our substantive conclusions about demographic factors but does produce other useful insights. We do not find support for Hypothesis 2, as the relationship between respondents’ positions on wealth redistribution and far-right support is neither significant nor statistically different between men and women. This is consistent with a range of scholarship suggesting cultural grievances motivate far-right support rather than economic preferences (e.g. Ivarsflaten, Reference Ivarsflaten2008). However, Model 2 provides unambiguous support for Hypothesis 1A. As hypothesized, conservative positions on gay equality – used as a proxy for a noneconomic, moral dimension of politics – are a significant predictor of far-right support among men. This, along with the sociodemographic profile outlined in the discussion of Model 1, suggests both a social and political situation in the party space for men who support far-right parties, somewhat akin to the ‘working-class’ authoritarianism discovered by Lipset (Reference Lipset1981) or the more recent ‘transnational’ cleavage in which far-right parties have been shown to be situated (Kriesi et al., Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2008). Conversely, tolerant, ‘liberal’ positions on gay equality predict support among women, demonstrating support for Hypothesis 1B as well. This suggests that any apparent right-wing noneconomic attitudes among female far-right supporters is wrapped up in attitudes toward immigrants and does not reflect a more general sociocultural conservatism. More importantly, however, coupled with the findings in Model 3 below, this result suggests that the far-right’s castigation of Islam as incompatible with European values toward women and gays (e.g. Zúquete, Reference Zúquete2008), as well as more generally (Mayer, Reference Mayer2013; Akkerman, Reference Akkerman2015), may find purchase among a subset of the female electorate.Footnote 16
Model 2 also controls for political interest to approximate the effect of lower subjective self-efficacy found to reduce women’s propensity to support far-right parties (e.g. Kitschelt and McGann, Reference Kitschelt and McGann1995; Mudde, Reference Mudde2007; Harteveld et al., Reference Harteveld, Dahlberg, Kokkonen and Van Der Brug2019). Interest is a necessary but not sufficient component of efficacy; as such, it is a useful lower-bound proxy.Footnote 17 Political interest is a significant predictor of female far-right support ceteris paribus, while it is not for men in the sample; however, the estimated coefficients are not statistically different (although P < 0.1). As above, if women are less politically confident than men, they may accordingly be less likely than men to buck a traditional mainstream party to support a party on the radical right, or more likely to concern themselves with social stigma. The positive relationship between political interest and far-right support among women may suggest that women require a higher level of political interest to engage with far-right politics, if their politics are less politically assertive on average (Mayer, Reference Mayer2013; see Table 3). Our interpretation of this difference remains provisional, however, and warrants further study.
Model 3 adds information on party platforms and leadership to existing voter-level variables. There are several noteworthy results. First, contrary to the prediction of Hypothesis 4, female leadership appears to negatively predict far-right support among women. There are several plausible explanations for this result. First, there are only four far-right parties led by women in the sample, one of which – the Norwegian FrP – has a notoriously pronounced gender gap (Immerzeel et al., Reference Immerzeel, Coffé and van der Lippe2015). More substantively, this suggests descriptive representation may not be especially meaningful to the type of women likely to support far-right parties.Footnote 18 Indeed, earlier research has suggested individual rather than contextual effects are more important in explaining the gender gap (Immerzeel et al., Reference Immerzeel, Coffé and van der Lippe2015). Thus, we suggest individual characteristics better explain women’s motivations for far-right support as well. However, because there are only four female far-right party heads in the study, this finding too warrants further consideration.
The next two rows added to Model 3 address Hypothesis 5 and Hypothesis 6. We find no support for Hypothesis 5, as positions advocating welfare state expansion are positively associated with far-right support for both men and women. One explanation for these results might be that far-right parties have maintained ambiguous economic platforms for much of their existence (De Lange, Reference de Lange2007; Ivarsflaten, Reference Ivarsflaten2008), and CMP may simply be capturing more mature and better organized far-right parties with their count of positive mentions of government programs.Footnote 19 More tellingly, however, we find that far-right parties that exhibit support for divorce, abortion, and secularism are more likely to earn support from women. That is an important finding in and of itself, suggesting together with the above that any narrative account of far-right support invoking ‘working-class authoritarianism’ derives chiefly from our understanding of why men support the radical right. Indeed, reflecting on the individual-level findings suggests a relationship between conservatism, workplace situation, and far-right support among men. The apparent sociodemographic situation of prospective male far-right voters also corresponds to conservative positions more generally at the individual level as moral conservatism, blue-collar work, and a relative lack of formal education are all also consistent with the extant image of the ‘left-behind’ (male) far-right voter (e.g. Rydgren, Reference Rydgren2013; Ford and Goodwin, Reference Ford and Goodwin2014). For women, Model 3 tells a different story. Rather than blue-collar work, there is a correlation between routine nonmanual (service, sales, and clerical) work and far-right support, undermining the generalizability of the blue-collar, ‘left-behind’ archetype of far-right supporters (see Roodujin, Reference Rooduijn2018), and suggesting the possibility of a female, ‘left-behind’ counterpart in the service proletariat (Mayer, Reference Mayer2013).
Although the predictive power of anti-immigrant attitudes is (unsurprisingly) strong among both men and women, other covariates similarly imply a different voter profile. As compared with support for other parties, sociocultural progressivism predicts far-right support among women at both the individual and party level. Thus, the significant covariates for women in the sample suggest women who support far-right parties differ from both men who support the far-right and women who support conservative parties. Considering individual and contextual covariates, the findings above suggest that far-right parties’ strategic use of Islam as a foil to European values might have some purchase among women in Western Europe. Other accounts of far-right Islamophobia have speculated as to this possibility (Zúquete, Reference Zúquete2008; Scott, Reference Scott2009; Campbell and Erzeel, Reference Campbell and Erzeel2018), but little empirical progress has been made on that score heretofore. Finally, these findings control for far-right party vote share in the most recent election, which is unsurprisingly significant for both men and women, simply indicating that more successful far-right parties have a larger number of supporters.
Discussion
This analysis is an early scholarly step toward elucidating the complex relationship between voting behavior, gender, and far-right populism. We have suggested that the sometimes murky picture of far-right party voters is actually due to an incomplete treatment of gender. As our analysis shows, multiple characteristics predicting far-right support differ between men and women. Where there is a consistent relationship between blue-collar work and far-right support among men, most of the women who support far-right parties are employed in routine nonmanual (service, sales, and clerical) work.
Moreover, while anti-immigrant attitudes are correlated with far-right support among both men and women, other forms of social conservatism – operationalized here as attitudes toward gay equality – only predict support among men. Strikingly, tolerance toward gays and lesbians predicts greater far-right support among women. This, coupled with the finding that negative mentions of traditional morality (i.e. support for divorce, abortion, and secularism) in far-right party platforms predict support among women but not men, suggests some far-right parties’ cultural progressivism – often but not exclusively paired with castigation of Islam as anti-modern and an anathema to European values – might attract women to the far-right (Akkerman, Reference Akkerman2015; Campbell and Erzeel, Reference Campbell and Erzeel2018). Indeed, this might suggest the strategy by which some far-right parties have rhetorically defended liberal values in the first place. Future research might clarify this interaction with analyses of campaign data, or voter studies at the national level where larger samples for particular parties and candidates are available.
These findings unsettle dominant narratives about support for far-right parties. Existing work paints a picture of culturally, morally conservative men in certain occupations expressing support for radical right parties based on perceptions of declining status – implying a fixed group of male voters (perhaps, ‘working-class authoritarians’) available to right-wing populists. But the correlates of female support are different. Blue-collar work and cultural conservatism seem to only predict far-right support among men. For women, a picture emerges of someone engaged in routine nonmanual work – service, sales, or clerical occupations – for whom cultural progressivism on issues outside of immigration might resonate.
How do we reconcile this voter profile, between nativism and individual progressivism? It suggests the prevalence of a type of progressive chauvinism: ‘equality and tolerance for me, not for thee’. Far-right parties, for example, the Danish People’s Party, have gained a lot of traction in support by advocating policies scholars have described as welfare chauvinism, wherein individuals support broad social safety nets so long as they exclude immigrants from accessing entitlements (Careja et al., Reference Careja, Elmelund-Præstekær, Baggesen Klitgaard and Larsen2016). This is a strategy adopted by the far-right that successfully diffuses to mainstream parties (Schumacher and Van Kersbergen, Reference Schumacher and Van Kersbergen2016). That far-right parties simultaneously offer what we term progressive chauvinism may broaden their base, attracting a new type of (female) supporter just as the social democratic parties of the left experienced historically unprecedented declines. Further research in this area might examine far-right policy framing, issue linkages and ownership within party systems, and voter mobilization. By examining the socio-structural roots of the party family, and taking seriously the large number of female supporters who have heretofore largely been overlooked in analyses of far-right support, we have identified several important predictors of female far-right support distinct from their male compatriots. As far-right parties gain in popularity, it is essential that comparative approaches to voting behavior push beyond simplistic narratives of far-right supporters as simply jackbooted radicals or ‘angry young men’. The results suggest that men and women have different profiles and motivations for supporting the far-right, and that the way gender has been encoded in research on the far-right may have obscured important features of the party family. A more nuanced view of far-right supporters – and party positioning to expand their base – reveals distinct, gendered dimensions.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1755773920000405.
