‘Sexually modern nativist voters’: do they exist and do they vote for the populist radical right?

ABSTRACT

Populist radical right (PRR) parties have attracted anti-migration voters by claiming to serve the interests of nationally defined in-groups. Recently, several European PRR parties have shifted focus from protecting traditional values to protecting so-called modern Western values, including lesbian, gay, bisexual, and transgender rights. Here, we study whether PRR parties managed to attract voters who hold these modern values. In contrast with previous research, we do not position these voters with ‘modern values’ as necessary opponents of nativist voters. European Social Survey data linked to 29 elections demonstrate that the likelihood to vote for PRR parties is somewhat lower among lesbian and gay rights people, but not for those who have strong anti-migration attitudes. However, this effect differs across countries and in some cases, the ‘sexually modern nativists’ have the highest PRR voting likelihoods. This is not directly related to the parties’ position on modern family values, but to the party system and alternative right-wing parties present.

KEYWORDS:

• Voting
• populism
• radical right
• LGBT
• Islam
• attitudes
• populist radical right

Introduction

The classic position of populist radical right (PRR) parties has been to defend the traditional family as the location where national identity was, and should be, reproduced. More recently, however, some of Europe’s new PRR politicians have taken a different position; Pim Fortuyn¹ (Dudink 2016) and Geert Wilders (Akkerman 2015; De Lange and Mügge 2015), for example, project an image of tolerance towards gays and lesbians, arguing that such tolerance is part and parcel of the Western democratic tradition (Andersen 2013). From this point of view, these politicians argue, that it is imperative that tolerance towards LGBT people (lesbians, gays, bisexuals and the transgendered) should be protected from the perceived Islamic threat. This tension between Islam and ‘secular modernity and tolerance’ (Andersen 2013, 109) draws attention to discourses surrounding sexual identities and the framing of the PRR parties’ anti-migration message. But the question is: does such a discourse resonate with voters? According to some unscientific Internet polls, Wilders’ Party for Freedom (PVV) was the most popular among gay voters (see Spierings and Zaslove 2015a). Yet, at the same time, parties such as the Front National have been at the forefront of protests against legalizing same-sex marriage and less strict adoption laws (Akkerman 2015).

This apparent split in the PRR party family raises a series of important questions. For example: to what extent does the electorate mirror the parties’ positions? And, do the parties’ positions play a role in a voter’s decision to vote for PRR parties? One study suggests that, in Northern European countries, the PRR parties might actually disproportionally attract ‘pro-gay, but anti-immigrant’ voters (Spierings and Zaslove 2015a). However, little to no systematic research exists on either the role of attitudes towards LGBT issues in PRR voting or whether PRR parties disproportionally manage to attract (or repel) voters supporting these attitudes, given the parties’ claims of protecting or condemning unconventional or untraditional families and relationships. Combining programmatic voter and conflict theories with frame theory would suggest that the parties’ positions affect the relation between attitudes and vote choice.

To date, most empirical studies have highlighted the strong impact of anti-migration attitudes on voting PRR (Immerzeel 2015; Ivarsflaten 2008; Lubbers, Gijsberts, and Scheepers 2002; Norris 2005), emphasizing that cultural concerns are more important than economic concerns (Goodwin 2013; Ivarsflaten 2008; Lucassen and Lubbers 2012; Norris 2005; Rydgren 2008). The argument is that in Western Europe citizens perceive migrants as a threat since they are thought to possess different cultural practices and beliefs that would, in rather general terms, undermine national identity and cohesion. Sniderman and Hagendoorn (2007) argue that in most European countries, the largest source for conflict is over sexual norms, since there are few other issues on which migrants and natives in Europe vary so strongly. But, as already noted, European countries differ on the extent to which a progressive sexual norm is tied to the construction of national identity. This is often reflected in the programme of PRR parties, that is, regarding their focus and their framing of the countries national identity. Supply-side studies have shown that it is particularly the ethnicizing of other issues that is typical to PRR politics (Faist 1994; Yılmaz 2012); however, previous voter studies have not shown the relevance of sexual issues. In short, we question the extent to which those who support LGBT freedoms but hold anti-migration attitudes – sexually modern nativist voters – vote for PRR. In the process, we take both the demand and the supply sides into consideration. We focus on how anti-migration and pro-LGBT emancipation attitudes interact and/or reinforce or counteract each other.

This study uses data from over 30,000 respondents nested in 29 elections from 10 countries: Austria, Belgium, Denmark, Finland, France, Italy, the Netherlands, Norway, Sweden, and Switzerland. The voter data come from the six European Social Survey (ESS) rounds, which are coupled with Comparative Manifesto data from the electorally most successful PRR parties in Western Europe: the Austrian Freedom Party (FPÖ), the Flemish Interest (VB), the Danish People’s Party (DF), True Finns (PS), the National Front (FN), the Northern League (LN), the List Pim Fortuyn/Party for Freedom (LPF/PVV), the Norwegian Progress Party (FrP), the Sweden Democrats (SD), and Swiss People’s Party (SVP) respectively.² First, we assess, for the different countries and elections, how large the group of ‘sexually modern nativists’ is. We then analyse whether this particular group of voters (at the intersection of anti-immigrant and pro-LGBT attitudes) has a disproportionally high likelihood of voting PRR. And lastly, we test the supply-side dependency of the relationship by modelling if and how the position a party takes on the modern family is mirrored by support for PRR parties.

Theoretical background

A demand–supply approach: party position and popular attitudes

There is a growing consensus regarding the core ideological characteristics of the PRR. Mudde (2007) argues that the PRR combines populism with nativism and authoritarianism. As Mudde (2007) notes, populism is a thin-centred ideology which juxtaposes the ‘pure people’ with the ‘corrupt elite’ (Mudde 2007, 23). However, given that populism is a thin-centred ideology, it must attach itself with another ideology. The PRR combines populism with nativism and authoritarianism. Nativism is an ideology that seeks to define membership in the national community by way of belonging to ‘the native group’; outsiders are, thus, seen as ‘fundamentally threatening to the homogenous nation-state’ (Mudde 2007, 19). Authoritarianism refers to the belief that there should be ‘a strictly ordered society’ and that transgressions of this order should be ‘punished severely’ (Mudde 2007, 23).

The extant literature on the PRR recognizes that explaining the rise, success, and also failure of PRR parties relies on a combination of both supply and demand (Mudde 2007; Rydgren 2007; Van Kessel 2015). Supply-side factors are often split into internal and external supply side: the former looking at party ideologies, organization, and framing, the latter focusing on the institutional context, such as party systems and/or electoral systems (Mudde 2007; Van Kessel 2015). Demand, on the other hand, focuses on those voter characteristics that may be structural and attitudinal, involving both individual positions and individual attitudes, as well as the larger socio-structural context that affects individual positions and attitudes (Mudde 2007). Thus, looking at both demand and supply, we can see that the success of PRRs is dependent not only on the choices made by voters driven by their personal (group) interests, attitudes, and issues they evaluate as important, but also on the way PRR parties frame political issues (Rydgren 2005). It may therefore be possible that the influence of attitudes towards LGBT issues on voting PRR actually stems from the interaction between supply and demand. To determine whether this is the case, our demand focus will be on the voters’ attitudes and our supply focus will be on the parties’ positions.

Cultural threat: anti-migration and attitudes towards homosexuality

Regarding the popular demand for PRR parties, previous research has shown again and again that anti-immigration attitudes, often labelled as ‘ethnic threat’, are the strongest predictor in PRR voting, with political cynicism as a relevant runner-up (e.g. Ivarsflaten 2008; Norris 2005; Rydgren 2008; Van der Brug, Fennema, and Tillie 2000). This finding is commonly explained through realistic conflict theory, which suggests that migrants and natives are in competition over scarce resources (Coser 1956; Rydgren 2007), whereby natives are considered to be the out-group, threatening the material resources of the in-group and a threat to dominant cultural norms (McLaren 2003; Schneider 2008), often referred to as symbolic threat (Stephan and Stephan 2000). Threat ascription as such is, however, not restricted to migrants. Also, LGBT people have been seen as contesting existing sexual moral and norms of the nation. Research on prejudice and group-focused enmity, for instance, found that prejudice towards one group is associated with prejudice towards another (Zick, Pettigrew, and Wagner 2008) and more concretely also unfavourable attitudes towards LGBT people and immigrants have often been found to be (strongly) correlated (Inglehart 1997; Persell, Green, and Gurevich 2001; Whitley and Aegisdottir 2000). The overlap between attitudes towards migrants and LGBT people is even considered to be the central component in the non-economic dimension of the political landscape across politics in the West (Van der Brug and Van Spanje 2009). Green–Alternative–Libertarian (GAL) or postmaterialist values (including pro-migration and pro-LBGT attitudes) would be negatively associated with, and are often opposed to, Traditional–Authoritarian–Nationalist (TAN) values (Bakker et al. 2015; Inglehart 1997). From this perspective, one would expect that the more pro-LGBT emancipation, the less likely a person is to vote for PRR, and the more anti-migration, the more likely a person is to vote for PRR.

At the same time, focusing on competition over prevailing culture norms of a society and the rights that have been granted based on certain norms (Ivarsflaten 2005; Lucassen and Lubbers 2012), questions this juxtaposition. Regarding sexual norms, Sniderman and Hagendoorn (2007) found that, in the Netherlands, migrants are far more sexually conservative than natives. Perceiving migrants as possessing conflicting sexual norms could then actually be at the heart of anti-immigrant attitudes. Empirically, this is, for instance, reflected in the reinforcing impact of liberal sexual norms and anti-migration attitudes on voting for populist radical right in Norway and Sweden (Spierings and Zaslove 2015a).

This apparent contradiction in the literature might be resolved by realizing that while anti-migrant attitudes and anti-LGBT attitudes are associated, this association does not exclude possible pockets of voters (including LGBT people themselves) that combine pro-LGBT with anti-migration positions. Moreover, when the perceived cultural fault line between the migrant and the heartland cultures is perceived to revolve around the acceptance of homosexuality, then pro-LGBT citizens (and therefore also LGBT natives) can be expected to be led more by their anti-migration stance, because they weigh this more strongly against other issues. That would imply that this ‘sexually modern nativist’ group is the most likely to vote for PRR. A less strong interpretation would predict that pro-LGBT citizens might be less inclined to vote for PRR, but that this negative inclination is out weighted by perceiving a strong cultural threat among the group of voters who is in favour of LGBT emancipation but strongly anti-migration. In this case, the sexually modern nativists would be on par with other nativists, whereas among less anti-migration voters, pro-LGBT attitudes make more of a difference.

Shifting party positions on sexuality

So far, we have assumed a rather homogenous understanding of the ethnicization of sexual politics on the political right. However, the perceived cultural conflict over sexual rights, between migrant groups and the so-called native population that Sniderman and Hagendoorn (2007) refer to, might be context dependent. For instance, the aforementioned authors positioned their work in the Netherlands, a country known to be one of the more sexually tolerant countries, but Western countries actually display quite a strong variation in the granting of rights to and political discourses around sexual minorities (Van den Akker, Van der Ploeg, and Scheepers 2013). In their study on voting PRR, Spierings and Zaslove (2015a) only found hints of this mechanism for some of the countries included.

Considering this diversity across Western Europe and as suggested by the demand–supply approach, the pattern described above is expected to be dependent on the political context and parties’ positions. A disproportionally strong attraction of sexually modern nativist voters to PRR is most (or only) likely if the PRR party not only aims to diminish or halt migration, but if it has also incorporated pro-LGBT stances, at least in an ethnicized framework, claiming to provide particular protection against the cultural threat to LGBT freedoms.

PRR parties tend to defend the homogenous nation state, including the traditional family, and traditional moral values. However, not all PRR parties have always been explicit regarding their positions on women, as they tend to take the position that it is up to women and families to make their own decisions, implying that the parties support the dominant social order. At the same time, PRR parties clearly oppose affirmative action as put forward by the ‘left-wing, feminist political elite’ (Akkerman 2015; De Lange and Mügge 2015; Spierings and Zaslove 2015b). Also, when it comes to sexuality and the traditional family, PRR parties have supported traditional positions regarding, for example, same-sex marriage (Akkerman 2015; Spierings and Zaslove 2015b). In other words, support for homosexuality did not fit into the traditional picture of PRRs, but with the growing emphasis on Islam, an important turning point appears to have been reached.

In most cases, PRRs juxtaposed ‘natives’ with immigrants in rather general terms. But which groups the PRRs have specifically emphasized to be the main opponents of nationals’ interests has changed over time (De Lange and Mügge 2015; Zúquete 2008). Whereas in the 1990s most of the PRRs focused on asylum seekers, people from former colonies, or non-Western immigrants in general, since 9/11, the parties have most explicitly focused on Muslims. Researchers have stressed that the cultural dissimilarities between European majorities and Muslims are emphasized more strongly, with Muslim asylum seekers particularly being singled out as outsiders (Lucassen and Lubbers 2012; Sniderman and Hagendoorn 2007). Following this change, more and more PRR parties have become defenders of liberal democracy, framing it as a European tradition of secularism and tolerance vis-à-vis what they view as a backward and perhaps even a medieval Islam (Andersen 2013; Betz and Meret 2009; Zúquete 2008).

While PRR have not often explicitly sought to protect gay and lesbian citizens, the growing emphasis on Islam has created a space for some populist radical right parties to implicitly, if not explicitly, defend the rights of (native) gay and lesbians. This has especially been the case in the Netherlands (Dudink 2016). Other PRR parties have also shifted position and became more ambiguous on LGBT issues. The Swedish Democrats, for example, were never very explicitly in favour of same-sex marriage, but stated that, as Islam forms a threat to sexual minorities, LGBT people should vote for them;³ and even though the Danish People’s Party voted against same-sex marriage, they did defend the rights of gays when a Gay Pride Parade was attacked by Muslims (Andersen 2013, 109).

The question is: considering this diversity in PRR parties’ condition and national contexts, can attitudes towards LGBT people be a relevant predictor in voting for PRR parties across Western Europe? Bringing the demand and supply side together suggests that the effect for sexually modern nativists we described earlier should only be expected in countries where these parties have come to emphasize the contrast between ‘us, the sexually-modern natives’ and ‘them, the traditional migrants’. Conversely, as long as these parties promote traditional family and sexual norms, it will be unlikely for voters with anti-migration but pro-LBGT attitudes to vote for the PRR.

Hypotheses

Based on the existing literature, we can formulate the following two hypotheses.

H1: The more a voter holds anti-migrant attitudes, the more likely it is they will vote for a PRR party.

H2: The more a voter holds pro-LGBT attitudes, the less likely it is they will vote for a PRR party.

Considering the focus on Islam as a threat to Western liberal-democratic culture, including tolerance towards sexual minorities, we formulate the following new micro-level expectation.

H3: The positive impact of holding strong anti-migrant attitudes on the likelihood of voting PRR is stronger among voters who hold pro-LGBT attitudes than among voters who hold traditional attitudes towards LGBT issues.

Finally, in order to include the different positions on the modern family and LGBT emancipation among PRR parties, we add a demand–supply hypothesis.

H4: The disproportionately positive likelihood of anti-migrant, pro-LGBT attitudes voters to vote for the PRR is stronger if the PRR party takes a less traditional position regarding family formation and LGBT rights.

Data and methods

Survey data and dependent variable

We use all six rounds of the ESS data collection (ESS 2002–12b) and have selected the Western European countries and surveys in which the respondents could choose to vote for a PRR party. This refers to the respondents’ self-reported choice of vote in the then most recent national election. We organized the data so that each respondent is nested in the most recent election in their country.⁴ This allows us to combine voter- and party-level data, as the Manifesto Project includes party position per election (see below). Three elections were excluded because the number of PRR voters in those elections where too low to make reliable inferences (<25), resulting in the inclusion of 29 elections from 10 countries. For all countries, only the largest PRR party was included, either because there were no data on the second and smaller party in the Manifesto database or because the number of voters was too small. Party choice was recorded through two values: a vote for the PRR (1) or a vote for another party (0). Citizens who did not vote were not included.

Table 1 provides an overview of all elections included by year and by country, also indicating the PRR party in that election. Almost all parties included are unequivocally part of the PRR party family (see Immerzeel 2015; Mudde 2007; Spierings et al. 2015; Van Kessel 2015). We also included the LPF for the 2002 and 2003 Dutch elections. As noted, there is debate over the status of the LPF; but we treat the LPF as functional equivalent to a PRR in the Dutch case (Rydgren and van Holsteyn 2005). Moreover, the parties’ strong pro-LGBT rhetoric (De Lange and Mügge 2015; Dudink 2016) makes it a crucial case for our fourth hypothesis in the disaggregated analyses.

Table 1. The 29 elections and PRR parties included in the analyses.

	Austria	Belgium (Flanders)	Denmark	Finland	France	Italy	Netherlands	Norway	Sweden	Switzerland
1999		VB								SVP
2000
2001			DF			LN		FrP
2002	FPÖ				FN		LPF
2003		VB					LPF			SVP
2004
2005			DF					FrP
2006	FPÖ						PVV
2007		VB	DF	PS	FN					SVP
2008
2009								FrP
2010		VB					PVV		SD
2011			DF	PS						SVP
2012					FN		PVV

Note: DF: Danish People’s Party; FN: National Front; FPÖ: Austrian Freedom Party; FrP: Progress Party; LN: Northern League; LPF: List Pim Fortuyn; PS: True Finns; PVV: Party for Freedom; SD: Sweden Democrats; SVP: Swiss People’s Party; VB: Flemish Bloc/Flemish Interest.

Models

As the dependent variable is dichotomous and respondents are nested in different elections,⁵ we have estimated multilevel logistic regression models, using the generalized linear mixed model function in SPSS. These multilevel models include a random intercept that captures the difference between voters in different elections and higher level explanatory variables can be included without deflating their p-values, which is crucial for election-level variables and explaining cross-election heterogeneity. As described elsewhere (Mood 2010), coefficients of logistic regression models cannot be compared across models and interaction terms should also be interpreted with care. For that reason, this study takes multiple approaches. Besides a pooled multilevel model, we also present disaggregate models per election, which provides insight into how effects differ by country by election and which contextual factors might be crucial, as well as help detect outliers (Spierings 2016). In addition, we will present the effects in terms of probabilities based on sample averages.

Independent voter-level variables

The core micro-level explanatory variables in our hypotheses are anti-migration attitudes and attitudes towards LGBT rights. We measure anti-migration as a scale based on six items that ask people’s views on allowing different immigrant groups into a country and the (negative) impact of the presence of immigrants,⁶ which all tap into one underlying dimension.⁷ These items are recoded to create six scales running from 0 to 1 and averaged. Second, attitudes towards LGBT rights, or more precisely LG rights, are measured by one item included in all ESS rounds: Gay men and lesbians should be free to live life as they wish. This variable is measured on a scale of five from completely disagree to completely agree.

We expect that the influence of strong anti-migration attitudes differs for the most pro-LGBT right people (or vice versa). This can be incorporated in two ways in the models. The first and seemingly most straightforward way is to include the two variables discussed above as well as their interaction term. This, however, cannot account for possible non-linear direct or interaction effects, whereas particularly the latter is the core of our theoretical expectation, as we focus on strong pro-LGBT and anti-migration citizens compared to all the rest. The second option is to create sets of dummies for both variables and add an additional dummy for our group interest similar to an interaction term (Brambor, Clark, and Golder 2006). This solves the issue above, but depends on defining certain cut-off points. For modelling the context-dependency of attitudes’ effects on party positions, the second approach might also be slightly easier, as it avoids needing to include and interpret three-way interactions effects. In order to draw reliable conclusions, we have applied both methods and in the table with the pooled models we present both the dummy approach and the linear interaction models. For the other tables, the appendici provide the additional models. They show very similar results.

For attitudes towards lesbian and gay people, a dummy was created for each of the five scores. On anti-migration, five groups were made ranging from 0 to 0.2, 0.2 to 0.4, and so on. The last two groups are considered, respectively, to be respondents holding moderately strong and extremely strong anti-immigrant attitudes. As the descriptive analyses will show below, it is exactly these groups on which PRR electorates voters differ. The additional dummy for sexually modern nativists is therefore coded ‘1’ for everyone who is strongly anti-migration, and most lesbian and gay rights. As will be shown below, roughly 6.5% of all respondents score 1 on this dummy.⁸ If our contentions hold, we should find a positive coefficient for the sexually modern nativist group.

In addition to the core variables, a set of common control variables is included (see Immerzeel 2015; Ivarsflaten 2008; Lubbers, Gijsberts, and Scheepers 2002; Rydgren 2008; Spierings and Zaslove 2015a). We included a set of dummies as a detailed classification of people’s occupational position, distinguishing between the self-employed, professionally employed, clerically employed, service-sector employed, blue-collar employed, agricultural occupations, military, unemployed, other non-employed, and employed but unknown. ‘Authority or law & order’ attitudes were measured by a scale (from 0 to 1) based on three items. Populist or anti-politics attitudes could be included as an index on political trust (from 0 to 1) based on the two items measuring the trust people have in parties and politicians (cf. Akkerman, Mudde, and Zaslove 2014; Schumacher and Rooduijn 2013; Spierings and Zaslove forthcoming). In addition to these core variables in the literature on voting PRR, we added an index that measured political engagement (from 0 to 1) and several demographic characteristics: education (years of schooling), age (years), sex (1 = female), religious attendance (7-point scale from ‘never’ to ‘every day’), and religious commitment (11-point scale from not at all to very religious) (cf. Immerzeel, Jaspers, and Lubbers 2013).

Election-level variables

To measure the positions of the PRR parties, we use the Manifesto data (Volkens et al. 2014). This is the only database covering all parties and countries.⁹ The Manifesto data include items on family morality issues, which echoes issues regarding gay and lesbian rights, family issues, and the idea that individuals should be able to have the freedom to live their lives as they choose. Evidently, it would be better to have concrete data on all parties’ LGBT emancipation positions, but these are not available in any comparative project. The traditional morality issue seems a reasonable proxy, including statements about the prohibition and suppression of immorality, divorce, and abortion, and the maintenance of the traditional family versus support for modern family composition.¹⁰

Based on the percentages of positive and negative sentences regarding morality issues, we have categorized the 29 PRR party platforms in 5 groups, the lowest scores indicating more traditional views on family life. The parties with only negative sentences form the lowest two groups: the 5 party platforms which devote most relative attention to the issue were coded ‘0’; the 12 others ‘1’. The platforms that included both positive and negative positions but had an overall negative score were given a ‘2’. The five platforms which did not pay explicit attention to the issue were coded ‘3’, as they have not made the traditional family part of their manifesto, indicating that it is not an important issue for the party. The last and only party with more positive than negative sentences was given the highest score of ‘4’. Not surprisingly, this was the Dutch LPF in 2003; the party was named after its openly gay leader: Pim Fortuyn (see Dudink 2016).

Analyses

The prevalence of ‘sexually modern nativism’

Before turning to the explanatory analyses, let us first establish the extent to which strong support for lesbians and gays living their lives freely is combined with strong anti-migration attitudes, ‘sexually modern nativism’. Figures 1 and 2 illustrate that among PRR voters, people with (moderately and extremely) strong anti-migration attitudes are far more prevalent (Figure 1). Not surprisingly, those who possess particularly strong positive attitudes towards LGBTs are less common (Figure 2). These differences are in line with the generally accepted dimensionality of politics in which these two points of view are considered part of opposite poles, GAL (Green–Alternative–Libertarian) versus TAN (Traditional–Authoritarian–Nationalist). Voters close to the latter are expected to be PRR voters (see also Table 2 below). It is worth mentioning though that people with explicit anti-immigration attitudes are also in the minority among PRR voters, and so are the PRR voters expressing anti-LGBT attitudes.

Figure 1. Voting for PRR parties and anti-migration attitudes.

Figure 2. Voting for PRR parties and pro-LGBT attitudes.

Table 2. Percentage of voters with ‘sexually modern nativist’ attitudes per election.

	Austria	Belgium (Flanders)	Denmark	Finland	France	Italy	Netherlands	Norway	Sweden	Switzerland
1999		6%								2%
2001			8%			4%		3%
2002	7%				11%		8%
2003		9%					10%			2%
2005			6%					3%
2006	7%						9%
2007		6%	6%	3%	13%					3%
2009								3%
2010		10%					10%		2%
2011			8%	3%						2%
2012					11%		8%

Note: Weighted using the by the ESS provided design and stratification weight.

Shifting our attention to the voters who deviate from the overall pattern and focusing more on the diversity among the population instead of the overall averages, we find that sexually modern nativists, who combine positions on both ends of the GAL–TAN scale, are no extreme rarity: in total, 6.4% of our sample takes a strong anti-migration position while also very strongly supporting freedom for lesbians and gays. Table 2 shows that the overall number of sexually modern nativists varies from 2% in Sweden and Switzerland to well over 10% in France, with Flanders and the Netherlands also touching the 10% mark. While the percentages of sexually modern nativists vary somewhat by year, none of the countries shows a clear increasing or decreasing trend. The differences mainly stress differences between countries. More generally, it is up for debate whether one considers the average of 6.4% a substantially significant group. In terms of number, we are talking about thousands of voters. For instance, even in Sweden, the 2.1% represents about one hundred thirty thousand voters; and in the Netherlands, the 8.4% in 2002 translates to about eight hundred thousand voters. In other words, the percentages are relatively small, but in terms of potential votes, this group is sizable in multiparty systems. Based on our models, we will provide some more calculation of what these figures mean in terms of votes and seats.

A comparison of the prevalence of sexually modern nativist voters and the party manifesto classifications does not show a substantially or statistically significant relationship. For instance, among the voters in France, we find the largest share of sexually modern nativists. This is the case even though the French National Front on average scores a 1 on modern morality, putting it in the bottom three, together with the Northern League (IT) and Flemish Interest (BE). In general, it should, however, be remembered that the Manifesto Project’s indicators for family morality issues measure a broader spectrum of issues than the parties’ position on LGBT people.

The impact of sexually modern nativism on voting PRR

Table 3 presents the pooled models in which voting for PRR is explained. Model 1 includes anti-migrant attitudes and support for lesbians and gays living their lives freely, as well as the control variables, which behave as expected, validating the overall models. Our core variables show that the first and strongest indicators for voting PRR are the anti-migration attitudes. The influence of anti-migration attitudes is ordinal as the likelihood to vote for the PRR rises with the strength of the anti-migration attitudes, but the effect is not linear and, in terms of logged odds, with smaller differences between the strongly anti-migration voters. For the attitudes towards gay and lesbian people, a negative ordinal (but not linear) effect is found as well. The coefficient for the most anti-LGBT voters does not completely fit this, but this might indicate that the most anti-LGBT voters are a small group of highly traditional voters consistently voting (orthodox) Christian. Overall, the models in Table 3 as well as the additional analyses (see Appendix 1) support the first two hypotheses.

Table 3. Multilevel models explaining voting for a populist radical right party in 29 elections across 10 European countries.

	Model A1 Interval variables base model		Model A2 Interval variables interaction model		Model B1 Dummy variables base model		Model B2 Dummy variables interaction model
	Logged odds	SE	Logged odds	SE	Logged odds	SE	Logged odds	SE
Core micro-level variables
Anti-migration attitudes	4.53***	0.12	3.06***	0.31
1: not against migration (ref)					ref		ref
2					1.37***	0.15	1.35***	0.15
3					2.39***	0.15	2.36***	0.15
4					3.09***	0.15	3.00***	0.16
5: extremely anti-migration					3.55***	0.16	3.45***	0.17
Pro-LGBT rights	−0.08***	0.02	−0.37***	0.06
1: strongly against					−0.01	0.12	0.01	0.95
2					0.22*	0.09	0.22**	0.09
3 (ref)					ref		ref
4					−0.08	0.06	−0.09	0.06
5: strongly in favour					−0.19**	0.07	−0.28**	0.07
Interaction: anti-migration × pro-LGBT att.			0.48***	0.09
Dummy: Sexually modern nativist; ref = no							0.21**	0.08
Micro-level controls
Occupation; ref = employed: blue-collar
Self-employed	−0.32***	0.08	−0.32***	0.08	−0.31***	0.08	−0.31***	0.08
Employed: professional	−0.75***	0.07	−0.74***	0.07	−0.75***	0.07	−0.75***	0.07
Employed: clerical	−0.39***	0.10	−0.39***	0.10	−0.41***	0.10	−0.41***	0.10
Employed: service-sector	−0.19*	0.09	−0.18*	0.09	−0.20*	0.09	−0.19*	0.09
Agricultural occupation	−0.23	0.30	−0.23	0.30	−0.17	0.30	−0.17	0.30
Military occupation	0.05	0.40	0.39	0.41	0.06	0.40	0.05	0.40
Unemployed	−0.05	0.11	−0.05	0.11	−0.03	0.11	−0.03	0.11
Non-employed	−0.36***	0.06	−0.36***	0.06	−0.36***	0.06	−0.36***	0.06
Unknown	−0.30	0.17	−0.29	0.17	−0.25	0.17	−0.25	0.17
Law & Order scale (0–1)	0.51***	0.11	0.50***	0.11	0.55***	0.11	0.55***	0.11
Political trust scale (0–1)	−0.13***	0.01	−0.13***	0.01	−0.14***	0.01	−0.14***	0.01
Political engagement scale (0–1)	−0.04	0.02	−0.04	0.02	−0.03	0.03	−0.03	0.03
Education: years of schooling	−0.05***	0.01	−0.05***	0.01	−0.05***	0.01	−0.05***	0.01
Age in years	−0.01***	0.00	−0.01***	0.00	−0.01***	0.00	−0.01***	0.00
Sex: female (ref = male)	−0.32***	0.04	−0.32***	0.04	−0.30***	0.04	−0.30***	0.04
Religious attendance (0–6)	−0.15***	0.02	−0.16***	0.02	−0.16***	0.02	−0.16***	0.02
Religious commitment (0–10)	−0.01	0.01	−0.01	0.01	−0.01	0.01	−0.01	0.01
Intercept	−1.92***	0.27	−1.00***	0.32	−1.95***	0.29	−1.89***	0.31
Model statistics
BIC	232,608.528		233,931.478		235,011.510		235,252.115
N	Nelection=29; Nindividual=39,271		Nelection=29; Nindividual=39,271		Nelection=29; Nindividual=39,271		Nelection=29; Nindividual=39,271

Note: All models are two-level models with random intercepts at the election level.

Source: ESS round 1 to 6.

*p < .05.

**p < .01.

***p < .001.

Part of the explanation of this effect being rather weak might also be that some voters with strong pro-gay rights attitudes strongly oppose the PRR, whereas others support PRR parties disproportionally as we hypothesized. Model 3 sheds more light on this. The dummy indicating those who combine strong anti-migration attitudes with pro-LGBT attitudes shows a positive coefficient and is statistically significant. Moreover, after including this dummy, the impact of pro-LGBT attitudes becomes somewhat weaker. These results are in line with Hypothesis 3, but the size of the effect is more difficult to assess given that these are logged odds coefficients. In Figure 3, the predicted probabilities of voting PRR are given, based on the coefficients in Model 2 and sample averages, for the most and least pro-LGBT voters per position on migration. This figure helps us to further interpret the results as it shows that the likelihood of voting PRR is indeed lower among those who are pro-LGBT emancipation. Actually, in terms of probabilities, the gap between those who are pro- and anti-LGBT emancipation widens when the anti-migration attitudes become stronger. However, and that is crucial for this study, this gap decreases again among the strong anti-migration voters (scores 4 and 5). For instance, among people without a strong or a neutral opinion on migration, the gap in likelihood to vote for PRR between pro- and anti-LGBT emancipation voters is 2.6 percentage points, while among the moderately strong anti-migration voters, it is only 1.0 percentage point. Or phrased differently, being rather neutral versus being a moderately strong anti-migration increases one’s likelihood of voting PRR by 8.3 percentage points among the most anti-LGBT voters, but 9.8 percentages points among the most pro-LGBT voters. At the same time, the figure shows that we cannot conclude that sexually liberal nativist are the most prone to vote for PRR. However, among the strong anti-migration voters, attitudes towards LGBT people are a weaker predictor of voting PRR. The additional analyses (see Appendix 1) confirm our third hypothesis, but it should be noted that the effect is small in terms of voting probabilities. Part of the story might be that there is a strong effect in some elections or countries, but no effect in others, which is the next step in our analysis.

Figure 3. Predicted probabilities of voting for PRR parties. Note: estimations based on Model A2 in Table 3 and sample averages.

The context dependency of sexually modern nativism’s impact

The analysis presented in Model 2 of Table 3 has been rerun per election (country-year) to see to what extent the effect varies. Since some of the more extreme subgroups with anti-migration and anti-LGBT attitudes become rather small after splitting the sample in 29 election-specific samples, we combined scores 1 to 3 for each variable into one group and took that as a reference in the dummy-based models. The ‘sexually modern nativist’ variables can be interpreted in the same way as above: a positive and statistically significant relationship indicates a disproportional tendency to vote for the PRR among sexually modern nativists compared to non-nativist progressive voters or nativist traditional voters. Table 4 shows the direction and significance of the interaction coefficients for each and every one of these 29 elections. Evidently, for precise interpretation, the main coefficients are crucial and we will include those below when focusing on the contexts in which an impact of sexually liberal nativism was found. For now, it suffices to say that in each election, there was a general positive main effect of anti-migration attitudes, whereas the effect of attitudes towards lesbian and gay people was less robust across elections.

Table 4. The additional logged odds on voting PRR for sexually modern nativists in the 29 elections.

	Austria	Belgium (Flanders)	Denmark	Finland	France	Italy	Netherlands	Norway	Sweden	Switzerland
1999		++								±±
2000
2001			−			+		±±
2002	±±				−		−
2003		+					−			±
2004
2005			+					−
2006	++						−
2007		+	−	+	+					±±
2008
2009								±
2010		–					+		±±
2011			±	−						–
2012					+		–

Note: Only coefficient for interaction dummy; models are similar to pooled ones; not multilevel though. Underlined coefficients are statistically significant (p < .10) in the interval-based models; coefficients in bold are statistically significant in the dummy-based models;± indicates the direction of the interaction, whereby a + is the expected direction; ++ and – indicate the coefficient being larger than (−) 1.

Above all, Table 4 shows that the effects found above are not universal, but that the voting behaviour of sexually modern nativists is context dependent. In 9 of 29 elections, we find a positive statistically significant or marginally significant effect and all are positive and more substantive than the average effect found in Table 3.¹¹ No statistically significant relationships pointing in the other direction were found. Moreover, in four countries, for all or multiple elections, an effect was found: Austria, Norway, Sweden, and Switzerland.

To assess these effects further, we have plotted the predicted probabilities for the four countries in which a clear effect of sexually modern nativist was found (see Appendix 2 for the underlying models). For each of the countries, it holds that under the condition that voters are in favour of migration, the probability to vote for the PRR is larger among voters with strong anti-LGBT attitude than among voters with strong pro-LGBT attitudes. However, the figures show that this alters when voters have stronger anti-migration attitudes. Under the condition that voters have strong anti-migration attitudes, in Norway, Sweden, and Switzerland, people with pro-LGBT attitudes have a higher probability of voting PRR than people with anti-LGBT attitudes, whereas in Austria, the probabilities have then become the same. To give an indication of the effect sizes, we calculated the impact in terms of votes for the Swedish 2010 Riksdag elections. If we take out ‘bonus’ of the sexually modern nativist group, this translates roughly to the loss of a full seat in parliament for SD (Figure 4).

Figure 4. Predicted probabilities of voting for the PRR parties for four countries.

These results raise the question why this effect is found in some countries and not in others. This study is based on the assumption that both the demand and supply sides of PRR politics should be brought together in order to systematically assess and explain the dynamics at play. Based on both the voter (demand) and party (supply) literatures, we formulated the expectation that sexually modern nativist voters are likely to vote disproportionally PRR if the – by definition anti-migration – PRR party took a relatively more modern position on LGBT rights and freedoms – which is likely to be framed as part of the parties’ anti-Islam agenda. To test this claim, we added to the multilevel models (Table 3; Models 2) the election-level variable that measures the family morality position of the parties and the interaction between this variable and the dummies for sexually modern nativist attitudes. This did not provide a substantially or a statistically significant interaction. Moreover, following Mood’s (2010) discussion in interpreting interactions in logistic regression models, we also calculated whether there is a difference between the Manifesto platform scores for the cases for which we did not find a relationship in comparison with the ones for which we did (i.e. a statistically significant coefficient as reported in Table 4). There is no statistically significant difference between these groups either.¹² Our basic expectation formulated in Hypothesis 4 is refuted.¹³ In the conclusion, we will reflect on this in greater detail.

Conclusion

In a number of Western and Northern European countries, PRR parties have begun to emphasize the protection of LGBT people, rights, and freedoms against traditional Muslims, allegedly aiming to Islamize the West. In their view, Europe’s progressive, modern culture would be endangered by such traditional views regarding the emancipation of women and LGBT people. This raises the question as to what extent voters with anti-immigrant but pro-LGBT attitudes (what we called ‘sexually modern nativists’) vote for PRR parties. To account for both demand- and supply-side explanations of voting behaviour, we alluded to both the voters’ attitudes and the parties’ position on LGBT issues. We found that sexually modern nativists are more likely to vote for PRRs than sexually traditional nativists, supporting our key hypothesis. The group of sexually modern nativists turned out to be a very clear minority in all countries under study, but their numbers can translate into multiple seats in parliament when it comes to elections. Moreover, this finding is of interest by itself: Why do some of the sexually modern voters hold nativist attitudes and others not? This is an interesting question for further research.

We found quite some variance between the countries under study on the extent to which sexually modern nativists were found to be more likely to vote for PRRs. In particular, in Norway, Sweden, and Switzerland, an effect was found that sexually modern nativist were more likely to vote for the PRR, and in Austria, the sexually modern nativist voters catch up with the rest. The variance between countries could not be explained by the differences between PRRs in their position on traditional family values, refuting one of our key hypotheses. Arguably, the strongest case of anti-Islam, pro-gay rights rhetoric was found in the Netherlands, with both Fortuyn (2002) and Wilders (2006, 2010, and 2012) (Akkerman 2015; De Lange and Mügge 2015; Dudink 2016),¹⁴ but no effect has been found that suggested that sexually modern nativists were more likely to vote for these parties in the Netherlands (see also Spierings and Zaslove 2015a).¹⁵

A more fitting explanation for the country difference comes from broadening our supply-side scope by placing the PRR party in the larger party system, and particularly placing it within the context of the message of mainstream parties (i.e. non-PRR parties that are sceptical towards immigration). It has been argued that the presence of PRR parties in the party system has led mainstream right-wing parties to adopt stronger anti-migration positions (e.g. De Lange 2012). In some countries, these other parties are liberal-conservative parties, whereas in others, they are conservative or Christian-democratic parties. If the main anti-migration alternatives were traditional, we would expect a sexually modern nativist to vote for PRR parties, because this alternative does not match the voter’s views on LGBT rights. If, on the other hand, the second-most anti-migration party is a liberal party that supports LGBT rights, sexually modern nativist voters might decide to vote for that party instead of the PRR party.

At first sight, this idea does explain that no relationship is found in Denmark, Belgium, and the Netherlands, where Venstre (DK), Open VLD (BE), and VVD (NL) are credible alternatives as conservative-liberal parties who are generally supportive of LGBT rights and a liberal family morality. Also, but in a somewhat less straightforward way, an alternative is available in Finland, France, and Italy: the major right-wing parties that are at least more liberal on LGBT issues than the PRR or include clear liberal factions.¹⁶ In the six countries in which no clear effect was found, sexually modern nativists have a viable alternative when they are in the voting booth. This focus on the alternative anti-migration parties also seems to fit the cases in which a disproportional likelihood to vote for the PRR was found. Most straightforwardly, in Austria and Switzerland, the main alternative right-wing anti-migration parties are the (traditional) Christian-democratic ÖVP (AU) and CVP (SW).¹⁷ In Sweden, the main right-wing alternative, the Moderate Party, is relatively pro-LGBT rights, but not strongly anti-migration in the period we studied. As the Moderate Party did make a right turn on this issue recently, it is an interesting test case for our suggested explanation here.¹⁸ Similarly, the Norwegian case deserves more in-depth attention as the second-most anti-migration party – the Norwegian Conservative Party – is rather traditional, for instance, voting against same-sex marriage, but a few MPs deviated from the party line.¹⁹

The dynamics we proposed here should be studied in greater detail before strong claims can be made. The Islamizing or ethnicizing of progressive values has only relatively recently become a populist radical right party strategy to attract voters and it seems that the traditional parties that claimed to serve the interests of sexual minority groups have no clear answer to this shift among the PRR. Interviews with sexually modern nativist in the different countries could possibly provide more insight into the motivations to vote for or not to do so for the PRRs. Only for the Netherlands, the Manifesto data have explicitly revealed pro-gay, but nativist stances. Many of the other parties have been politically ambiguous, for example, voting against same-sex marriage, but referring to protecting gays when discussing the perceived threat of the Islamization of Europe. Therefore, further comparative research will require better indicators on PRR parties as well as their main contenders regarding positions on LGBT emancipation.

Disclosure Statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Niels Spierings is Assistant Professor in Sociology at Radboud University. His research interests include political behaviour, gender and sexuality, populism, social media, Islam, migration, democratic politics, and quantitative, qualitative and feminist research methods. On these topics he has published journal articles and monographs. In 2015 he co-edited a special issue on gender and the populist radical right in Patterns of Prejudice with Andrej Zaslove.

Marcel Lubbers is Professor in (Political) Sociology at Radboud University, Nijmegen. He has published internationally on understanding radical right voting and euroscepticism. His current work includes a focus on immigrant integration.

Andrej Zaslove is Assistant Professor, Department of Political Science, Radboud University, Institute for Management Research, Nijmegen, the Netherlands. His research focuses on populism (with a current emphasis on measuring populism, foreign policy and populism, and gender and populism). In 2015 he co-edited a special issue on gender and the populist radical right in Patterns of Prejudice with Niels Spierings.

Notes

1 We include the Lijst Pim Fortuyn as a PRR party, although it is a borderline case. Mudde (2007, 59) considers it to be a neo-liberal populist party, while others consider it to be a PRR party (De Lange and Art 2011). Following Rydgren and van Holsteyn (2005), we consider the LPF to be a functional equivalent of the PRR for the core question in this study.

2 We focus on PRR parties in Western Europe for two reasons. First, classifying parties in Central and Eastern Europe is tricky, as the boundary between the PRR and other types of populism is not always clear. Second, there are significant contextual differences between Western Europe and Central and Eastern Europe, both regarding supply and demand.

3 See T. Brandel “SD siktar in sig på hbt-röster”. Svenska Dageblat. 28 July 2010.

4 If two survey rounds were held between elections respondents in both surveys are grouped in one higher level unit (e.g. the 2002 Austrian elections). If data were collected overlapping an election, we split the sample by the day of interview and linked respondents to different elections (e.g. ESS1 in Belgium).

5 Technically speaking, these elections are also embedded in countries. Approximately 15% of the variation is found at the country level (24% at the election level, and the rest at the individual level). However, our random intercepts capture these differences and including an additional level does not influence the results. We therefore present the simplest models in the text. In Appendix 1, the results from a three-level model and a country-fixed-effects model are given.

6 To what extent do you think [country respondent] should allow people from the poorer countries outside Europe to come and live here? (Allow many to come and live here; Allow some; Allow a few; Allow none.); How about people of a different race or ethnic group from most [country] people?; How about people from the poorer countries outside Europe?; Could you say it is generally bad or good for [country]’s economy that people come to live here from other countries? (Bad 0 through Good 10); Would you say that [country]’s cultural life is generally undermined or enriched by people coming to live here from other countries? (Undermined 0 through Enriched 10); Is [country] made a worse or a better place to live by people coming to live here from other countries? (Worse 0 through Better 10) (ESS 2002–12a).

7 Exploratory Factor Analysis: Eigen value 3.777; Bartlett’s test p<.001; KMO 0.843; lowest loading 0.689; Cronbach’s alpha 0.881.

8 Models were rerun with two ‘interaction dummies’ separating the moderately and extremely strong anti-migration respondents. Both showed highly similar interaction coefficients, but the latter group was only 1.5% and therefore did not reach statistical significance at conventional levels.

9 The Chapel Hill expert data are not coded per election and do not include Norway and Switzerland.

10 Other codes also include items relating to sexuality but tend to conflate it with other minorities, which is problematic for this study. For example, the manifesto items on equality measure positive attitudes towards equality policies, conflating sexual equality, class equality, and the position of underprivileged groups. Similarly, the item on ‘underprivileged minority groups’ can include statements supporting gay and lesbian people as well as ethnic minorities.

11 Given the small group size, we include marginally significant coefficients for now. Mostly, these were found for countries that showed a conventionally statistical significant relationship for another election as well.

12 A Mann–Whitney U test, for instance, shows a p value of .381; the two averages being 1.55 and 1.00, on the scale from 0 to 4.

13 Results can be obtained from the authors.

14 In the case of Wilders, it is actually more than claims making: the party supported the abolishment of the ‘defiant clerks’ (clerks who did not want to marry same-sex couples despite the law) in the parliamentary vote, even though this could have estranged the orthodox Christian party which government often needed to reach a majority (the PVV supported the minority government) in the Senate.

15 In terms of votes and political impact, this means that if the PVV would manage to increase the voting probability among the sexually liberal nativists of about 7% with only 10 percentage points, it gets an additional seat (they currently hold 12).

16 The Finnish National Coalition Party is a viable liberal and anti-migration party. For instance, it supports same-sex marriage (Bucken-Knapp et al. 2014; E. Tessieri, “National Coalition Party and Perussuomalaiset lead anti-immigration drive in Finland” Migrant Tales, 12 September 2014). In France and Italy, the major right-wing parties (Les Républicains [former UMP] and Forza Italia) are broad coalitions that include liberals and liberal-conservatives. For instance, Berlusconi has been said to be steering the party back towards its liberal roots (See “Pd-azzurri: asse sui diritti (senza Ncd)" Corriere della Sera, 4 January 2014; G. Dinmore "Berlusconi seeks return to liberal roots” Financial Times, 22 July 2012).

17 The liberal FDP is far less anti-migrant; the CVP is the foremost alternative to the SVP.

18 See M. Crow, “Sweden’s Moderate Party Flipflop On Immigration” The European Guardian – The Home of Europe’s New Right, 8 May 2015. The smaller strongest anti-migration party in Sweden, the Christian-democrats, is more traditional and thus not a real alternative either.

19 See S. Picheta “Norway new gay marriage law also grants new parental right” Pinknews, 12 June 2008.

Appendix 1. Alternative models testing sensitivity of results to estimation technique and operationalization.

Table

	Interval with country as additional level		Interval with fixed effect for country effects		Dummies with country as additional level		Dummies with fixed effect for country effects
	Logged odds	SE	Logged odds	SE	Logged odds	SE	Logged odds	SE
Anti-migration; ref = 1: not against migration
2					1.35***	0.15	1.35***	0.15
3					2.36***	0.15	2.36***	0.15
4					3.00***	0.15	3.00***	0.16
5: extremely anti-migration					3.46***	0.16	3.46***	0.17
Anti-migration attitudes – continuous (0–1)	3.07***	0.31	3.08***	0.31
Pro-LGBT rights
1: strongly against					0.01	0.12	0.01	0.12
2					0.22**	0.09	0.22**	0.09
3 (ref)
4					−0.09	0.06	−0.09	0.06
5: strongly in favour					−0.28**	0.07	−0.28**	0.07
Pro-LGBT rights – interval (1–5)	−0.37***	0.06	−0.36***	0.06
Sexually modern nativist (ref = no)					0.21**	0.08	0.21**	0.08
Interaction anti-migration × pro-LGBT rights	0.48***	0.09	0.47***	0.09
Micro-level controls included

Note: All models are two-level models with random intercepts at the election level, unless mentioned otherwise.

Source: ESS round 1–6.

*p < .05.

**p < .01.

***p < .001.

Appendix 2. Interaction models per country for which initial interaction effects were found.

Table

	Austria		Norway		Sweden		Switzerland
	Dummy	Interval	Dummy	Interval	Dummy	Interval	Dummy	Interval
Anti-migration; ref = 1: not against migration
2	0.57		1.09***		2.04^#		1.53***
3	1.57*		1.94***		3.79***		2.52***
4	2.08**		2.48***		4.27***		2.98***
5: extremely anti-migration	3.09***		2.76***		4.99***		4.07***
Anti-migration attitudes – continuous (0–1)		2.91**		2.04**		2.49		2.19**
Pro-LGBT rights; ref = 1: strongly against
2	0.02		0.28		−1.81		0.49^#
3	−0.39
		−0.22		−2.23^#		0.47
4	−0.80*		−0.11		−1.83^#		0.47^#
5: strongly in favour	−1.75***		−0.07		−2.28^#		−0.15
Pro-LGBT rights – interval (1–5)		−0.79**		−0.41**		−0.86		−0.62***
Sexually modern nativist (ref = no)	1 . 32**		0 . 29		1 . 19^#		0 . 61*
Interaction anti-migration × pro-LGBT rights		0 . 78*		0 . 71**		1 . 44^#		1 . 08***

Note: Only coefficients for interaction dummy are given; control variables are included in the models though (see Table 3).

Source: ESS round 1–6.

***p < .001.

**p < .01.

*p < .05.

^#p < .10.