Radical right parties have become a permanent feature of most West European polities since the 1980s, but Germany has bucked this trend for decades. While parties to the right of the Christian Democrats and the Liberals occasionally found support in subnational elections, far-right politicians faced very high political and social hurdles in Germany, and electoral success at the national level remained elusive. This has changed only recently, when in 2017 the populist radical right “Alternative for Germany” (AfD) garnered 12.6 per cent of the vote and became the first new party to enter the Bundestag since the 1990s.
Two factors can help to understand this unexpected development. First, the AfD did not start out as a radical right party but as an outfit that combined soft euroscepticism with economic liberalism and socially conservative policies (Arzheimer 2015). Although the AfD, from its very beginning, attracted various stripes of right-wingers as members, its leadership initially comprised mainly of disappointed members of the German elite including many professors, lawyers, doctors, and former centre-right politicians, who carefully avoided any associations with traditional German right-wing extremism or the modern radical right parties in other European countries. This strategy made it possible for the party to gain access to the mainstream media and win subnational political representation in a string of state-level elections. Only in 2015, when the AfD’s then-leader and many of his supporters left the party, the AfD adopted a populist stance and began to focus on immigration, refugees, and Islam as their new core issues, while the politics of the Euro zone took a back seat.[suffusion-multic] [suffusion-column width=’1/2′] Second, the party’s transformation was followed by Angela Merkel’s unexpected decision to temporarily suspend the Dublin Regulation, which in turn led to the unorderly arrival of hundreds of thousands of asylum seekers in Germany. As a result, the salience of the asylum/migration issue rose, and the AfD’s re-positioning has proven very successful ever since. While the AfD remained just under the electoral threshold in the 2013 general election, a remarkable result for a new party, and had some respectable electoral successes from 2014 to early 2015, it has done spectacularly well since 2016. Our core assumption is that behind this development is a dramatic change in the motives of AfD voters. More specifically, the ideological shift of the AfD within a fundamentally changed environment induced voters to decide on the basis of migration attitudes. By testing this assumption, we also ascertain whether the AfD is in any sense a special case as previous research on the party would suggest, or whether support for them is driven by the same
perceptions of ethnic threat that radical right electorates in other countries hold.
[/suffusion-column] [suffusion-column width=’1/2′] [/suffusion-column] [/suffusion-multic]
The remainder of this paper is structured as follows. In the next section, we present our conceptual framework and spell out our theoretical expectations. We then review the existing research on the AfD’s ideology and voters and locate our own contribution within this larger body of knowledge. Following that, we analyse data from the 2013 to 2017 series of German Longitudinal Elections Study (GLES) tracking polls, a series of high-quality quarterly academic election surveys, to trace the evolution of the party’s electorate. Our most important finding is that attitudes towards immigration, which were unrelated to the AfD vote in 2013, are now its most important driver. The final section presents our conclusions, and discusses avenues for further research.
In the 1980s, a new party family on the political right has emerged in Europe. The most prominent approach to describe, delineate, and differentiate these parties was developed by Cas Mudde (2007), who calls this family the “radical right”.
According to Mudde, radical right parties combine nativism – a tendency to regard non-native elements as a threat to the homogeneous nation-state – with authoritarian tendencies (Mudde 2007, 21–23). What makes them “radical” is their opposition to some core principles of liberal democracy including pluralism, constitutional protection of minorities, and a civic conception of citizenship (Mudde 2007, 25).
A large sub-group of the party family is also populist: they pit the “pure” and homogeneous people against an allegedly corrupt and self-serving elite and claim to speak for the “silent majority” or the “true nation” (Mudde 2007, 23). Consequently, parties of the populist radical right are often fervent supporters of direct democracy and could therefore be dubbed “illiberal” democrats (Canovan 1999).
Only a small sub-group of the radical right openly opposes democracy per se. In a departure from his earlier work, Mudde calls these parties the “extreme right” (Mudde 2007, 24).
While radical right parties have become a fixture in most West European countries, Germany has long been a negative case. Although previous analyses have shown that there has been considerable latent demand for radical right policies in Germany, national electoral success for radical right parties remained elusive (Berbuir, Lewandowsky, and Siri 2014).
The research literature has identified two related reasons for this: first, parties at the far end of the left-right spectrum remained ideologically beholden to the legacy of National Socialism (Kitschelt 1995, 203-6), which rendered them unelectable and increasingly irrelevant for many potential voters. Second, because of their association with National Socialism and the resulting stigmatisation, these parties failed to attract capable politicians (Art 2011, 79). The unusually broad ideological appeal of the mainstream Christian Democrats (CDU) and their standing electoral alliance with their Bavarian sister party CSU, whose strong anti-immigration stance left little room for a radical right competitor (Lubbers, Gijsberts, and Scheepers 2002, 359) provide a further and complementary explanation. In a recent contribution, Art (2018, 80) even argues that in the past, the German parties, media, and civil society actors “had executed containment [of the radical right] close to perfection”.
Consequently, attempts to found modern radical right parties never got off the ground or remained confined to local/state-level politics (Decker 2008, 120; 129–31). The rise of the AfD since 2013 represents the first nation-wide deviation from this pattern.
In our view, this structural break can be explained by the fact that the AfD began as a soft-eurosceptic, market-liberal and socially conservative outfit that only transformed into a fully-fledged populist radical right party once it had gained a foothold in several state parliaments. The strong social norm against voting for extremists in post-war Germany would have stigmatized the party and only the gradual transformation allowed the party to leverage their full electoral potential.
While this transformation was particularly fast and dramatic, it is by no means unprecedented in European politics. The United Kingdom Independence Party (UKIP) and the Hungarian Fidesz are cases in point. Just like the AfD, UKIP was founded by conservative intellectuals pushing hardline EU policies before they leveraged upon the stereotypical radical electorate (Ford and Goodwin 2014b). Fidesz in Hungary started out as a centre right-wing party in the 1990s (and is still a member of the European People’s Party) but became one of the strongest radical right forces in Europe (Krekó and Mayer 2015). While both UKIP and Fidesz differ from the AfD in important ways, their respective trajectories illustrate how more traditional centre-right parties can change their electoral fortunes by crossing over to the radical right.
As a consequence of the AfD’s transformation and the Christian Democrats’ new leniency on citizenship and asylum, latent anti-immigration sentiment could come to sustain the not-so-new party. Accordingly, our hypotheses are these:
As the AfD’s right-wing stance hardened over time, the effect of general ideology on the AfD vote increased
As the AfD’s nativism increased and the party took ownership of the immigration issue, the effect of anti-immigration sentiment on the AfD vote increased
As leading market liberals left and the party became more anti-elite and (even) more isolated, the effect of market-liberalism on the AfD vote decreased
We do not test the effect of eurosceptic attitudes on the AfD vote, because the AfD’s euroscepticism was first derived from and constrained by the party’s ordo-liberal principles, then became tied to its growing nativism. Moreover, because of the (relatively) low salience of EU politics in Germany, eurosceptic attitudes are measured rarely and inconsistently in the available data sets.
Before we turn to testing our hypotheses, we need to address two points. First, we summarise the scholarly evidence on the AfD’s leadership and ideology and complement this presentation with an original content analysis of the party’s social media posts. Second, we will review the (few) existing studies on the AfD’s electorate.
Unsurprisingly, the meteoric rise of the AfD has prompted a flurry of academic enquiries into its nature. Researchers have collected and analysed information on the party’s manifestos, leadership, and parliamentary candidates.
During its founding phase in 2012/2013, the AfD attracted a broad coalition of right-wingers as members (Arzheimer 2015), ranging from ordo-liberal economists to Christian fundamentalists, but the party’s top-tier was dominated by (mostly) men who belonged to Germany’s elite and had been, or could have been, members of Germany’s traditional centre-right parties. Of the original 58 signatories of the “Manifesto for an Electoral Alternative 2013” that predated the formal founding and registration of the party by some months, 28 were academics and 14 more held a PhD or medical doctorate, earning the party the nickname “professors’ party.”2 With this ideological and social profile, the AfD could very well be described as a “conservative challenger to the CDU/CSU” (Dilling 2018, 86), operating in a space that had been occupied by the Christian Democrats before Merkel moved the party towards the centre, by, inter alia, ending the draft, abolishing nuclear energy, and introducing a national minimum wage.
Accordingly, the AfD’s candidates in the 2013 federal election were more market liberal but no more authoritarian than their competitors from the CDU (Jankowski, Schneider, and Tepe 2016). However, although the party’s first and very short manifesto highlighted the need to attract large numbers of skilled immigrants, the AfD candidates were even at this early stage more willing to curb immigration than their CDU counterparts (Ceyhan 2016).3
Moreover, Franzmann (2016) suggests that the AfD’s leadership occasionally toyed with a “populist style” around the 2013 election, but later moderated its tone. This chimes with Schmitt-Beck’s (2017) finding that the AfD’s 2013 electorate comprised two distinct groups: early deciders, who were motivated by the party’s stance against the Euro bailout packages, and a somewhat bigger group of late deciders, who were primarily drawn by the AfD’s dog whistle messages (Lees 2018, 12) on immigration. Nonetheless, most authors agree that during the first two years of its existence, the AfD was a soft-eurosceptic party, but not (yet) populist or radically right (Berbuir, Lewandowsky, and Siri 2014; Arzheimer 2015; Grimm 2015; Schmitt-Beck 2017).
The issue of the AfD’s radical right-wing populism became moot early in July 2015: two months before Merkel’s controversial decision to suspend the Dublin Regulation and after an intense struggle between moderate and more radical forces over the course of the party, which had dominated coverage of the AfD since at least January 2015, the leading moderates including five of the seven MEPs left the AfD after their main proponent Bernd Lucke was ousted as party co-chair (Dilling 2018, 98).
After this de-facto split, the AfD quickly radicalised. Xenophobic and populist positions that had before been at least controversial within the party became mainstream. A ban on any co-operation with the islamophobic Pegida movement was relaxed and later lifted completely. The onset of the so-called refugee crisis in September 20154 – “a gift to the party”, according to one leading AfD politician – further fuelled these developments. Even openly racist statements and attempts to minimise the Holocaust by party leaders were no longer beyond the pale (Art 2018, 81).
Lucke’s successor Frauke Petry, who had come to power with the tacit support of the most radical factions within the AfD, made headlines by floating the idea that refugees could be shot at the border. She also tried to raise her profile by forming a vague alliance with the respective leaders of the Austrian FPÖ, the Dutch PVV, the Italian Lega, and the French FN. However, Petry could not, or would not, keep up with the pace of the AfD’s radicalisation. Her long-term strategy for developing the AfD’s coalition potential was not even debated by the 2017 party conference, and she was quickly side-lined as the AfD’s frontrunner. Immediately after the federal election, she stepped down and subsequently left the party (Dilling 2018, 99). As a result of these developments, there is now “growing scholarly consensus that the AfD fits Mudde’s (2007) conceptualisation of the populist radical right” (Hansen and Olsen 2018, 3).
For the first time in more than six decades, there is now a radical and nationally viable alternative to Germany’s highly successful Christian democrats. The issue of ideological competition between the centre and the radical right is a complex one. While some authors claim that raising the saliency of immigration will benefit the radical right (Arzheimer 2018), others have argued that historically, the centre right has owned the issue (Bale 2008), and that under certain conditions, the centre-right can still curb radical right success by taking a tough line on immigration (Pardos-Prado 2015; Downes and Loveless 2018).
There is little doubt that the AfD’s new emphasis on migration was crucial for the party’s success, and that Merkel’s decision to allow hundreds of thousands of refugees stranded in neighbouring countries to enter Germany became the focus of the AfD’s attack on the CDU/CSU. We will therefore return to the question of issue saliency in a later section.
It is not immediately clear how this ideological trajectory can be best quantified and tracked. While the party has published manifestos and position papers on the federal level that reflect the party’s transformation (Dilling 2018, 86–89, Rosenfelder 2017), these are few and far between. Manifestos at the state level and input provided for voting advice applications (Linhart 2017) complement the national platforms, but they do not just reflect variation over time but also the considerable differences between state-level party chapters.
In our view, social media data provide the best insight into the pace and shape of the AfD’s transformation. From its very beginning, the party has relied on Facebook in particular for both internal communication (Schreiber 2018) and for addressing their supporters and potential voters. Of all German parties, the AfD has by far the strongest presence on Facebook, and their many private and institutional accounts form a hub for right-wing discourses in the German-speaking internet (Schelter et al. 2016, Stier et al. 2017). For our purposes, the AfD’s central account is particularly interesting. It is not linked to a single politician but to the national party as a whole, and has posted several times per day on average since the party was founded in 2013. As of November 2018, it has been “liked” by 433,000 users, more than the Christian Democrats (183,000) and the Social Democrats (187,000) combined.
For the analysis, we look at all 3,886 messages that the AfD posted between March 11, 2013 (the day they started using Facebook) and September 24, 2017 (the day of the latest federal election) on their own page. We use a basic keyword-based approach to track how attention to the party’s core issues varies over time.5 Our findings are very clear (see Figure 1): the share of posts mentioning the issues which brought the party into life in 2013 – the Euro and the European institutions (solid line) on the one hand, and Greece (dotted line) on the other – has declined considerably, particularly after the height of the Euro crisis in summer 2015.6 Asylum and immigration (dash-dotted line), on the other hand, skyrocketed in the second half of 2015 and have remained prominent since then. Finally, the “Islam” issue (dashed line) only took off over the course of 2016 and 2017, after the immediate salience of the migration crisis has decreased somewhat. Thus, the AfD is not a single-issue party any more, but has developed a clear focus on common populist radical right topics: migration, Europe, and Islam. In sum, the social media data show how the AfD transformed from a soft-eurosceptic to a populist radical right party.
Figure 1: Selected issues raised by the AfD in their Facebook posts, 2013-2017
In a bid to cross-validate our findings, we also looked at data from the Chapel Hill Expert Survey (CHES), which covered the AfD in its 2014 and 2017 rounds. Compared to 2014, the party moved to the right on the general left-right and on the cultural dimension. On the economic dimension, there is a slight move to the left, which is in line with the European radical right’s well-known welfare-chauvinist tendencies. More importantly, the salience of the economic dimension dropped dramatically, while the importance of the cultural dimension rose (see Figure A1 in the appendix).
Because the AfD is still a relatively new party, only a handful of studies have looked into their national voter base, some of which examine very specific research questions. Moreover, most existing studies provide only snapshots of the electorate at a specific point in time or in a given election.
The 2013 federal election poses a special problem because there were so few AfD voters, but using the 2013 GLES short-term campaign panel Schwarzbözl and Fatke (2016) show that even then, AfD voters were more sceptical of European integration and of immigration than any other group of voters. Similarly, Schmitt-Beck’s aforementioned analysis of the GLES 2013 rolling cross section campaign survey and post-election panel wave demonstrates that early deciders were chiefly motivated by euroscepticism, whereas later decisions were driven by anti-immigration attitudes (Schmitt-Beck 2017).
Using a cross-sectional non-academic survey conducted in December 2016, Lengfeld (2017) claims that the “losers of modernisation”, whom he defines as voters with a low socio-economic status (SES), were no more likely to vote for the AfD than other voters. Lengfeld takes this as evidence that the AfD vote has nothing to do with modernisation and/or globalisation but is rather driven by a strong preference for cultural homogeneity and a fully autonomous and independent nation-state. It is, however, important to note that Lengfeld’s survey contains no items that actually tap into these alleged attitudinal drivers.
Similarly, an analysis of the 2015 wave of the Socio-Economic Panel (SOEP) by Bergmann, Diermeier, and Niehues (2017) shows that most AfD leaners come from the middle quintile of the income distribution. However, the authors also present some evidence, collected from the SOEP and the headline findings of a smattering of other surveys, that AfD voters and supporters are more pessimistic than others about the economy and their future personal economic situation, and that they worry much more about crime, immigration, and social cohesion.
Lux (2018) argues that Lengfeld’s operationalisation of socio-economic status was inadequate and that the number of AfD voters in his survey was much too low for a proper analysis of subgroups anyway. Using data from Germany’s 2016 General Social Survey (ALLBUS), he identifies (weak) effects of low SES. Like Lengfeld, he also finds a small effect of subjective deprivation. Using data from the 2016 ALLBUS, the 2016 European Social Survey (ESS) and the 2016 wave of the GLES long-term panel, Tutic and Hermanni (2018) also try to improve on Lengfeld’s operationalisation of low SES as an indicator of “loser of modernisation” status. However, neither of these analyses incorporates attitudes on immigration or other genuinely political attitudes.
The same is not true for Hambauer and Mays (2017). Using a December 2015 GLES tracking poll, they show that three months after the onset of the so-called refugee crisis AfD voters were less willing to accept refugees than voters of other parties. This holds across a whole host of subgroups (i.e. refugees from war zones, people who flee from famine or natural disasters, or people who are persecuted for their religious beliefs). Having said that, the willingness to accept refugees from war zones was surprisingly high even amongst AfD voters (81 per cent). Another tracking poll conducted in June 2016 when scepticism had become much more widespread shows that a negative view of the refugee situation, fear of refugees, the feeling of being kept in the dark about the situation, a sense of being negatively affected by the influx of refugees, an identification as being on the political right and a general dissatisfaction with the way democracy works in Germany all had very strong and highly significant positive effects on the probability of voting for the AfD.
Similarly, Goerres, Spies, and Kumlin (2018), who conducted an online survey in May 2016, find that being critical of the institution of political asylum and welfare chauvinism, i.e. the wish to restrict welfare benefits to the native population, are very strong predictors of AfD support (measured as propensity to ever vote for the party). More generally, support for redistribution from the rich to the poor reduces support for the AfD.
This contrasts with Hansen and Olsen (2018), whose analysis of the 2017 GLES post-election survey shows no significant effects of attitudes on redistribution. One should, however, bear in mind that they use a different dependent variable (vote choice), a different model (multinomial logit), and a different indicator attitudes towards redistribution. More importantly perhaps, Hansen and Olsen (2018) also find a very strong effect of anti-immigrant attitudes, which is in line with Dilling’s (2018) analysis of the same data set.
What all these studies have in common is that they look at the AfD’s electorate at a single point in time or, in the case of Schmitt-Beck (2017), over the span of a period of less than three months. The first study that takes a longer view is that by Bieber, Rossteutscher, and Scherer (2018), who employ a series of 14 GLES tracking polls, which span a period from May 2013 to September 2016. Building on the conception of “anti-party parties” developed in the 1990s (Poguntke and Scarrow 1996), they identify three potential motives for supporting the AfD, which correspond with three equivalent sub-groups of AfD voters: “anti-party sentiment”, i.e. a dislike for the established parties, “loyal voting” that is grounded in holding core political convictions aligned with the aims of the anti-party party, and “protest voting” that signals dissatisfaction with an ideological shift of a previously preferred party. Somewhat surprisingly, they take migration scepticism, worries about the European debt crisis, and left-right self-placement as indicative of the latter.
While we agree that such a “tactical” (Alvarez, Kiewiet, and Núñez 2018) pattern of protest voting may apply to some AfD voters, we would question its operationalisation by Bieber, Rossteutscher, and Scherer (2018). At a minimum, they would need to ascertain that a tactical protest voter a) has a first preference for another party, b) perceives this party as to the left of her own position, and c) believes that the original party will respond to protest voting by moving back towards the right. In our view, estimating the coefficients for the attitudinal variables unconditionally across all respondents (Bieber, Rossteutscher, and Scherer 2018, 451) renders them indistinguishable from the effects of simple issue voting or, in the case of left-right self-placement, ideological voting.
Moreover, we are worried about the inclusion of identification with the AfD in the model, which Bieber, Rossteutscher, and Scherer (2018) use as an indicator for “loyal” voting. While we agree with Arzheimer (2018, 145–46) that a fully specified model of radical right voting should take party identifications into consideration, the AfD was just over three years old at the end of the period under study. Even if we assume that stable long-term identifications had already formed by 2016, these would largely have been the result of issue preferences that were not met by the established parties. Controlling for party identification will therefore induce post-treatment bias (King and Zeng 2006, 147–48), i.e. the effect of issue voting will be underestimated.
To summarise: previous research suggests that the remarkable growth of the AfD’s support, which more than doubled between the 2013 and 2017 elections, is the result of an equally remarkable change in the socio-economic and attitudinal factors that predispose voters towards the AfD. More specifically, there is some initial evidence that over the course of these four years the AfD, initially a de facto breakaway from the centre-right parties, began to attract a clientele that closely resembles the prototype of the radical right’s electorate. What is lacking so far (with the partial exception of Bieber, Rossteutscher, and Scherer (2018) is a longitudinal model that tracks the development of the AfD’s electorate over this crucial period, using consistent operationalisations for socio-economic and socio-cultural attitudes. In the next section, we aim to close this gap.
Micro data for the analyses come from the long-term tracking component of the German Longitudinal Election Study (GLES), a quarterly online survey that monitors voting intentions and public opinion for the German academic community.7 For the period from May 2013 to September 2017, there are 18 individual samples with approximately 1,000 respondents each. They are supplemented by ten smaller (n 500) “booster samples” that were collected regionally around state-level elections during that time. In each case, the surveys were in the field for about two weeks, and the exact date of each interview is recorded in the data set. Because of this design, the long-term tracking component is ideally suited for the task at hand. It covers the whole period of the party’s existence, with short intervals of three months or less between individual polls.
Our analysis of the GLES tracking data builds and improves on the existing research in several crucial ways. First, ours is the only analysis to cover the full period from the AfD’s appearance in opinion polls in May 2013 to their first entry in the Bundestag in September 2017. It is therefore particularly well suited for looking into the transformation of the AfD’s electorate over time.
Second, we acknowledge the fact that some crucial attitudinal items that were not included in each and every of the GLES tracking polls for reasons of cost and space are effectively missing completely at random and hence excellent candidates for multiple imputation (MI). By using MI, we are able to include many more individual samples than previous analyses, providing much improved temporal coverage. Adding the 2017 surveys, making use of the booster samples, and MI collectively allow us to include twice as many samples, three times as many AfD voters, and four times as many total respondents in our model as Bieber, Rossteutscher, and Scherer (2018) did.
Third, against the backdrop of this dense time-series, we do not rely on somewhat arbitrary “periods” in the development of the AfD and its electorate. Instead, our model includes the main effect of and interactions with calendar time to directly capture changes in the baseline of AfD support and in the magnitude and direction of the effects of its predictors (Firebaugh 1997).8 Fourth and finally, we account for the clustering of our observations in time and the resulting non-independence by estimating a multi-level model.
While online interviews raise concerns regarding the representativeness of the resulting sample, GLES participants were recruited offline, which should ameliorate these problems. This is confirmed by the distributions of some key demographic variables (gender, education, region). They closely resemble those in the ALLBUS studies, which relies on face-to-face interviews and is considered the gold standard in survey research in Germany. The only major difference is with respect to age: the GLES respondents are markedly younger, but because demographic variables are controlled for, this should not constitute a problem.
The dependent variable is the intention to vote for the AfD in the next federal election. While it would have been interesting to compare AfD voters with non-voters and supporters of the small right-wing extremist parties, even the former are severely underrepresented in German election studies, while the latter can hardly be captured at all. Therefore, voting intention was dichotomised, with 1 representing an intention to vote for the AfD and 0 expressing support for any other party, and a binary logit model was specified.
From our hypotheses, it follows that general ideology, anti-immigration sentiment, and support for market-liberalism are our core independent variables (see Table 1). Controlling for general ideology in addition to the more specific attitudes implies a much more conservative test for their relative effects.
Table 1: Operationalisation of attitudinal variables
Left (1) / right (11)
More welfare benefits, even if that means higher taxes
(1) / lower taxes, even if that means fewer welfare benefits (11)
It should be made easier (1) / more dicult (11) for
foreigners to move to Germany
Market liberalism and anti-immigration sentiment are missing by design in five/ten polls, respectively. Because this information is missing completely at random (MCAR) and because item non-response rates are otherwise low9, the data were multiply imputed using chained equations (MICE) so that all 28 individual polls could be included in the analysis.10 While multiple imputation adds an additional layer of complexity to the model, it greatly boosts the number of usable cases, and more importantly, the temporal coverage (see Figure A2 in the appendix).
Because the electorates of radical right parties have a well-known socio-demographic profile (Rydgren 2013), we include gender, employment status, current or previous occupation, region, age group, and religious affiliation. However, previous research has also shown that these variables often have no or very small direct effects once attitudes are controlled for (e.g. Arzheimer and Carter 2009, 1000–1001).
Finally, all variables were interacted with the date on which the interview was conducted (in calendar days, centred on the mid-point of the period) so that changes in the strength and direction of the associated effects can be tested. To simplify the analysis, time is assumed to have an effect that is linear in the logits.
Although the date of the interview is controlled for, respondents are “nested” within individual days and hence will be subject to common random political shocks. To account for these, a two-level structure that includes a random intercept at the interview-date level was specified.11 Parameters were estimated using Stata’s (15.1) meqrlogit procedure.
Table 2: Multi-level model of AfD voting: main effects (fixed)
note: * = p<0.05; ** = p<0.01; *** = p<0.001
Table 3: Multi-level model of AfD voting: interaction effects (fixed)
male × time
employee × time
public servant × time
self-employed professional × time
other self-employed × time
farmer etc. × time
in training × time
medium × time
high × time
east × time
catholic × time
other christian × time
other religion × time
no religion × time
unemployed × time
30-39 × time
40-49 × time
50-59 × time
+60 × time
left/right ideology × time
tax/welfare × time
anti-immigration × time
note: * = p<0.05; ** = p<0.01; *** = p<0.001
Table 4: Multi-level model of AfD voting: random effect and number of cases
note: *** = p<0.001
Before estimating the parameters of the full model, an intercept-only model was specified to assess the importance of the “nesting” (not shown). The estimated variance at the upper (poll) level was .072. Assuming the fixed variance of 3.29 at the individual level that is implied by the logistic structure of the model (see Goldstein, Browne, and Rasbash 2002), the variance partition coefficient (VPC) is 0.022, i.e. about 98 per cent of the total variation occurs at the individual level. This means that the multi-level structure is of little substantive interest and can be treated as a mere inconvenience.
Including the individual-level variables in the model results in further drop of the contextual variance, which is now estimated at 0.044 (see Table 4).12 While this number is very small, it should not be completely discounted. Substantively, a variance of 0.044 means that for 50 per cent of the polls, the logit will be shifted by up to points on the logistic scale. For 90 per cent of the polls, the shift will be in the range of .
Because of the presence of multiplicative interaction terms, the interpretation of the regression coefficients requires great care. The main effect of time in Table 2 refers only to those respondents who fall in the reference categories of all the dichotomous/polytomous variables13 and place themselves at the midpoint of the ideological scales. For them, the effect of time is not statistically significant, so their likelihood of voting for the AfD does not change.
Conversely, the main effects of all other variables in Table 2 refer to the mid-point of the observation period (July 2015). At this time, men were substantively more likely to support the AfD than women. This overrepresentation is hardly surprising and rather consistently documented in the literature (Givens 2004; Coffé 2018). Living in the territory of the former GDR also increased the odds of an AfD vote. Occupation had no significant effect, but highly educated respondents were less likely to vote for the AfD. In line with expectations, Catholics were less likely to vote for the AfD than Protestants, whereas non-religious voters were substantially more likely to support the AfD. With respect to age groups, our analyses support what has previously been noted in commercial surveys: voters beyond the age of 60 were significantly less likely to vote AfD, presumably because they identify with one of the established parties. However, the finding that people aged 30 to 39 years were significantly more likely to support the AfD than even younger voters is unexpected. One should, however, bear in mind that the effect is small and refers only to July 2015, and that the differences between this and the two next age groups are themselves not statistically significant. Finally, positioning oneself on the political right greatly increased the likelihood of an AfD vote, or, put differently, voters who see themselves as left of centre would hardly ever consider voting for the party. A restrictive position on immigration, on the other hand, exerted a sizeable effect on the probability of an AfD vote, even after controlling for general left-right ideology. Favouring higher taxes and more welfare benefits had a negative effect, albeit a weak one.
To understand how these effects have developed over time, one needs to turn to the interaction effects (see Table 3 and Figure A3 in the appendix). Because the value of multiplicative interaction terms depends on the arbitrary scaling of the constituent terms, their statistical significance is less relevant than their direction and magnitude (Brambor, Clark, and Golder 2006).14
For most of the socio-demographics, the AfD’s profile seems to become clearer over time. Because the explanatory variables’ relationship to this probability is non-linear to begin with and is further complicated by the interaction terms, the impact of any given variable is best assessed by graphing model-adjusted predictions and (average) marginal effects. In such a plot, differences in probabilities are calculated by varying the value of the focal variable while holding all other variables at their actual values, thereby integrating over them.
As an example, Figure 2 shows that in May 2013, the estimated level of support for the AfD was four percentage points higher for men than for women. By September 2017, this gender gap had grown by two percentage points, but because of the wide confidence intervals, one cannot be sure that the increase is real. In other words, while the change in the effect is not significant, the effect itself is: one can be very certain that the AfD has disproportionate support amongst men, because the lower bounds of the confidence interval are far away from zero.
Our results for the attitudinal variables are much more clear-cut. Even in 2013, when the AfD was a new party, the average marginal effect of a one-point change on the general ideology scale was a 1.5 point increase in the probability of an AfD vote (see Figure 3). By 2017, the size of the effect had almost doubled, reflecting the clearer profile of the party. Hypothesis 1 is thus confirmed.
Figure 3: Marginal effect of left-right ideology on AfD support over time
Net of this effect of general ideology, market liberalism has no relevant effect on the AfD vote (see Figure 4). Because the variable is coded so that higher values correspond to a preference for higher taxes/more welfare spending, the trend is in line with hypothesis 3, but the effect was very weak to begin with and only achieves statistical significance during the middle of the observation period.
Figure 4: Marginal effect of economic ideology on AfD support over time
Attitudes on immigration (see Figure 5), on the other hand, had no discernible effect in 2013, but for 2015, their marginal effect is already estimated at more than 1.5 percentage points, even though general ideology is controlled for. In 2017, the magnitude of the effect had at least doubled, resoundingly confirming hypothesis 2: a one-point change on the eleven-point scale now increase the probability of an AfD vote by roughly four percentage points. Anti-immigration attitudes now provide by far the clearest separation between supporters of the AfD and all other voters, and their political importance is hard to overstate.
Figure 5: Marginal effect of attitudes on immigration on AfD support over time
Importantly, the AfD’s success was apparently not contingent on an increase in anti-immigration sentiment per se. Although a very large number of asylum seekers arrived in 2015 and 2016, the GLES data show that for most of the four-year period, immigration attitudes were not significantly different from their long-term average of seven points (see Figure 6). Their variance and kurtosis have also remained essentially constant, suggesting that there was no opinion polarisation (see Figure A4 in the appendix). What has changed is the importance of these attitudes for the AfD vote: support for the AfD is now predominantly motivated by preferences for restrictive immigration policies, reflecting the party’s changing ideological appeal and the generally increased salience of the issue for Germans’ voting behaviour (see Figure 7).
Figure 6: Immigration attitudes over time
Figure 7: Self-reported salience of immigration attitudes in Germany, 2013-2017
The point of departure for this paper was the observation that for the first time, Germany has a populist radical right-wing party that is entrenched at the national level. The rise of the AfD was facilitated by the fact that it began its political life as a softly eurosceptic, non-radical party that attracted voters and members from a relatively broad spectrum. The leadership crisis and transformation in 2015 could have been the end of the party, but the arrival of hundreds of thousands of asylum seekers firmly put the main issue of the radical right on top of the public’s agenda, resulting in a massive increase of support for the AfD.
In the process, the AfD’s electorate changed, reflecting the transformation of the party. They now closely resemble the somewhat stereotypical image of populist radical right-wing voters in Western Europe. Crucially, general right-wing views and negative attitudes towards immigration, which played no or only a minor role in the early days, are now the main drivers of AfD support. Taken together, we find evidence for the transformation of the AfD’s electorate towards the prototype of the West European populist radical right-wing voter.
Our results speak to the existing literature in several ways. First, our findings support evidence from other European countries on the importance of anti-immigrant sentiments (Arzheimer 2018; Ford, Goodwin, and Cutts 2011; Lubbers, Gijsberts, and Scheepers 2002; Rydgren 2008; Brug, Fennema, and Tillie 2000). While the AfD framed their approach as a specifically “German Alternative” to Merkel’s “there is no alternative” policies, they in fact mobilised on the issue of (perceived) ethnic threat, just like any other radical right party (Arzheimer 2018). Second, the findings on the socio-demographics point to a group that presumably feels that they are “left behind” and miss out on the benefits of modernisation (Betz 1994; Ford and Goodwin 2014a; Goodwin and Milazzo 2005). Lastly, our findings contribute to a related debate on the (ir-)relevance of neo-liberal economic attitudes for radical right voting (Arzheimer 2009; Ivarsflaten 2005; Oesch and Rennwald 2018; Lange 2007).
Furthermore, we would like to encourage future research to make more use of social media data. Previous research has shown that the traditional media play a crucial role in the success of the radical right (Berning, Lubbers, and Schlueter 2019; Vliegenthart, Boomgaarden, and Van Spanje 2012; Ellinas 2018). The analysis of social media data does not only help us to better understand how voters consume and react to news within and outside radical right filter bubbles but also provides us with a means to gauge how political actors present themselves.
Our analyses do not come without limitations. We focus on changes over time and draw on repeated cross-sectional data. We therefore cannot examine intra-individual variation, e.g. whether the transformation of AfD affects general ideological positions or immigration preferences of AfD supporters. Nonetheless, our time-series analyses show neither an increase in anti-migration sentiment nor a polarisation in migration attitudes. Future studies might evaluate how the AfD’s transformation relates to potential ideological shifts of the electorate in the medium- or long-run.
Our interpretations are also limited to the AfD’s short history. What the transformation means for the long-term prospects of the party is necessarily an open question. On the one hand, the AfD has now found a clearly defined niche, in which radical right parties flourish in many West European societies. On the other hand, attitudes towards immigration and immigrants are still relatively positive in Germany, and the salience of the issue has declined recently.
Moreover, the internal strife within the party has not stopped with Lucke’s or Petry’s ouster. Both have founded their own, allegedly moderate right-wing parties, though they have remained electorally irrelevant so far. Most recently, they were joined in this by Andre Poggenburg, a former regional party leader, who broke away and formed yet another splinter party because the AfD is not radical enough in his view. More generally, the ongoing conflict between the party’s openly radical wing and the more compromising groups has again come to the fore following the secret service’s highly publicised decision to closely monitor the former.
Such intense factionalism is not uncommon in the radical right, and in two prominent cases – the splits of the Front National in 1999 (Ivaldi 2003) and of the FPÖ in 2002 (Luther 2003) – the respective parties suffered massive electoral losses, from which it took them years to recover. While the AfD has benefitted in the past from its broad range of positions, personalities, and political styles, the party may well have well reached the point where the ongoing quarrels and the increasingly visible extremist tendencies within the party are putting voters off.
Alvarez, R. Michael, D. Roderick Kiewiet, and Lucas Núñez. 2018. “A Taxonomy of Protest Voting.” Annual Review of Political Science 21 (1): 135–54. doi:10.1146/annurev-polisci-050517-120425.
Art, David. 2011. Inside the Radical Right. the Development of Anti-Immigrant Parties in Western Europe. Cambridge: Cambridge University Press.
Art, David. 2018. “The Afd and the End of Containment in Germany?” German Politics and Society 36 (2, SI): 76–86. doi:10.3167/gps.2018.360205.
Arzheimer, Kai. 2009. “Protest, Neo-Liberalism or Anti-Immigrant Sentiment: What Motivates the Voters of the Extreme Right in Western Europe?” Comparative Governance and Politics 2: 173–97. doi:10.1007/s12286-008-0011-4.
Arzheimer, Kai. 2015. “The Afd: Finally a Successful Right-Wing Populist Eurosceptic Party for Germany?” West European Politics 38: 535–56. doi:10.1080/01402382.2015.1004230.
Arzheimer, Kai. 2018. “Explaining Electoral Support for the Radical Right.” In The Oxford Handbook of the Radical Right, edited by Jens Rydgren, 143–65. Oxford University Press. doi:10.1093/oxfordhb/9780190274559.013.8.
Arzheimer, Kai, and Elisabeth Carter. 2009. “Christian Religiosity and Voting for West European Radical Right Parties.” West European Politics 32 (5): 985–1011. doi:10.1080/01402380903065058.
Bale, Tim. 2008. “Turning Round the Telescope. Centre-Right Parties and Immigration and Integration Policy in Europe.” Journal of European Public Policy 15 (3): 315–30. doi:10.1080/13501760701847341.
Berbuir, Nicole, Marcel Lewandowsky, and Jasmin Siri. 2014. “The Afd and Its Sympathisers. Finally a Right-Wing Populist Movement in Germany?” German Politics 24 (2): 154–78. doi:10.1080/09644008.2014.982546.
Bergmann, Knut, Matthias Diermeier, and Judith Niehues. 2017. “Die Afd: Eine Partei Der Sich Ausgeliefert Fühlenden Durchschnittsverdiener?” Zeitschrift Für Parlamentsfragen 48 (1): 57–75. doi:10.5771/0340-1758-2017-1-57.
Berning, Carl C., and Elmar Schlueter. 2016. “The Dynamics of Radical Right-Wing Populist Party Preferences and Perceived Group Threat: A Comparative Panel Analysis of Three Competing Hypotheses in the Netherlands and Germany.” Social Science Research 55 (1): 83–93. doi:10.1016/j.ssresearch.2015.09.003.
Berning, Carl C., Marcel Lubbers, and Elmar Schlueter. 2019. “Media Attention and Radical Right-Wing Populist Party Sympathy: Longitudinal Evidence from the Netherlands.” International Journal of Public Opinion Research, 31 (1): 93-120. doi:10.1093/ijpor/edy001.
Betz, Hans-Georg. 1994. Radical Right-Wing Populism in Western Europe. Houndmills, London: Macmillan.
Bieber, Ina, Sigrid Rossteutscher, and Philipp Scherer. 2018. “The Metamorphoses of the Afd-Electorate: From a Eurosceptical Protest Party to a Right-Wing Alternative?” Politische Vierteljahresschrift 59 (3): 433–61. doi:10.1007/s11615-018-0103-y.
Brambor, Thomas, William Roberts Clark, and Matt Golder. 2006. “Understanding Interaction Models. Improving Empirical Analyses.” Political Analysis 14: 63–82.
Brug, Wouter van der, Meindert Fennema, and Jean Tillie. 2000. “Anti-Immigrant Parties in Europe: Ideological or Protest Vote?” European Journal of Political Research 37 (1): 77–102.
Canovan, Margaret. 1999. “Trust the People! Populism and the Two Faces of Democracy.” Political Studies 67: 2–16.
Ceyhan, Sara. 2016. “Konservativ Oder Doch Schon Rechtspopulistisch? Die Politischen Positionen Der Afd-Parlamentskandidaten Im Parteienvergleich.” Zeitschrift Für Politikwissenschaft 26 (1): 49–76. doi:10.1007/s41358-016-0020-9.
Coffé, Hilde. 2018. “Gender and the Radical Right.” In The Oxford Handbook of the Radical Right, edited by Jens Rydgren, 200–211. Oxford University Press.
Decker, Frank. 2008. “Germany: Right-Wing Populist Failures and Left-Wing Successes.” In Twenty-First Century Populism. the Spectre of Western European Democracy, edited by Daniele Albertazzi and Duncan McDonnell, 119–34. Houndmills: Palgrave Macmillan.
Dilling, Matthias. 2018. “Two of the Same Kind? The Rise of the Afd and Its Implications for the Cdu/Csu.” German Politics and Society 36 (1): 84–104. doi:10.3167/gps.2018.360105.
Downes, James F., and Matthew Loveless. 2018. “Centre Right and Radical Right Party Competition in Europe: Strategic Emphasis on Immigration, Anti-Incumbency, and Economic Crisis.” Electoral Studies 54: 148–58. doi:10.1016/j.electstud.2018.05.008.
Ellinas, Antonis A. 2018. “Media and the Radical Right.” In The Oxford Handbook of the Radical Right, edited by Jens Rydgren, 269–84. Oxford University Press.
Firebaugh, Glenn. 1997. Analyzing Repeated Surveys. Thousand Oaks, London, New Delhi: Sage.
Ford, Robert, and Matthew J. Goodwin. 2014a. Revolt on the Right. Explaining Support for the Radical Right in Britain. London: Routledge.
Ford, Robert, and Matthew J. Goodwin. 2014b. “Understanding Ukip. Identity, Social Change and the Left Behind.” The Political Quarterly 85 (3): 277–84. doi:10.1111/1467-923X.12099.
Ford, Robert, Matthew J. Goodwin, and David C. Cutts. 2011. “Anti-Immigrant, Politically Disaffected or Still Racist After All? Examining the Attitudinal Drivers of Extreme Right Support in Britain in the 2009 European Elections.” European Journal of Political Research 50 (3): 418–40.
Franzmann, Simon T. 2016. “Calling the Ghost of Populism: The Afd’s Strategic and Tactical Agendas Until the Ep Election 2014.” German Politics 25 (4): 457–79. doi:10.1080/09644008.2016.1201075.
Givens, Terri E. 2004. “The Radical Right Gender Gap.” Comparative Political Studies 37: 30–54.
Goerres, Achim, Dennis C. Spies, and Staffan Kumlin. 2018. “The Electoral Supporter Base of the Alternative for Germany.” Swiss Political Science Review 24 (3): 246–69. doi:10.1111/spsr.12306.
Goldstein, Harvey, William Browne, and Jon Rasbash. 2002. “Partitioning Variation in Multilevel Models.”
Goodwin, Matthew J., and Caitlin Milazzo. 2005. UKIP. Inside the Campaign to Redraw the Map of British Politics. Oxford: Oxford University Press.
Grimm, Robert. 2015. “The Rise of the German Eurosceptic Party Alternative Für Deutschland. Between Ordoliberal Critique and Popular Anxiety.” International Political Science Review 36 (3): 264–78. doi:10.1177/0192512115575384.
Hambauer, Verena, and Anja Mays. 2017. “Wer Wählt Die Afd? – Ein Vergleich Der Sozialstruktur, Politischen Einstellungen Und Einstellungen Zu Flüchtlingen Zwischen Afd-Wählerinnen Und Der Wählerinnen Der Anderen Parteien.” Zeitschrift Für Vergleichende Politikwissenschaft, online first. doi:10.1007/s12286-017-0369-2.
Hansen, Michael A., and Jonathan Olsen. 2018. “Flesh of the Same Flesh. a Study of Voters for the Alternative for Germany (Afd) in the 2017 Federal Election.” German Politics, online first. doi:10.1080/09644008.2018.1509312.
Ivaldi, Gilles. 2003. “The Fn Split: Party System Change and Electoral Perspectives.” In The French Party System, edited by Jocelyn J. A. Evans, 137–51. Manchester, New York: Manchester University Press.
Ivarsflaten, Elisabeth. 2005. “The Vulnerable Populist Right Parties. No Economic Realignment Fuelling Their Electoral Success.” European Journal of Political Research 44: 465–92. doi:10.1111/j.1475-6765.2005.00235.x.
Jankowski, Michael, Sebastian Schneider, and Markus Tepe. 2016. “Ideological Alternative? Analyzing Alternative Für Deutschland Candidates’ Ideal Points via Black Box Scaling.” Party Politics 23 (6): 704–16. doi:10.1177/1354068815625230.
King, Gary, and Langche Zeng. 2006. “The Dangers of Extreme Counterfactuals.” Political Analysis 14: 131–59. doi:10.1093/pan/mpj004.
Kitschelt, Herbert. 1995. The Radical Right in Western Europe. a Comparative Analysis. Ann Arbor: The University of Michigan Press.
Krekó, Péter, and Gregor Mayer. 2015. “Transforming Hungary – Together? An Analysis of the Fidesz-Jobbik Relationship.” In The East European Radical Right in the Political Process, edited by Michael Minkenberg, 183–205. Routledge.
Lange, Sarah L. de. 2007. “A New Winning Formula?: The Programmatic Appeal of the Radical Right.” Party Politics 13 (4): 411–35. doi:10.1177/1354068807075943.
Lees, Charles. 2018. “The ‘Alternative for Germany’: The Rise of Right-Wing Populism at the Heart of Europe.” Politics, online first. doi:10.1177/0263395718777718.
Lengfeld, Holger. 2017. “Die ‘Alternative Fur Deutschland’: Eine Partei Für Modernisierungsverlierer?” Kölner Zeitschrift Für Soziologie Und Sozialpsychologie 69 (2): 209–32. doi:10.1007/s11577-017-0446-1.
Linhart, Eric. 2017. “Politische Positionen Der Afd Auf Landesebene: Eine Analyse Auf Basis von Wahl-O-Mat-Daten.” Zeitschrift Für Parlamentsfragen 48 (1): 102–23. doi:10.5771/0340-1758-2017-1-102.
Lubbers, Marcel, Mérove Gijsberts, and Peer Scheepers. 2002. “Extreme Right-Wing Voting in Western Europe.” European Journal of Political Research 41: 345–78.
Luther, Kurt Richard. 2003. “The Self-Destruction of a Right-Wing Populist Party? The Austrian Parliamentary Election of 2002.” West European Politics 26: 136–52.
Lux, Thomas. 2018. “Die Afd Und Die Unteren Statuslagen. Eine Forschungsnotiz Zu Holger Lengfelds Studie Die ‘Alternative Für Deutschland’: Eine Partei Für Modernisierungsverlierer?” KZfSS Kölner Zeitschrift Für Soziologie Und Sozialpsychologie 70 (2): 255–73. doi:10.1007/s11577-018-0521-2.
Mudde, Cas. 2007. Populist Radical Right Parties in Europe. Cambridge: Cambridge University Press.
Oesch, Daniel, and Line Rennwald. 2018. “Electoral Competition in Europe’s New Tripolar Political Space: Class Voting for the Left, Centre-Right and Radical Right.” European Journal of Political Research 57: 783–807. doi:10.1111/1475-6765.12259.
Pardos-Prado, Sergi. 2015. “How Can Mainstream Parties Prevent Niche Party Success? Center-Right Parties and the Immigration Issue.” Journal of Politics 77 (2): 352–67. doi:10.1086/680040.
Poguntke, Thomas, and Susan E. Scarrow. 1996. “The Politics of Anti-Party Sentiment: Introduction.” European Journal of Political Research 29: 257–62.
Rosenfelder, Joel. 2017. “Die Programmatik Der Afd: Inwiefern Hat Sie Sich von Einer Primär Euroskeptischen Zu Einer Rechtspopulistischen Partei Entwickelt?” Zeitschrift Für Parlamentsfragen 48 (1): 123–40. doi:10.5771/0340-1758-2017-1-123.
Rydgren, Jens. 2008. “Immigration Sceptics, Xenophobes or Racists? Radical Right-Wing Voting in Six West European Countries.” European Journal of Political Research 47 (6): 737–65. doi:10.1111/j.1475-6765.2008.00784.x.
Rydgren, Jens. 2013. Class Politics and the Radical Right. London, New York: Routledge.
Schelter, Sebastian, Felix Biessmann, Malisa Zobel, and Nedelina Teneva. 2016. “Structural Patterns in the Rise of Germany’s New Right on Facebook.” In 2016 Ieee 16TH International Conference on Data Mining Workshops (Icdmw), edited by Carlotta Domeniconi, Francesco Gullo, Francesco Bonchi, Josep Domingo-Ferrer, Ricardo Baeza-Yates, Zhi-Hua Zhou, and Xindong Wu, 440–45. International Conference on Data Mining Workshops. Los Alamitos: IEEE Computer Society, Conference Publishing Services. doi:10.1109/ICDMW.2016.120.
Schmitt-Beck, Rüdiger. 2017. “The ‘Alternative Für Deutschland in the Electorate’: Between Single-Issue and Right-Wing Populist Party.” German Politics 26 (1): 124–48. doi:10.1080/09644008.2016.1184650.
Schreiber, Franziska. 2018. Inside Afd. Der Bericht Einer Aussteigerin. München: Europa Verlag.
Schwarzbözl, Tobias, and Matthias Fatke. 2016. “Außer Protesten Nichts Gewesen? Das Politische Potenzial Der Afd.” Politische Vierteljahresschrift 57 (2): 276–99. doi:10.5771/0032-3470-2016-2-276.
Stier, Sebastian, Lisa Posch, Arnim Bleier, and Markus Strohmaier. 2017. “When Populists Become Popular: Comparing Facebook Use by the Right-Wing Movement Pegida and German Political Parties.” Information Communication & Society 20 (9): 1365–88. doi:10.1080/1369118X.2017.1328519.
Tutic, Andreas, and Hagen von Hermanni. 2018. “Sozioökonomischer Status, Deprivation Und Die Affinität Zur Afd – Eine Forschungsnotiz.” Kölner Zeitschrift Für Soziologie Und Sozialpsychologie 70 (2): 275–94. doi:10.1007/s11577-018-0523-0.
Vliegenthart, Rens, Hajo G. Boomgaarden, and Joost Van Spanje. 2012. “Anti-Immigrant Party Support and Media Visibility: A Cross-Party, over-Time Perspective.” Journal of Elections, Public Opinion and Parties 22 (3): 315–58. doi:10.1080/17457289.2012.693933.
Figure A1: Expert assessments of the AfD’s ideological positions 2014 vs 2017 (Chapel Hill Expert Survey)
[Figure A1 about here]
Figure A2: Number and timing of usable interviews without (upper panel) and with (lower panel) multiple imputations
[Figure A2 about here]
Figure A3: Estimated overall increase in AfD support over time
[Figure A3 about here]
Figure A4: Variance and kurtosis of immigration attitudes in Germany, 2013-2017
[Figure A4 about here]
2014 Chapel Hill Expert Survey
Polk, Jonathan, Jan Rovny, Ryan Bakker, Erica Edwards, Liesbet Hooghe, Seth Jolly, Jelle Koedam, Filip Kostelka, Gary Marks, Gijs Schumacher, Marco Steenbergen, Milada Vachudova and Marko Zilovic (2017). “Explaining the salience of anti-elitism and reducing political corruption for political parties in Europe with the 2014 Chapel Hill Expert Survey data,” Research & Politics (January-March): 1-9.
2017 Chapel Hill Expert FLASH Survey
Polk, Jonathan, Jan Rovny, Ryan Bakker, Erica Edwards, Liesbet Hooghe, Seth Jolly, Jelle Koedam, Filip Kostelka, Gary Marks, Gijs Schumacher, Marco Steenbergen, Milada Vachudova and Marko Zilovic (2017). “Explaining the salience of anti-elitism and reducing political corruption for political parties in Europe with
Rattinger, Hans; Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Weßels, Bernhard; Wolf, Christof; Bieber, Ina; Scherer, Philipp (2018): Long-term Online Tracking, T20 (GLES). GESIS Data Archive, Cologne. ZA5720 Data file Version 3.0.0, doi:10.4232/1.13016
Rattinger, Hans; Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Weßels, Bernhard; Wolf, Christof; Bieber, Ina; Scherer, Philipp (2015): Long-term Online Tracking, T21 (GLES). GESIS Data Archive, Cologne. ZA5721 Data file Version 3.1.0, doi:10.4232/1.12231
Rattinger, Hans; Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Weßels, Bernhard; Wolf, Christof; Bieber, Ina; Scherer, Philipp (2015): Long-term Online Tracking, T22 (GLES). GESIS Data Archive, Cologne. ZA5722 Data file Version 3.0.0, doi:10.4232/1.12232
Rattinger, Hans; Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Weßels, Bernhard; Wolf, Christof; Bieber, Ina; Scherer, Philipp (2015): Long-term Online Tracking, T23 (GLES). GESIS Data Archive, Cologne. ZA5723 Data file Version 2.0.0, doi:10.4232/1.12421
Rattinger, Hans; Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Weßels, Bernhard; Wolf, Christof; Bieber, Ina; Scherer, Philipp (2015): Long-term Online Tracking, T24 (GLES). GESIS Data Archive, Cologne. ZA5724 Data file Version 1.2.0, doi:10.4232/1.12279
Rattinger, Hans; Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2015): Long-term Online Tracking, T25 (GLES). GESIS Data Archive, Cologne. ZA5725 Data file Version 2.1.0, doi:10.4232/1.12280
Rattinger, Hans; Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2015): Long-term Online Tracking, T26 (GLES). GESIS Data Archive, Cologne. ZA5726 Data file Version 1.1.0, doi:10.4232/1.12281
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2015): Long-term Online Tracking, T27 (GLES). GESIS Data Archive, Cologne. ZA5727 Data file Version 1.1.0, doi:10.4232/1.12282
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald ; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2016): Long-term Online Tracking, T28 (GLES). GESIS Data Archive, Cologne. ZA5728 Data file Version 2.1.0, doi:10.4232/1.12521
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald ; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2016): Long-term Online Tracking, T29 (GLES). GESIS Data Archive, Cologne. ZA5729 Data file Version 1.1.0, doi:10.4232/1.12522
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald ; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2016): Long-term Online Tracking, T30 (GLES). GESIS Data Archive, Cologne. ZA5730 Data file Version 1.2.0, doi:10.4232/1.12540
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald ; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2016): Long-term Online Tracking, T31 (GLES). GESIS Data Archive, Cologne. ZA5731 Data file Version 1.1.0, doi:10.4232/1.12655
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald ; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2016): Long-term Online Tracking, T32 (GLES). GESIS Data Archive, Cologne. ZA5732 Data file Version 1.0.0, doi:10.4232/1.12625
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald ; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2016): Long-term Online Tracking, T33 (GLES). GESIS Data Archive, Cologne. ZA5733 Data file Version 1.0.0, doi:10.4232/1.12675
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald ; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2017): Long-term Online Tracking, T34 (GLES). GESIS Data Archive, Cologne. ZA5734 Data file Version 1.0.0, doi:10.4232/1.12731
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald ; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2017): Longterm-Online-Tracking, T35 (GLES). GESIS Data Archive, Cologne. ZA6815 Data file Version 1.0.0, doi:10.4232/1.12795
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald ; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Dietz, Melanie; Scherer, Philipp (2017): Longterm-Online-Tracking, T36 (GLES). GESIS Data Archive, Cologne. ZA6816 Data file Version 1.0.0, doi:10.4232/1.12872
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald; Weßels, Bernhard; Wolf, Christof; Dietz, Melanie; Bieber, Ina; Scherer, Philipp (2017): Longterm-Online-Tracking T37 (GLES). GESIS Data Archive, Cologne. ZA6817 Data file Version 1.0.0, doi:10.4232/1.12917
Hessen 2013 (ZA5737)
Rattinger, Hans; Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Weßels, Bernhard; Wolf, Christof; Bieber, Ina; Scherer, Philipp (2015): Long-term-Online-Tracking of State Election Hesse 2013 (GLES). GESIS Data Archive, Cologne. ZA5737 Data file Version 3.0.0, doi:10.4232/1.12422
Sachsen 2014 (ZA5738)
Rattinger, Hans; Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2015): Long-term-Online-Tracking of State Election Saxony 2014 (GLES). GESIS Data Archive, Cologne. ZA5738 Data file Version 2.0.0, doi:10.4232/1.12283
Brandenburg 2014 (ZA5739)
Rattinger, Hans; Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2015): Long-term-Online-Tracking of State Election Brandenburg 2014 (GLES). GESIS Data Archive, Cologne. ZA5739 Data file Version 2.0.0, doi:10.4232/1.12284
Thüringen 2014 (ZA5740)
Rattinger, Hans; Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2015): Long-term-Online-Tracking of State Election Thuringia 2014 (GLES). GESIS Data Archive, Cologne. ZA5740 Data file Version 2.0.0, doi:10.4232/1.12285
Baden-Württemberg 2016 (ZA5741)
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2016): Longterm Online Tracking at State Election Baden-Württemberg 2016 (GLES). GESIS Data Archive, Cologne. ZA5741 Data file Version 1.0.0, doi:10.4232/1.12531
Sachsen-Anhalt 2016 (ZA5742)
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2016): Longterm Online Tracking at State Election Saxony-Anhalt 2016 (GLES). GESIS Data Archive, Cologne. ZA5742 Data file Version 1.0.0, doi:10.4232/1.12532
Rheinland-Pfalz 2016 (ZA5743)
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2016): Longterm Online Tracking at State Election Rhineland-Palatinate 2016 (GLES). GESIS Data Archive, Cologne. ZA5743 Data file Version 1.0.0, doi:10.4232/1.12533
Mecklenburg-Vorpommern 2016 (ZA5744)
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2016): Longterm Online Tracking at State Election Mecklenburg-Western Pomerania 2016 (GLES). GESIS Data Archive, Cologne. ZA5744 Data file Version 1.0.0, doi:10.4232/1.12674
Schleswig-Holstein 2017 (ZA6819)
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald ; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2017): Long-term-Online-Tracking of State Election Schleswig-Holstein 2017 (GLES). GESIS Data Archive, Cologne. ZA6819 Data file Version 1.0.0, doi:10.4232/1.12851
Nordrhein-Westfalen 2017 (ZA6820)
Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Schoen, Harald ; Weßels, Bernhard; Wolf, Christof; Henckel, Simon; Bieber, Ina; Scherer, Philipp (2017): Long-term-Online-Tracking of State Election North Rhine-Westphalia 2017 (GLES). GESIS Data Archive, Cologne. ZA6820 Data file Version 1.0.0, doi:10.4232/1.12852
1 Corresponding author
2 The list of founding members was archived at https://web.archive.org/web/20120923000256/http://www.wa2013.de:80/index.php?id=200#c563 (November 12, 2018).
3 These candidate studies also suggest that the East German chapters positioned themselves to the right of their western counterparts as early as 2014.
4 From September 2015 until at least the end of 2016, the refugee crisis dominated the political debate in Germany. With almost 900,000 new applications for asylum in 2015 and around 280,000 additional requests in 2016, the country received by far the largest number of refugees in Europe in absolute terms, although Sweden received even more on a per-capita basis. To put this in perspective, in 1992 during the Yugoslav Wars, Germany received only half as many requests as in 2015. Although numbers fell dramatically after the EU “deal” with Turkey was made in April 2016, the consequences and resulting challenges for social and economic integration are still being felt today.
5 We used the following regular expressions (ignoring case): Europe: “euro”; Greece: “athen|griech”; Islam: “moslem|muslim|islam|kopftuch|scharia|ihad”; migration: “(zu|ein)wander|migrant|asyl|fl[üue]+cht|ausländer”.
6 Because the data are noisy, we are applying smoothing splines.
7 We merged datasets for 18 separate waves from the long-term online tracking series (T20-T37) with ten additional data sets from the state election series (ZA5737-ZA6820), using the most recent versions of these data that were available at the time of writing. As each version of data set has its own DOI and full bibliographic information, we provide the relevant citations in the online appendix. The fully merged and harmonised version is made available through the first author’s dataverse.
8 While one could simply trace changes in the univariate distributions of the well-known predictors of radical right voting amongst AfD voters, modelling changes in the effects of theses variables on the likelihood of a vote for the AfD is the most straightforward and efficient way to gain analytical leverage on the transformation from a multivariate perspective. These changes will be due to both the party’s transformation and to the dramatic events of 2015/16. Moreover, it is plausible that radical right parties try to raise the salience of the immigration issue in general and may further radicalise their supporters in particular (but see (Berning and Schlueter 2016) for evidence to the contrary from the Netherlands). However, even potential bidirectional causality would not constitute a problem, because we are primarily interested in whether the AfD’s electorate is moving towards the prototypical image of radical right voters.
9 It varies from 1 per cent for religious affiliation and unemployment status to 15 per cent for voting intentions.
10 Only variables that were completely observed (gender, education, region, and calendar time) were used to predict the missing values. Although diagnostic measures suggest that 15 imputations would have sufficed, were created to provide an additional margin of safety.
11 As a robustness check, we specified two alternative structures: respondents nested within samples, and respondents nested within days nested within samples. The results were virtually identical.
12 For ease of presentation, the results from the full model haven been split across three separate tables.
13 Young, western, protestant, working class females with low levels of formal education who are not unemployed.
14 Because time is measured in days, the absolute sizes of the interaction effects are very small and estimates in Table 3 are therefore reported in scientific notation.