Dynamics and determinants of right-wing populist mobilisation in Germany

Abstract Western Europe has recently experienced increasing protest mobilisation by right-wing populist movements. Although these movements are receiving increasing scholarly attention, systematic data on protest activities is limited. In this research note, original data is used to describe the protest activity of Germany’s Pegida movement across space and time and to explore the city-level determinants of protest mobilisation. The protest dataset records 373 events with more than 337,000 participants in major cities between 2014 and 2017. The data documents the involvement of right extremists during the protests and illustrates the movement’s nativist and anti-elitist orientation. The correlational analysis of the determinants of protest activity shows that protests are less likely in cities with a large foreign-born population and lower income levels. Moreover, protests correlate with vote shares for the right-wing populist party AfD, underlining the party-movement linkage and the local entrenchment of right-wing ideology.

, the use of street protests as a strategy by right-wing populists is a development of the last twenty years (Blee and Creasap 2010;Hutter and Borbáth 2019). Despite a recent surge of scholarly interest in far-right movements and their connection to party politics (Borbáth and Gessler 2020;Caiani 2019;Castelli Gattinara and Pirro 2019;Minkenberg 2019), systematic analyses of mobilisation dynamics of the far right are limited. Recent work by Castelli Gattinara et al. (2022) offers the first large-n comparative analyses for Europe and shows the importance of political and discursive opportunities as well as symbolic and organisational resources for far-right groups. However, there is considerable unexplored variation in protest activity at the subnational level, and we know relatively little about the local determinants of right-wing populist mobilisation.
This research note contributes to the ongoing efforts to better understand the dynamics of right-wing populist mobilisation by sharing insights from an original dataset on protest activity by the Pegida movement in Germany. 2 First, we describe mobilisation patterns across time and identify two major protest waves around the Charlie Hebdo shooting and during the so-called refugee crisis in 2015/2016. As our data records details on participants and slogans, we are able to document the participation of right-wing extremists in the demonstrations and the movement's nativist and anti-elitist orientation. Second, we explore the local determinants of protest by focussing on three widely acknowledged causes of right-wing populist mobilisation: economic grievances, cultural grievances and ideology (Gidron and Hall 2020). Results from non-parametric regression analysis show that Pegida protests are more likely in cities with a lower share of foreign-born residents and higher income levels. Moreover, protests correlate with vote shares for the right-wing populist party AfD, underlining the entrenchment of right-wing ideology at the local level.
Our data allows for subnational analysis of variation in right-wing populist protests and thus contributes to the growing literature on place-based identities and populism (Adler and Ansell 2020;Cramer 2016). Moreover, the data facilitates the study of party-movement linkages since it includes information about actors and participants.

Breeding ground for right-wing populist protests
Before describing our data in more detail, we summarise some key insights from research on the breeding ground for right-wing populism. Commonly, existing explanations are related to economic, cultural and ideological grievances (Castelli Gattinara et al. 2022;Gidron and Hall 2020). While some of these factors vary over time (e.g. the influx of refugees), others are rather stable (e.g. the share of the local population that is foreign-born) and can hardly explain the exact timing of protest. Nonetheless, these factors may be relevant when explaining why a movement does (not) diffuse to a particular locality, or why certain watershed events translate into mobilisation in certain locations.
The first factor is people's feeling of being economically disconnected (Grasso and Giugni 2016). Some citizens feel disregarded by mainstream politics and society as a whole. This feeling is reinforced by mainstream political parties' focus on the middle of society, to which these people do not feel they belong. In response, people distance themselves from others they perceive as posing a threat of further economic decline. In many cases, these are newcomers, such as migrants or refugees, whom they do not perceive as part of the community or their in-group (Gidron and Hall 2020). This separation often goes hand in hand with support for movements that weave the perceived dangers of a future economic decline and newcomers into a story. President Trump successfully used such a framing of taking care of the economically aggrieved population with his ' America First' slogan.
In addition to economic grievances, cultural grievances can be part of the breeding ground for right-wing populist movements. People tend to differentiate and positively exaggerate the merits of their group in relation to other groups. Group identities may originate from different things, but a distinction that is politically easily mobilised is that between natives and migrants (Ivarsflaten 2008). However, interactions between different groups can help reduce possible prejudices, discrimination and stereotypes (Pettigrew 1998). When many groups have contact with each other, mobilisation solely based on different group identities is thus ultimately more difficult because groups may change their attitudes due to their interactions with each other over time.
A third factor is the prevalence of right-wing attitudes in the population (Hawkins et al. 2020). For right-wing populist movements, it is much easier to mobilise people with a right-wing ideology than people with a more centrist or left-wing orientation. A person's education or economic situation does not fully determine their ideology. Ideology is thus a third level of the breeding ground that partially overlaps with economic and cultural grievances but is also important in itself. Localities where many people have a right-wing ideology provide a fertile breeding ground for right-wing populist groups because populists do not have to invest much to get such people to support their cause.
A growing literature highlights the importance of space in the study of populism. One example is Cramer's (2016) study of urban/rural divides in Wisconsin and support for Donald Trump. In this article, we build on this literature and focus exclusively on the structural conditions of right-wing populists' mobilisation to protests at the subnational level. In those cities where these structural conditions are present, it should be easier for a right-wing populist movement to mobilise people to participate in demonstrations. We examine whether this is the case and which of the possible factors from these three approaches have explanatory power in the following using the example of Pegida in Germany -a movement that mobilised 'the highest turnout of a right-wing protest in Germany since the end of World War II' (Rucht 2018: 235).

The case of Pegida in Germany
This section provides a general introduction to Pegida (short for Patriotische Europäer gegen die Islamisierung des Abendlandes [Patriotic Europeans Against the Islamicisation of the Occident]). The movement originated in a private Facebook group, and the first Pegida protests occurred in Dresden on 20 October 2014. The group quickly spread to numerous other German cities, but its main base remained in Dresden, where it holds regular protests until today. In contrast, the Pegida offshoots in towns and other countries (e.g. Austria, Netherlands, Norway) stopped their protest activities soon (Berntzen and Weisskircher 2016).
While the political culture in Saxony, the state of which Dresden is the capital, paved the way for Pegida's initial success, the prevalence of xenophobic attitudes in German society enabled the movement's diffusion (Virchow 2016). The movement's original theme -Germany as a venue for violent religious conflicts -quickly expanded to encompass broad calls for limiting immigration and better funding for the police and judiciary (Vorländer et al. 2018: 12-18). Compared to other major protests in Germany, attendees at Pegida protests were younger and more male (Daphi et al. 2021). Most were driven by a general dissatisfaction with the political system and published opinion in mainstream news media, while only a minority was motivated by still very broad goals of rejecting immigrants or religiously motivated violence (Vorländer et al. 2018: 88).
One reason for the movement's attractiveness to people beyond established right-wing groups was its various references to (successful) German protests in the past and its claims to represent the 'silent majority' (Volk 2020). For example, just like demonstrations against the socialist regime in the German Democratic Republic, rallies took place on Mondays and the slogan 'We are the people' (Wir sind das Volk!) was ubiquitous. However, public support increasingly diminished as Pegida's events became more racist and aggressive in their language. An exception was Dresden, where Pegida was founded and protests still took place at the time of writing.

Original data on right-wing populist protests
In order to analyse subnational variation in the emergence and dynamics of right-wing populist street mobilisation, we collected protest event data for the Pegida movement in Germany between 2014 and 2017. Our dataset includes information on events in 89 major cities with more than 100,000 inhabitants in Germany based on media reports from German newspapers. 3 We focus on major cities to obtain a sample of comparable cases and do not cover mobilisation dynamics in rural areas. We relied on more than 80 local and regional media outlets to reduce reporting bias. 4 Outlets with nationwide coverage are unlikely to capture the local dynamics we are interested in; comparable police data across state borders are currently unavailable (see Hutter 2014). For every city, we automatically retrieved all articles we found via a keyword search that included, among other things, the city name and the term Pegida. 5 Duplicate entries were deleted from the result list based on text similarity and research assistants analysed the final set of around 20,500 news reports. 6 Our event definition comprised all observable collective activities that aimed to support the Pegida movement publicly. 7 We only recorded events with a clear geographical scope and did not record online activities, for example. Most events were street demonstrations, although some protest events took the form of torchlight processions. In addition to the event type, we recorded dates, locations, actors, violence, and slogans. Please refer to the codebook in the Online appendix for more details. The basic unit of the dataset was the event report. Since there are often multiple media reports on a single event, we retained all individual reports. We aggregated them at the event level later to keep information about uncertainty in multiple reports, for example, regarding participant numbers (Weidmann and Rød 2015).
In total, we identified 373 protest events in 30 cities with more than 337,000 participants between 2014 and 2017. 8 To validate the data, we compared our timeline of events and participants with two additional data sources. Kanol and Knoesel (2021)  Before we explore variation in protest activity across German cities to examine the breeding ground of right-wing protests, we present descriptive findings from the data. Figure 1 shows total participant numbers per month over time. The solid line shows the average of all participant estimates as recorded in the source articles, and the dashed lines represent the minimum and maximum estimates. Following the movement's emergence in Dresden in October 2014, protests against 'Überfremdung' , i.e. the feeling of being overwhelmed by immigrants, spread to many cities across the country. Participation peaked first after the Islamist terror attack against the Charlie Hebdo magazine in Paris in January 2015. After protests declined during spring, a second wave started in the fall of 2015, when a large number of asylum seekers from Syria arrived in Germany. During the so-called refugee crisis, Germany experienced the biggest influx of asylum seekers in history. Events in the city of Cologne where migrants allegedly assaulted women on New Year's Eve sparked further protests in January 2016. After the European Union struck a deal with Turkey that led to reduced migration, Pegida lost traction. Figure 2 illustrates the spatial distribution of Pegida's activities in Germany. Dresden is the city with the most recorded protest events, followed by Leipzig. Nevertheless, the map shows that the movement was not limited to Eastern Germany. Therefore, explanations that focus solely on the Socialist past of Eastern Germany fall short. Local branches that often displayed their connection to the broader movement by adapting similar names emerged in Munich (Mügida), Nuremberg (Nügida), Cologne (Kögida) and Potsdam (Pogida). However, as the spatial distribution shows, Pegida was not active in all cities. For instance, Pegida did not hold significant rallies in the northern metropolitan areas of Bremen and Hamburg.
Our dataset also provides information on participants and their demands at the events. Figure 3(a) lists the most active actors at Pegida events and underlines the movement's proximity to radical right actors. In addition to Pegida's local branches, right-wing extremist groups such as Neo-nazis and hooligans as well as extremist parties like Pro NRW, The Right (Die Rechte) or the German Freedom Party (Die Freiheit) regularly participated in demonstrations. Figure 3(b) summarises the most  common slogans at Pegida demonstrations. 9 The claim to represent the people ('We are the people') with its historical connection to the resistance movement in the former GDR is the most common slogan (see also Volk 2020). Moreover, the claims showcase the movement's nativist ('Deport!') core and its opposition to mainstream media ('Liar press!').

Contextual predictors of protests
After introducing the data, we put it to use by exploring the 'short-term contextual' (Arzheimer and Carter 2006: 424) variables that predict protest occurrence. Following the scholarly discussion of the explanatory factors of right-wing populism, we merge our data with data on economic factors (unemployment, income tax per capita), socio-demographic characteristics (migration balance, population size, average age, foreign-born population and refugees), and political factors (AfD vote share, turnout). Including the vote shares for the AfD in the federal elections in 2013 allows us to determine the prevalence of right-wing ideologies before the first Pegida protests in 2014. All data were compiled by the Bertelsmann Foundation 10 and all predictors were lagged by one year to reduce endogeneity concerns.
Given that all explanatory variables are measured annually, we aggregated the total number of protest events at the city-year level (89 cities for four years). 11 Due to missing values on some predictor variables, our final sample comprised 327 observations. Following standard practice in the quantitative study of protest events, we ran a Poisson regression model with the yearly number of events as the dependent variable. 12 To uncover potential nonlinear relationships between structural predictors and protest, we used a semi-parametric generalised linear model (Simpson 2021;Wood et al. 2016) and a shrinkage version of cubic regression splines to penalise variables with little predictive power. 13 Finally, we included random intercepts for each city, federal state and year to account for dependencies in time and space.
The plots in Figure 4 display statistically significant relationships as shown by the model. More details on the smoothed terms are summarised in Table  A1 (Online appendix). Four contextual factors emerge as significant predictors. Unsurprisingly, cities with more inhabitants are more likely to experience protests. When it comes to economic factors, wealth measured as income tax per capita was positively correlated with protest occurrence. This finding does not necessarily contradict economic grievance theory, as in comparatively more affluent cities inequality can be a salient issue (Adler and Ansell 2020), and the fear of economic decline can fuel support for right-wing populists.
The share of the foreign-born population is negatively related to protest, suggesting that mobilising citizens in those areas is more difficult compared to cities where people have less contact with migrants. This evidence is in line with the contact theory, according to which interactions between different groups help reduce possible prejudices, discrimination and stereotypes (Pettigrew 1998). Finally, our model shows a positive relationship between AfD vote share in the federal elections in 2013, as a measure of right-wing ideology, and subsequent Pegida protests. While we cannot know if the AfD supporters participate in Pegida protests, we can show that the right-wing populist vote and mobilisation cluster geographically and that both are 'two sides of the same coin' (Grabow 2016).
Other potential explanatory factors are not significant in the model. Two crucial things have to be highlighted for the interpretation of the results. First, we cannot provide causal estimates since variation in contextual factors is not exogenous. Second, a correlation at the city level, for example, between wealth and protest, does not mean that members of that group (e.g. the rich) mobilise. Still, our analysis allows for identifying important predictors, thus shedding light on the breeding ground for right-wing populist mobilisation.

Conclusion
Previous research has identified an increase in far-right mobilisation in Europe over the last two decades, with notable cross-national variation in mobilisation trajectories (Castelli Gattinara et al. 2022: 11). This research note contributes to ongoing efforts to understand far-right mobilisation by analysing variation within Germany using an original dataset on protests organised by the Pegida movement. Our data shows two major waves of Pegida mobilisation between 2014 and 2017 and demonstrates significant spatial variation in protest occurrence beyond Eastern Germany. The description of participants and their claims shows the involvement of right-wing extremists and the movement's xenophobic and anti-elitist orientation. Our exploratory analysis of contextual factors suggests that economic and cultural grievances in isolation do not explain the emergence of far-right protests. Protests are more likely in wealthier cities and less likely in cities with a high share of foreign-born citizens. Our data complement the current research agenda on party-movement linkages and populist mobilisation and help to understand the dynamics of far-right mobilisation at the subnational level.

We follow Mudde and understand populism as '[…] as a (thin) ideology
that considers society to be ultimately separated into two homogeneous and antagonistic groups, the pure people and the corrupt elite, and which argues that politics should be an expression of the volonté générale (general will) of the people […]' (Mudde 2019, 7). 2. The data is available at https://doi.org/10.17605/OSF.IO/5238F 3. A list of all cities is included in Table A4 (Online appendix). We recorded protest events for 30 cities. The data is available at https://doi. org/10.17605/OSF.IO/5238F 4. We included all regional newspapers available on the online platform Wiso-Genios (Online appendix, Table A3). 5. The exact search string was '[city name] AND pegida AND (demo* OR protest* OR *demo OR*protest)' . 6. Each assistant had to undergo a comprehensive trial coding phase to ensure intercoder reliability. 7. We also include all events directed against. Due to space constraints, we do not provide details on counter-mobilisation in this research note (see Vüllers and Hellmeier 2022). 8. Since we are aware that our data does cover all events, we believe that these are conservative estimates. We cannot know how many participants attended more than one demonstration. Our data also comprises 421 counter-demonstrations (see Vüllers and Hellmeier 2022). 9. We recorded all slogans mentioned in the source articles (average: 2.2, maximum: 7). 10. The data is available at https://www.wegweiser-kommune.de 11. See Figure A2 (Online appendix) for an overview of the distribution of events. 12. The results from quasi-Poisson and negative binomial models are shown in Table A1 (Online appendix). Model diagnostics are summarised in Figure A3 (Online appendix). 13. We obtain comparable results when running generalised linear mixed-effects models without smoothing and penalising (Online appendix, Table A2).