How Democracy Works within a Populist Party: Candidate Selection in the Alternative for Germany

Abstract

Does the demand for more direct democracy by populist parties have any implications for their internal decision-making? To answer this question, a novel large-scale research project analyses the 2017 candidate selection of all Bundestag parties, including the populist Alternative for Germany. Some 1,334 individual nominations of seven parties are compared using quantitative indicators along three dimensions of intra-party democracy (IPD): competition between aspirants for candidacy, inclusion of members and nomination-related communication. It shows that the AfD is living up to its promise of practising grassroots democracy: in all results it ranks at the top by a wide margin. A new populist organizational model seems to have emerged following neither the classic hierarchical and leader-oriented mode of many other European right-wing populist parties nor the delegate assembly mode typical of German parties. Our further development of IPD concepts, newly elaborated measuring methods and surprising empirical evidence improve the understanding of democratic decision-making in populist parties.

‘Grassroots democracy – for one another instead of against one another’, ‘Grassroots democracy instead of high finance’ or ‘Direct democracy? Full throttle, like Switzerland’. These are but a few examples of election slogans uttered by the 2013-founded Alternative for Germany (AfD) during the 2017 German federal election campaign. Obviously, direct democracy is core to the appeal of the AfD as well as other right-wing populist parties, but does its intra-party decision-making meet its own standards or does it adhere to the standard organizational model of the populist right party family: hierarchical decision-making under the control of a charismatic leader? This article aims to gain knowledge about the newer party-based right-wing populism and its organizational dimension, which ‘has been partly neglected by the literature’ (Vittori 2020: 53) so far. In our opening of the black box, the AfD challenges the widespread perception that in right-wing populist parties only reduced or no intra-party democracy (IPD) can be found (Art 2018a). We show the opposite is true: for each indicator used to measure IPD, the AfD is in the top position.

To investigate IPD, candidate selection data are used. They are understood as a decisive part of the broader IPD, where crucial preliminary decisions are made as to who the elected representatives of the parties will be (Rahat 2009; Rahat and Shapira 2017). Elmer Eric Schattschneider's (1942: 64) insight that they are ‘one of the best points at which to observe the distribution of power within the party’ has hardly lost any of its relevance. Sharing this insight, this article compares the AfD's candidate selection with the established parties in Germany in the run-up to the 2017 general election.¹ Our analysis is based on – to our knowledge – an unprecedented quantitative data collection. For almost a year, a number of researchers participated in randomly selected nomination conferences at districts and for candidate lists in the local and regional party branches of the 16 German states. They took part in these meetings from beginning to end and recorded the rules adopted, communication between the aspirants for a candidature and party members, and information about the nominations (number of aspirants and attending party members, ballots, voting results, etc.). In sum, the standardized data include information on 1,334 single candidate selections from each of the seven parties being represented in the 19th German Bundestag.

In the subsequent section, we briefly discuss the state of research on decision-making in populist parties as part of their broader organizational context and to define our understanding of IPD by differentiating between a consensus-oriented and a majoritarian-oriented model. Only through this theoretical basis is it possible to understand and classify our IPD measurements. After having introduced our research design, we identify the democratic ideals promoted within the AfD's manifestos and party rules (Hazan and Rahat 2010). Applying established indicators from recruitment research and elaborating new ones, we then operationalize and analyse IPD in candidate selection (Höhne 2013; Rahat 2009). We then discuss potential explanations for the AfD's classification as majoritarian-oriented IPD, because it seems paradoxical that this party (like other populist radical right parties in Europe) advocates a political concept that has been described as ‘illiberal democracy’, yet, at the same time, its internal procedures are highly democratic. Finally, we summarize the theoretical and conceptual value of this article and its surprising empirical findings.

Populist parties, their organization and internal decision-making

Debating IPD in a pluralistic understanding seems to be redundant when it comes to populist parties and their concept of the ‘general will’. Both within the party and for the representative democracy, the populist claim to be the sole representative of the people (Caramani 2017) makes democratic contest of diverse opinions and different interests within the political system no longer necessary. One of the core notions within populists’ ‘thin ideology’ (Mudde 2014) – the homogeneous people – denies the very necessity for pluralism, not only at the state level but also internally within the party. When they advocate democracy, it is not for liberal democracy but plebiscitary democracy. Nadia Urbinati (2015: 480) introduced the paradoxical term ‘direct representative democracy’, meaning a multifaceted tendency in contemporary Western democracies to strengthen direct connections between the people and their elected politicians. This transformation is accompanied by populist parties that typically demand direct democracy (Best 2020). That in turn is an element in the ideologically charged campaign to set themselves apart from the intermediation of the established political parties.

What is known about the decision-making as part of the organization of populist parties? They ‘rely as little as possible on complex party organizations as intermediaries between citizens and politicians’ (Müller 2016: 36). But vague terminologies such as ‘outsider formations’ (Heinisch and Mazzoleni 2016a: 238) do not allow any conclusions to be drawn for the nature of their internal organization. Unfortunately, frequently cited studies do not, in large part, examine the decision-making in populist parties to any great extent (e.g. Betz 1994; Moffitt 2016; Mudde 2007; Müller 2016; Scarrow et al. 2017; Zaslove and Wolinetz 2018; Zulianello 2020). An exception is the anthology Understanding Populist Party Organization (Heinisch and Mazzoleni 2016b), which, however, lacks a chapter on Germany. Only recently, a ‘literature gap’ (Vittori 2020) was identified for the organization of populist parties. Nevertheless, there are two models, one established and a second, more recent one, that reduce the complexity of populist organizations, simplify their variance and help us to understand them.

The older model corresponds to the classic understanding of a right-wing party, often with an entrepreneurial founding context (Betz 1994; Minkenberg 1998). It covers older populist parties with a weak but centralized and elite-oriented organization, which was identified in the literature as ‘the main organization trait of populist parties’ (Art 2018a; Vittori 2020: 54). The inherent conflict between authoritarian tendencies of populist parties on the one hand and populist ideas such as popular empowerment on the other (Mudde 2007) manifested itself in this organizational model. According to this ‘standard model’ of party-based populism, internal decision-making is dominated by strong charismatic leadership, which is considered necessary to maintain the party's coherence (Heinisch and Mazzoleni 2016c). The concentration of power ‘is frequently accompanied by formal or informal mechanisms designed to restrain intra-party democracy’ (Heinisch and Mazzoleni 2016a: 227). This also limits the intra-party influence of a possible ‘movement component’ (Ellinas 2020: 50). An extreme example of this classical model is the Dutch Freedom Party, which presents itself as the authoritarian one-man-show of a charismatic leader (Otjes and Louwerse 2013). Cases with a more developed organization are the French Front National or the Italian Lega Nord (Heinisch and Mazzoleni 2016a).

Populism can be inherent in a contemporary democratic system (Pasquino 2008); conversely, democracy can also be part of populism. The more recent organizational model comprises populist parties that present themselves as populist grassroots movements, which are not necessarily right-wing organizations (Ruzza 2017). They can be marked by a tendency towards direct and bottom-up participation without intermediary channels by a complex party organization (Vittori 2020). Their claim to be a movement corresponds to the image of a network that lacks party affiliation and blurs the boundaries between members and supporters. Examples are the Italian 5 Star Movement (Mosca and Tronconi 2019) or Podemos in Spain (Kioupkiolis and Pérez 2018). Both use digital tools for the orchestrated participation of their members in the nomination of candidates as well as for adopting their manifestos (Lanzone and Rombi 2018). However, for both, an asymmetrical balance of power in favour of the party leaders has also been demonstrated (Caiani et al. 2021), which in turn refers to a central element of the older populist parties.

This raises an exciting question: what kind of IPD is practised in the younger populist parties? To answer this question, IPD, as a well-established concept in party research, is used (Cross and Katz 2013; Ignazi 2020; Rahat and Shapira 2017; Teorell 1999). Benjamin von dem Berge and Thomas Poguntke (2017) distinguish between a ‘plebiscitary’ and an ‘assembly-based’ ideal type of IPD. Their starting point is the inclusion of members in processes of IPD, but a broader involvement of party members does not necessarily mean much for the quality of IPD if no choice between at least two aspirants can be made. Furthermore, as complex organizations, parties do not function according to plebiscite logic (Cross 2018), even though party primaries are an increasingly important element of their decision-making (Kenig et al. 2015a; Scarrow 1999) and digital parties with new and immediate forms of participation channels are emerging (Gerbaudo 2018).

Following von dem Berge and Poguntke (2017), slightly different terms with partly modified meanings for both IPD types are proposed. Analogous to the two ideal types of representative democracy suggested by Lijphart (1984), which focus on the dominant decision-making mode of intra-state politics, the first type is referred to as ‘consensus-oriented IPD’ and the second one as ‘majoritarian-oriented IPD’. The intense direct participation of all members, especially the grassroots, reflects the ideal of the latter type. Decisions are usually made at general meetings or by ballots with a final majority vote. In the first IPD variant, it would be the consensus-oriented negotiations of party functionaries. It favours delegate assembly without overt confrontations. While this one is based more on balancing party-specific power issues and intra-party representation demands, the majoritarian-oriented IPD is connected to the mobilization power of party groups.

If one understands the parties’ decision-making as a chronological multi-stage process as it is suggested for candidate selection (Hazan and Rahat 2010; Höhne 2013), then the preliminary stage is crucial for the consensus-oriented IPD and the final stage for the majoritarian-oriented IPD. The preliminary stage is characterized by a more informal and unregulated decision-making process, while at the final stage decisions at an official nomination conference are made based on binding participation rights of party members. The final stage is visible and subject to greater numbers of rules for appropriate conduct. However, informal norms that apply in the earlier stage can provide information on who makes the first ‘cut’. Negotiations by influential party representatives can lead to a result in which only one aspirant remains at the end. This is called an ‘appointment system’ (Hazan and Rahat 2010: 72–86). Therefore, a steering agent was proposed as the decisive actor for the preselection of party leaders (Aylott and Bolin 2017) – its functional logic is the same for candidate selections (Kenig et al. 2015b).

Bringing together the considerations on our two IPD types and the more or less distinct populist organizational models, we base our analysis on the following assumptions: first, as a younger populist right party compared to its cognate European counterparts, the AfD claims to be a participation-oriented populist organization which relates to its demand for more direct democracy in general. Second, its proclaimed IPD matches the configuration of the parties' actual candidate selection processes. It will be assigned to the majoritarian-oriented IPD as it best fits the normative understanding of democracy of a populist party, especially its anti-pluralism.

Research design

German parties are not entirely free to choose their internal organization. According to the German constitution and party law, parties are obliged to organize their decision-making democratically. Within these boundaries, though, it is possible to implement a varied spectrum between the ideal IPD types that enables a fruitful analysis of the AfD and its established counterparts. Thus, the research design is a cross-sectional intra-state comparison that best demonstrates what is unique about the AfD's internal democracy.

To answer the question of how the AfD performs with regard to IPD, we analyse various aspects of its candidate nomination processes and compare the results with the other Bundestag parties. At the beginning of the analysis AfD's self-set specifications of IPD in its manifestos and party rules are determined. IPD in candidate selection is operationalized by: (1) the competition among aspirants; (2) the inclusion of party members; and (3) the communication at nomination assemblies.² While the inclusion is well established in candidate selection research (e.g. Schindler 2020), indicators are introduced below to measure competition and communication. Depending on the results the indicators lead to, the parties’ decision-making will be assigned to one of the two suggested ideal IPD types. The more aspirants and selecting members are involved and the more intense the communication between them, the more likely is the majoritarian-oriented IPD. Vice versa, low values of these three variables indicate the presence of the consensus-oriented IPD. Although this study focuses on the final stage of candidate selection, conclusions thereof will be drawn for the preliminary stage.

The aspirants’ competition for a nomination is considered one of the most meaningful indicators of IPD (Katz and Cross 2013; Rahat et al. 2008). It shows directly whether the members present – the only actors formally entitled to nominate candidates – had a real choice between alternatives. Indirectly, it indicates the (un)importance of the final decision stage. If, in turn, this stage is found to be relevant in terms of competition between aspirants, this would indicate the presence of majoritarian-oriented IPD. Conversely, a lower relevance of this stage would mean that IPD most likely follows the consensus-oriented mode. An absence of visible competition does not allow the assumption that there was no competition at all. The conclusion can only be that the final stage had no influence and majoritarian IPD was lacking.

Whether a more inclusive intra-party decision-making system is a more democratic one likewise depends on the understanding of IPD (Hazan and Rahat 2010; Kenig et al. 2015b; Rahat 2009). Its majoritarian variant strives to realize the basic democratic claim of equal participation. Accordingly, a delegate conference can be seen as a restriction on participation, whereas a general meeting allows all enfranchised party members to join. In contrast, the consensus-oriented IPD opposes the concept that sheer inclusion rates cannot assess the quality of IPD. Apart from its more functionary-driven decision-making logic, the admission of all members does not necessarily lead to broader inclusion in any case. Under the circumstances of declining member numbers in the context of party alienation from society (Borz and Janda 2020; Gauja 2017), a delegate conference with its more substantial participation commitment may show a higher inclusion rate than would be the case for a general meeting in the same party branch. In order to measure how inclusive or exclusive a candidate selection is, the relationship between a defined population base and the people who are either allowed to participate or actually participate was introduced (Hazan and Rahat 2010). Hence, ‘formal inclusion’ can be distinguished from ‘effective’ inclusion. In this study, actually participating members at candidate selection are set in relation to the members of the respective party branch.

Communication is rarely developed as an indicator of candidate selection research. Although it is questionable whether a short introductory speech or a limited Q&A round at a nomination conference offer sufficient information about an aspirant's qualification to become MP, they do at least help the selecting members, especially in very inclusive meetings, to get an impression of who the aspirants are, what they want politically and how they differ from each other in terms of certain visible or self-declared characteristics or qualifications. However, in the final stage of a decision-making process, there may be no need for communication. The requirement of a dialogue between aspirants and members depends on the applied IPD mode. Its majoritarian type presupposes much more interaction in the final stage compared to the consensus-oriented one. Due to their deeper involvement in party affairs – for example as functionaries, MPs or party board members – delegates can be considered better informed about the aspirants than rank-and-file members.

Our study is part of the large-scale project #BuKa2017 of the Institute for Parliamentary Research (IParl) to analyse German candidate selection, especially the demand of party members for a candidacy, the supply of candidacies with different chances to win a seat, the selectors and their role as gate-keepers. All parties elected at the general election in 2017 were part of the project. Between September 2016 and July 2017, labour-intensive field research took place at nomination conferences randomly selected by a polling institute. The parties had to be asked to permit our attending observation, which was particularly difficult and time-consuming with regard to the AfD due to its aversion to academic research, or rather the distrust and secrecy of populist parties in general (Mudde 2007). As the AfD won 97% of its parliamentary mandates via the proportional vote (91 of 94 MPs), in this analysis, the main focus is on the state party level.³ There, 48 of 96 list conferences of altogether seven parties with a total of 1,246 list places were examined. Additionally, 88 district races were directly observed. For details on the case selection, data collection methods, their validation, operationalization of the variables applied in this analysis and summary statistics see the Online Appendix.

Ambition and reality of the AfD's decision-making

High standards for IPD

The AfD presents itself as a self-proclaimed alternative to the ‘old parties’ (Altparteien), ‘cartel parties’ (Kartellparteien) or ‘system parties’ (Systemparteien) as it disparages its competitors. It also opposes professional politicians. Its election manifesto states, ‘The secret sovereign in Germany is a small, powerful political oligarchy which has developed within the existing political parties’ (AfD 2017: 8). Similarly, the party platform notes, ‘Behind the scenes, a small and powerful elite within the political parties is secretly in charge, and is responsible for the misguided development of past decades. It is this political class of career politicians whose foremost interest is to retain their own power base, status, and material well-being’ (AfD 2016: 7).

Unlike other right-wing populist parties, however, the AfD's outwardly directed demands meet with an internal response. From the outset, it has called for more direct democracy within parties. On its homepage, it presents itself as a participatory organization offering its members more rights than any other competitor; its spokespersons (this alternative term for chair was borrowed from the Greens) call their organization a grassroots movement (Koschmieder 2017).

Although the AfD's statute prescribes a delegate assembly for the federal level, a general meeting can take place as well (Bundessatzung der AfD 2015: §11). In fact, some national assemblies have been held as general meetings so far, but this becomes more and more difficult as the number of members increases.⁴ This distinguishes it from the other parties, which only held exclusive assemblies of delegates there. All members can participate in a consultative member survey (only electronically; this was conducted for the two top candidates for the 2021 federal election) or a membership ballot (as a party primary or postal vote) (Bundessatzung der AfD 2015: §20). Quotas for lists or party boards are denied because they would restrict the voting rights of the members (Bundessatzung der AfD 2015: §5). Its candidates for parliaments should prove that they have at least five years of professional experience outside politics (Bundessatzung der AfD 2015: §19), which corresponds to its general demand for the incompatibility of legislative and executive mandates and restrictions on terms of office.

Competition of aspirants

Comparing the results for the aspirants’ competition for a nomination, AfD candidate selection processes were by far the most competitive. First, this was measured with the rank competition index (RCI), which provides an aggregated result of the contests for a list of candidates within a party (Höhne 2013: 184–185). The index has a range of 0 to 1, with high values indicating competition for upper list positions or more competition in general. Competition was defined by more than one candidacy per list place. The list place values were assigned in descending order. For instance, a list with 10 places received a value of 10 for the first place and one for the last one. The values of the contested places were summed up and divided by the list place value of the whole list. While the AfD's value is at 0.8 (out of a maximum of 1), the Social Democrats (SPD) and Christian Democrats (CDU and CSU) (almost) approach 0, which shows that there was no competition at all (Table 1). The Left Party and the Greens with an RCI of 0.5 cannot yet be assigned to any IPD type, and the Liberals with a value of 0.3 tend towards consensus-oriented IPD.

Table 1. Rank Competition Index (RCI)

Note: RCI = ∑ rank value of the contested (more than one candidacy) list places/∑ rank value of all list places of a state party branch. A party's RCI is the mean value of the single RCIs.

Second, apart from a small number of senior positions, almost every list place was contested in the AfD. On average, five aspirants competed for a nomination (Figure 1). The Greens and the Left Party fielded 2.8 or 2.4 aspirants, respectively, while the remaining four parties had only about one. The bubbles in Figure 1 display the frequency of aspirant numbers per list place (ALP) in % of the respective party. The frequencies are adjusted per party to eliminate biases due to differing sample sizes. They indicate the competition's intensity and dispersal on the lists. The AfD's bubble sizes are almost equally distributed. This shows, for example, that three aspirants demanding a list place is as common as 11 aspirants. The lack of a discernible cluster effect marks a sharp contrast to the typical recruitment pattern in Germany, where challenged candidacies (Kampfkandidaturen) are very rare (Höhne 2017). In parties operating according to the consensus-oriented IPD, competition is usually circumvented by negotiations at the preliminary stage of the candidate selection process. This is illustrated by the larger bubbles at the bottom of the CSU, CDU, SPD, Liberals and, to a lesser extent, the Greens and the Left Party.

Figure 1. Aspirants Per List Place (ALP)

Third, to detect close races and better deal with novice candidates who have unlikely prospects, which are more common in new parties like the AfD, the list candidates’ initial support (CIS) was measured. CIS is another indicator to determine the intensity of competition at candidate selection. It sums the share of votes of the (finally nominated) candidates in the first ballot and divides it by the number of list places. The share of votes is based on the number of all votes cast, including invalid votes and abstentions. The analysed number of list places had to be reduced from 1,246 to 1,078 because the exact voting results were not available or recorded for 168 lower places. These were almost exclusively at the bottom of the lists, with negligible impact, so they can be excluded from analysis without affecting the results. CIS shows for the AfD significantly lower percentages than for the other parties (Figure 2). The bubble sizes display the relative frequency distribution of candidates’ initial support. The larger the bubble, the more candidates received the same initial support. The AfD's mean value of 45.3% is not only indicating the frequent presence of several promising aspirants. Rather, AfD candidates received less than 50% of the votes on average, while for the other six parties, most aspirants cleared the 50% hurdle for nomination in their first attempt. This dividing line is important because a lack of absolute majority usually makes a second ballot necessary – which was the standard procedure only in the AfD.

Figure 2. Candidates’ Initial Support (CIS)

Inclusion of party members

In Germany, the electoral law only allows for either a (formally) more inclusive general meeting or a (formally) more exclusive delegate assembly (Bundeswahlgesetz 1993: §21). The former means that any member who meets the few legal requirements, such as the minimum age of 18 years, can participate. In the latter case, only elected party members can join. The effective inclusion, which can be used as a reliable indicator for intra-party participation, ranges at the district level between 4% to around 30% (Table 2). The highest is achieved by the AfD, whose candidates are ultimately nominated at a general meeting. It is followed by the Liberals, the Left Party and the Greens. These smaller parties with regard to membership numbers hold general meetings, too, nearly without exception. Among the larger parties, the CSU proposes candidates exclusively, the SPD predominantly by delegates and the CDU increasingly at general meetings (Hellmann and Höhne 2020). Thus, it is less surprising that the AfD has the top position and the CSU the last, but rather that the AfD's inclusion rate is almost double that of the Greens with about 16%, although this ‘participation party’ also exclusively organizes general meetings.

Table 2. Effective Inclusion of Party Members

Note: In %; in parentheses = N randomly selected cases; effective inclusion = ∑ attendant members/∑ all members; own member numbers as of 31 December 2016, based on information from the central offices of the parties.

At the state level, the inclusion of members is significantly lower in all parties (Table 2). This can be explained by the dominance of the delegate assemblies. Especially in parties with a large member base, assemblies are usually the only viable option due to organizational and financial constraints (Höhne 2017). Only the smaller parties in the city states of Berlin, Bremen and Hamburg invite their members to general meetings, but not always. Here, too, the AfD reverses the picture: in 14 of 16 state-level associations, a general meeting was held. Of all parties examined, the AfD demonstrates the highest member inclusion, with around 23% on average. In contrast, the Left Party manages only 5%, the CDU and SPD 2%, and the CSU just over 0%. The AfD's general meetings at state level reach 26%, above the Greens (19%) and the Liberals (18%). However, it must be taken into account that it has fewer members overall (24,868 at the end of 2016), which proves to give it an advantage over parties with more members when organizing a general meeting.⁵

Nomination-related communication

The electoral law only prescribes an adequate amount of time for self-introduction (Bundeswahlgesetz 1993: §21). The nomination conference decides on the actual length of speeches, the number and selection of questions. Indeed, the intensity of aspirant-member communication differs sharply between the parties. If we compile all communication actions at list meetings – that is, all questions raised and statements made by the members selecting the candidates – the AfD again takes the top position (Table 3): on average, roughly three contributions were addressed to one of its aspirants. All other parties are below this level. The picture is even clearer when it comes to communication for a list place: the mean value for the AfD is almost eight contributions compared to 3.4 for the Left Party, rounded up to one for the Greens and nearly zero for the remaining parties.

Table 3. Nomination-Related Communication at List Conferences

Note: Contributions = questions, statements and supportive speeches in Q&A rounds.

The AfD's conferences were far more protracted than those of the other parties. At the state level, they extended not only into the evening but often over several days, or even consecutive weekends. On average, the AfD needed 28 hours to fill a list, while the CSU took only an hour and a half (Table 4). Diverging from the other parties, the AfD's conferences started with lengthy deliberations on the appropriate rules for its nominations. These included the voting procedure, the length of time that aspirants had to introduce themselves, the maximum number of questions addressed to them and the admission of journalists. At the actual nominations, much more time was spent on communication between the aspirants and the members, as well as on voting and counting the votes.

Table 4. Time Taken to Fill a List

Note: Arithmetic mean in hh:mm (without breaks); in parentheses = N randomly selected nomination conferences/all cases.

All indicators measuring intra-party democracy in candidate selection show the highest values for the AfD compared to the other parties in the Bundestag (Figure 3). In line with the significant competition among members for nomination, inclusive decision-making and intense communication at nomination conferences, the AfD's examined IPD is strongly rooted in the majoritarian-oriented type, while the CDU, CSU and SPD are clearly to be found in the consensus-oriented one. Between these two poles are the Greens, the Left Party and the Liberals, which are less clear-cut, oscillating between the two IPD poles. For them, the data show different accentuations, although they leaned towards the consensus-oriented IPD in general. While the Greens showed comparatively moderate competition of aspirants, the Left Party's IPD is characterized by a great willingness to engage in discourses, and the Liberals distinguished themselves by a respectable mobilization of members to nominate their district candidates.

Figure 3. Summary of the AfD's IPD Compared to the Other German Parties

Discussion

In this study, we argue that the AfD's revealed majoritarian-oriented IPD fulfils several functions: regarding its organization, it is a mechanism to deal constructively with its members’ dissatisfaction with wider representative democracy (Höhne 2018). As current research has shown, IPD can keep idealistic grassroots members motivated (Koo 2020). The AfD has been associated with movements like the xenophobic and nationalistic Pegida (Arzheimer and Berning 2019) or, more recently, with demonstrations against the freedom restrictions implemented to fight the COVID-19 pandemic. Hence, activists and voters with populist attitudes (Voogd and Dassonneville 2020) may feel attracted to a party that positions itself as more democratic compared to other parties with supposed oligarchies of professionalized politicians. Regarding party competition, the AfD's IPD is an integral part of its campaign to differentiate itself from its competitors; it is seen as a ‘lever to challenge established parties’ (Von dem Berge and Poguntke 2017: 137). The calculus behind this is not to force its IPD model on others, thereby sacrificing its organizational unique selling point, but to expose them, knowing full well that they would hardly turn away from the established routines of the consensus-oriented IPD. But to what extent does the novelty of the AfD account for the empirical findings, and how likely is it that its IPD will hold? Two further explanations should be considered: on the one hand the party's internal ideological pluralism and on the other its level of political experience.

The party has changed its ideological course several times, at first from a single-issue party against European fiscal union to an ethno-nationalistic one in the context of the European refugee crisis (Arzheimer 2015; Berbuir et al. 2015; Schmitt-Beck 2017). After the conservative market-liberal and Eurosceptic wing – formerly dominating the party – was considerably weakened, most of its followers left the party (Franzmann 2019). Second, since the European elections and the elections in three eastern German states in 2014, the party's orientation has shifted further to the right. Since then, the conflict line has been how to deal with the far right, both internally and externally. Frauke Petry, co-leader between 2015 and 2017, failed to bring the party back to a more moderate course (Art 2018b). The struggles of the party wings have led to profound changes in the AfD's manifestos, accompanied by tangible shifts in membership. But – and this is the crucial argument here – the majoritarian-oriented IPD was stable over time. Conflicts about the party's strategic course will probably remain commonplace, especially between the party branches in the east and west. Accordingly, organizational conditions that would allow a shift towards the consensus-oriented IPD are lacking.

As far as political experience is concerned, this can again be seen both in retrospect and in the future. The AfD is a rather young party, but its candidate selection process researched here was by no means its first. In 2013, it narrowly missed the 5% hurdle to enter the Bundestag, with 4.7% of the second votes. However, the party gained access to state funding. Before the general election in 2017, it had already completed 14 candidate selections in state-level associations and one nationwide one for the 2014 European elections. The resulting parliamentary mandates meant a dramatic improvement in access to human and financial resources (Schroeder et al. 2018) as, in general, reaching the threshold of representation is an essential step for a party's institutionalization (Arter and Kestilä-Kekkonen 2014; Bolleyer and Bytzek 2016). Accordingly, the AfD was already experienced when selecting its candidates for the 2017 election.

Viewed prospectively, the question arises whether growing experience could lead to the shedding of numerous alternative organizational elements hitherto embraced, as was observed for the early Green/New Left parties. In any case, incumbency effects of MPs will be more frequent in the AfD's future candidate selections. Nevertheless, two (further) arguments speak against a fundamental change in its IPD. First, being internally democratic is much more central to the AfD's self-image than for the Greens, which started as an ‘anti-party party’ whose main focus was not only on anti-establishment policies but in particular on ecology and post-materialist policies. Second, there is no evidence that the AfD's gain in political experience so far had a mitigating effect on its IPD. Studies examining the early AfD suggest that the examined candidate selection was not significantly different from that for the 2013 general election (Koschmieder 2015 emphasizes a very competitive situation in 2013; see also Reiser 2018). Accordingly, the AfD was described as having a ‘base of confident militants and a culture of internal debate’ (Lewandowsky 2014: 4). Its founder Bernd Lucke failed to strengthen his position as (sole) party spokesman and resigned in July 2015 (Jäger 2019). To date, there is a lack of a strong leadership personality and the party is still led by a team leadership of two speakers, whose existence is attributed to the integration of the competing party wings as well as the will to share power (Lees 2018). Finally, it is worth noting that despite COVID-19 and the fact that delegate assemblies would offer better virus control, the AfD sticks to inclusive general meetings for its candidate selection to the 2021 federal election. Again, this is not unique to the constituencies, but very much so to the state level. There, only three delegate assemblies but 12 general meetings and one postal vote of all members in Baden-Wuerttemberg were held.⁶

Conclusion

Starting from the further development of a theoretical IPD concept (Von dem Berge and Poguntke 2017), an empirical research programme for the comparative analysis of parties was launched. The suggested consensus-oriented IPD was contrasted with the majoritarian-oriented one. For the application of these IPD models, established indicators for measuring candidate selection were used and new indicators introduced. On this basis, the decision-making of the AfD was precisely measured and compared to the other Bundestag parties. The novel large-scale data set, which is the result of strenuous field research, revealed robust empirical insights. They form an indispensable piece of the puzzle for the scholarly debate on the organization of the newer right-wing populism.

Principles of internal democracy are promoted in the AfD's manifestos as well as by its rules. This commitment corresponds very well with the party's internal ways of working: most of the candidate selections for the general election in 2017 were fiercely contested, both for the ‘promising’ and the ‘hopeless’ positions. Almost every nomination conference was based on the membership principle, leading to a high effective inclusion rate. Intensive aspirant-member communication, bringing transparency to the decision-making criteria, was an integral part of its candidate selection. Nearly all nominations, especially for the lists of the state-level associations, were very time-consuming. Thus, the intra-state comparison of the 2017 candidate selection of the seven current Bundestag parties brought to light the highest IPD level for the AfD, assigned to its majoritarian variant.

In keeping its promise to practise grassroots democracy, the AfD stands for a hitherto unknown decision-making model for a right-wing populist party. The empirical findings sharply revise the widespread understanding of older right-wing parties as centralized organizations with limited internal democracy (e.g. Heinisch and Mazzoleni 2016a). At the same time, the AfD's IPD partly resembles the younger populist organizational model, in which extensive participation can be found (e.g. Vittori 2020), but unlike the 5 Star Movement or Podemos, there was no evidence of party-internal plebiscites or any elements of central control (Caiani et al. 2021). Explanations for the AfD's majoritarian-oriented IPD have been discussed here, emphasizing its party-internal (to mobilize members and supporters) and party-external functions (to differentiate itself from competing parties) which underline its anti-elitism and anti-pluralism. In the absence of longitudinal data, an indication of durability is that many aspects of the party's life (of over eight years) have changed in the recent past, but not its mode of democratic decision-making.

To determine how durable the AfD's deviant case of the standard populist organizational model will be in the future and how it relates to other younger populist organizations, further research on its IPD is necessary. Longitudinal and cross-sectional approaches are needed in order to draw a more generalizable and more detailed picture of (new) right-wing populist parties and their IPD. Our theoretical IPD concept and applied indicators in candidate selection can be used to detect IPD for a comparison of populist parties – for example, left- and right-wing ones (March 2017), oppositional and governmental ones (Albertazzi and McDonnell 2015) or depending on their geographical place of activity (Manow 2018) – and to classify long-term changes within a party, such as in response to institutionalization (Bolleyer and Bytzek 2016). Studies could analyse the IPD in several manifestations as a dependent and an independent variable of, for example, a party's organization, internal and external representation, as well as its electoral success.