Intra-party decision-making in contemporary Europe: improving representation or ruling with empty shells?

ABSTRACT Political observers agree that parties in European parliamentary democracies are more likely than previously to give party members opportunities to vote in decisions about party policies or personnel. Observers are less agreed about the implications of these apparent procedural trends. Some, including Peter Mair, saw them as evidence of the hollowing-out of party democracies; others have seen them as enhancing citizens’ opportunities for meaningful political participation. Because this is ultimately an empirical question as well as a normative one, these radically conflicting interpretations make it crucial to examine which interpretation is best supported by usage to date This is the task we undertake in this article. We use data from the Political Party Database Project (PPDB) to investigate the extent to which parties in 26 European countries have adopted and employed intra-party ballots. We also ask whether there is evidence that such procedures are changing intra-party relationships. We find that balloting of party members is indeed widely used, but it is by no means universal. We find much less support for the implication that such ballots are associated with less competitive contests, or that the new devices are generally used in ways that devalue party-member bonds.


Introduction
In Ruling the Void, Peter Mair warned against the hollowing out of political parties. He argued that there was a trend towards parties becoming leaderdominated organizations, accompanied by professionalization (and outsourcing) of party communications and the atrophying of parties' grassroots linkages. In his words, '[T]he result is a new form of democracy, one in which the citizens stay home while the parties get on with governing' (2013, p. 98). This indictment is arguably an extension of Kirchheimer's catch-all party predictions, which foresaw parties moving away from cleavage-based representation to form broader and more opportunistic voting coalitions. Yet Mair's conclusions about the consequences of the perceived changes were considerably more pessimistic than Kirchheimer's.
Mair was not alone in diagnosing a politically consequential trend towards the weakening of parties' organizational linkage mechanisms. Some scholars have connected these changes in parties' grassroots mobilizing approaches with an overall personalization (McAllister, 2007) or even presidentialization of party leadership (Poguntke & Webb, 2005Webb, Poguntke, & Kolodny, 2012). According to these diagnoses, the more that the party leader becomes the party message, the less room there is for party membership organizations to play a role in setting party policies. In leader-centred parties, party organizations serve as fan clubs, not as two-way transmission belts linking the party with its core supporters (McAllister, 2007). Party leaders may bolster their personal mandate through parties' use of new leadership selection mechanisms, ones which give individual party members a direct say in their selection (Passarelli, 2015, p. 15;Poguntke & Webb, 2005). Other authors have made similar arguments about the malign effects brought about by empowering individual members while potentially undermining the accountability and participation opportunities provided by networks of local party organizations (Accetti & Wolkenstein, 2017;Ignazi, 2020;Mair, 1994, p. 16). Mair and his co-authors as well as others linked these trends with the apparent sharp drops in party membership from the 1980s onwards, trends that were particularly evident in the once membership-rich parties of Western Europe (Mair & van Biezen, 2001;van Biezen & Poguntke, 2014;van Biezen, Mair, & Poguntke, 2012). All these authors diagnose a hollowing out of parties' traditional structures that both reflects, and further stimulates, increasing electoral volatility. They foresee the declining ability of any partiesincluding new onesto capture and consolidate partisan support in ways that endure across the ups and downs of electoral cycles.
Yet not all observers have been uniformly pessimistic about all aspects of the diagnosed trends. Most importantly, those who study one of the most striking changesgiving party members a direct say in selecting party leaders and making other key decisionsroutinely use the term 'intraparty democracy' to describe these membership ballots (Hazan & Rahat, 2010). Whether or not this usage implies that alternative decision-making procedures are less democratic than direct balloting of members (see Cross & Katz, 2013 for a refutation of this conclusion), it certainly implies that using such ballots is a legitimate alternative for democratic parties. Many would go further, and argue that inclusive (non-elite) decision-making is a virtue. Some also argue against the notion that party ballots necessarily undermine the traditional mediating organizations of party structures, pointing out that parties could use participation rights as a membership incentive, and could thus rebuild their grassroots organizations by offering members a more direct say in party decisions (Faucher, 2015;Gauja & van Haute, 2015;Scarrow, 2015).
These widely differing interpretations are fundamentally empirical questions, at least as much as normative ones. After more than two decades in which European parties have made increased use of membership ballots it is now possible to draw more informed conclusions about whether the individualization of party-member links through direct balloting of members is associated with a (paradoxical) marginalization of party members or an increasing dominance of party leaders. Examining this evidence is the task we undertake in this paper, in which we investigate how political parties in contemporary European democracies seek to use inclusive decision mechanisms to build and sustain partisan support coalitions. Using data from the Political Party Database Project (PPDB), we investigate the extent to which parties in 26 European countries have adopted and employed intra-party ballots to settle important decisions, including selecting party leaders and legislative candidates, writing manifestos, and resolving internal policy debates. We also ask whether there is evidence that such procedures might be changing relationships between party leaders and party members, or between members and other party supporters. We find that balloting of party members is indeed a widely used option, though it is by no means the dominant one. However, we find little support for the argument that such ballots are associated with less competitive contests at the leadership level. We also find little support for the notion that they are generally used in ways that devalue party-member bonds.

The individualization of membership linkage and intra-party power dynamics
For at least a half century, political scientists have been diagnosing ongoing and seemingly consequential changes in how political parties in European democracies use their organizations to support their electoral coalitions. Thus, in the 1960s, both Leon Epstein (1967) and Otto Kirchheimer (1966) detected a shift away from organizational reliance on grassroots structures and activists, towards more centralized and professional party organizations. They saw this change in campaign styles as likely to shift intra-party power balances in favour of party leaders. In the 1970s and 1980s, a slew of studies on 'party decline' seemed to confirm this shift away from membership-based party organization. According to some observers, this decline in partisan participation and allegiance stemmed not only from shifts in party strategies, but also from changes in citizens' participatory preferences (Barnes & Kaase, 1979;Dalton, 1988;Inglehart, 1977). This literature, along with the erosion of the social cleavages which had once structured much of European politics, the electoral losses experienced by some long-dominant parties in the 1970s and 1980s, and the rise of green (and largely later) populist challengers, raised the question of how established parties would respond to the new problems of interest aggregation that these developments constituted (Lawson & Poguntke, 2004)and of how successful their responses might be.
In the 1990s and early 2000s, scholars disagreed about the extent and implications of trends in parties' relations with their closest supporters. Their diagnoses tended to highlight the same eye-catching changes, including increased electoral volatility, new party emergence, the collapse of longestablished parties (in some countries), the overall decline of party membership enrolment, and parties' increasing use of intra-party ballots to give members a direct say in party decision-making, including in leadership selection (Aylott & Bolin, 2020;Pilet & Cross, 2014;Sandri, Seddone, & Venturino, 2015). Yet despite using very similar evidence, scholars reached different conclusions about the implications of such changes for party-based representation, particularly with regards to the use of membership ballots. We can identify three distinctive, if sometimes overlapping, perspectives.

Professionalized parties and entrenched political elites
This strand of argument is closely associated with Peter Mair, and with the Katz-Mair cartel party thesis. According to this perspective, politics has become a career, and parties have transformed into vehicles designed to support the careers of individual politicians. Such a diagnosis is not far removed from the political ambition approach that views parties tout court as the result of the incentives which lead politicians to invest in them (Aldrich, 1995;Schlesinger, 1994). What is different in the cartel party thesis and related arguments is the assertion that the resulting parties are empty shells whose primary purpose is to secure stable careers for the individuals at the top of the party; these sorts of parties are designed for winning elections, not for representing the interests and priorities of distinct slices of the population (Katz & Mair, 1995, p. 22;Mair, 2013, p. 88f;Panebianco, 1988, p. 274). In such parties, leaders may implement member ballots as a way of evading obstacles to vote-seeking that might be imposed by party congresses or by mid-level elites who are more wedded to fixed ideas about what and whom the party stands for, and what that implies for party priorities. Indeed, using such devices may be one of the markers of a cartel party, which is specifically manifest in an erosion of the boundary between formal members and supporters, particularly through the spread of primary elections; use of direct votessometimes merely to ratify, other times to decide among alternatives determined by the party leadership; and by-passing of party congresses or meetings, in which communication and coordination among the members/supporters and from them to the center is facilitated, in favor of direct, unmediated, and one-way communication (e.g. via direct mailing or e-mail) from the center to the members/supporters. (Katz & Mair, 2009, p. 761) As this suggests, not only may cartel parties use direct balloting to individualize party-member relations, thus weakening the ability of members to mount coordinated opposition to leaders' priorities. They may further undermine the potential power of party members by allowing non-members to participate in these ballots. In this view, direct ballots can (and likely will) be used to reduce the ability of the grassroots organization to challenge party elites.

Personalized or presidentialized parties
A more extreme variant of the careerist view of political parties identifies a tendency for both governing and opposition parties in parliamentary democracies to become more leader-centred. This trend is described as personalizationor even presidentialization. While the terms are often used interchangeably, these concepts denote distinct albeit related phenomena. In the first instance, personalization simply means that politics become more centred around individual politicians while collective actors such as parties become less important (Rahat & Kenig, 2018). Whereas personalization can happen on all levels of the political system, presidentialization refers more specifically to an increasingly leader-centred functioning of political systems, incorporating electoral processes and behaviour, and the relationships between leaders and their parties. Such developments are driven by various factors including the erosion of cleavage politics, the changing media environment, the internationalization of political decisionmaking, and the growth of the state (Poguntke & Webb, 2005. In the professionalized cartel parties, multiple elites work towards the long-term stability of the political status quo, and the cartel system helps provide job security for party functionaries even when their party does not come out first in an election. In contrast, presidentialized parties are not necessarily long-term projects. While they have often evolved out of cartel parties and can be regarded as their most extreme incarnation, presidentialized parties tend to become merely vehicles to serve personal ambitions of their leaders. Often these are the ambitions of one person, sometimes to the point that the party becomes equated with that leader. As such, the party structures of such a 'personal party' (Calise, 2015) may not be designed with an eye to stability of political succession or to organizational entrenchment. When they grow out of established parties, the political and organizational viability of the party on the ground erodes; new parties are not necessarily formed with long-term intentions, and may be challengers to the established parties (think of Nigel Farage's Reform UK party in Britain or PIS in Poland). In such parties, the role of the voluntary organization is to rally behind the leader, but the voluntary base is not given a meaningful say.
Populist versions of such parties may put a great deal of emphasis on grassroots mobilization and consultation, and that can include the use of member ballots for personnel or policy decisions, but for the most part this participation is designed to endorse the political directions that have been prechosen by party leadership (Mudde & Kaltwasser, 2014;Pedahzur & Brichta, 2002;Wolinetz & Zaslove, 2018;Zaslove, 2012). Thus, even more than in the cartel party interpretation, the personalized and presidentialized party interpretation (including its populist variant) suggest that intra-party ballots will be used in ways that guarantee resounding mandates for party leaders and their projects. Compared with other decision mechanisms, votes taken by member balloting are thus much more likely to produce lopsided or even unanimous verdicts.

More responsive and permeable parties
A third set of perspectives takes a more optimistic view of changes in parties' organizational approaches. This perspective, often touted by parties themselvesparticularly new partiesdepicts parties as capable of fighting back against their waning appeal by giving supporters more direct control over party decision-making processes. This view emphasizes the idea of parties as vehicles for improving representation by strengthening internal transmission belts that elevate policies and personnel that are endorsed by party members. The same mechanisms that some observers criticize as populist are here seen as innovative and inclusive approaches that bring new viewpoints into politics. From this perspective, grassroots organizations should be designed to enhance legitimacy, including by helping parties keep aligned with the priorities of their target electorate. This is why parties and party leaders invoke their relationships with their members/supporters as part of their electoral appeals. It is also why parties may find it desirable to strengthen ties with members (and boost party membership numbers) by giving the grassroots more say (e.g. Garland & Brett, 2014;Kitschelt, 1988;Niedermayer, 2012;Poguntke, 1987). By increasing intra-party democracy, and affording the base more control over elites, parties are (or at least could be) improved vehicles for representing the views of their closest supporters (Lehrer, 2012;Teorell, 1999). Such developments may be enhanced by the advent of modern communications technology, which offers new opportunities for grassroots engagement in decision-making processeseven radical forms of 'liquid democracy' (Caiani, Padoan, & Marino, 2021;Litvinenko, 2012;Mosca, 2020). On this view, contemporary political parties are not necessarily evolving towards becoming empty shells; instead, at least some of them are changing to become more responsive to those who are active in the party, and to supporters whose views may be articulated in meaningful contestation and deliberation. As part of this evolution, members take advantage of the participation opportunities afforded by the party ballots, and the outcomes of these ballots are not foregone conclusions.
These three sets of diagnosis have emerged from a similar reading of the facts, interpreted through different lenses, but the first and second have much darker (though slightly different) implications for democracies. In this paper, we ask how these perspectives stack up today. In particular, we are interested in two questions about the supposed trends in party-supporter relations.
First, all these diagnoses were propelled by the seeming spreading appeal of intra-party ballots, with party members gaining a direct say in important party decisions. While they disagreed on whether ballots represented grassroots empowerment or a weakening of party intermediary structures in favour of 'plebiscitary' decision-making, the predictions agreed that such devices were going to become a common feature of parties in European parliamentary democracies. In light of such consensus, we start by asking what the evidence tells us about the contemporary adoption and usage of procedures in contemporary European democracies.
Second, and more importantly for our investigation here, to the extent that such procedures are used, we ask what the evidence reveals about their impact on intra-party power dynamics. We focus on two aspects: To begin with, how many European parties now explicitly give voting rights to non-members? For students of traditional party politics, accustomed to the distinction between members and other supporters, this blurring of membership boundaries has been note-worthy when it happens, but how common is it in practice? In addition, we ask whether experience with these ballots suggests that they are on the whole less democratically authentic than alternative decision mechanisms, such as party congress votes. In other words, is there evidence that such ballots are most often being used as part of a shift to stronger leaders with more personal mandates, or does evidence support the interpretation that they are being used in conjunction with giving members a meaningful say in party affairs?
Taken together, our analyses provide a unique data-driven inspection of the adoption and circumstances of use of procedures based on party membership ballots, some of which have been much publicized, but which have received less systematic scrutiny across a large number of political parties. Our aim here is not to explain what is driving the changes that we find. That is beyond the scope of this paper, though as suggested by the preceding discussion, there is no shortage of possible explanations. Instead, our project here is to study where European parties currently stand with regards to plebiscitary procedures, and to assess whether the usage and the impact match much-discussed predictions about the implications of adopting such mechanisms. As will be seen, our findings suggest that contemporary realities lie somewhere between our three stylized diagnoses, being neither as universally disempowering as pessimists suggest, nor as broadly empowering as optimists may hope.

Data and methods
We investigate these questions using information provided by the Political Party Database (PPDB) Round 2, covering party structures and practices. The PPDB is a collaborative data collection project through which scholars who are experts about parties in a particular country provide information about parties' statutory rules, their resources, and the rules and conduct of key decision-making procedures. While the second round of PPDB data collection covers 286 parties on five continents, this paper focusses on European parties included in the collection, mirroring the emphasis of the literature with which we are concerned. This gives us a total sample of 188 parties in 26 countries, although the specific findings reported below often cite lower figures due to missing data. For most countries, the data include all parties represented in the lower house of the legislature when the data were collected. Thus, our sample includes parties across the ideological spectrum, and of diverse ages. The Round 2 data were collected in 2017-2020, but the reported decision-making rules span 2011-2019, because coders were asked to report on rules and practices in each party's most recent decision event (selecting a candidate, a party leader, etc.). The PPDB Round 2 data report on (at most) one decision event per party (more information about the PPDB Round 2 data can be found in the online appendix).
To answer our first question about the extent to which contemporary parties offer their members opportunities to vote on important national party decisions, we investigate the diffusion of members' voting rights with respect to four areas: leadership selection, candidate selection, manifesto production, and policy decisions. Elsewhere we have investigated the extent to which party statutes formally recognize the use of membership ballots for party decision-making, and we have classified parties according to an index of Plebisicitary Intra-party Democracy (PIPD) (Poguntke et al., 2016). In this paper, we disaggregate the evidence on the index component 'decision-making', and go beyond formal rights to investigate the extent to which plebiscitary practices have actually been implemented. In other words, our interest here is not only whether such votes are formally recognized in party statutes, but also whether various types of party-wide membership votes have been held at least once during the period of our data collection. The PPDB variables of interest for all our analyses are described in the online appendix, as is any necessary re-coding.
Beyond this, in order to answer questions about the possible impact of ballot usage on intra-party power dynamics, we also look more closely at party leadership votes. We focus here because this is the decision area that theories of personalization and cartelism highlight as being most vulnerable to pseudo-democratic theatre, that is, to apparently more inclusive procedures actually being used to consolidate leadership power. Thus, we expect that leadership decisions will be most revealing of the extent to which membership ballots are enhancing or eroding intra-party linkage. Moreover, leadership contests are undeniably crucial to the strategic and programmatic direction that a party takes, so are intrinsically among the most importantindeed, are arguably the most importantdecisions parties take. First, we investigate whether leadership selection decisions which involve membership ballots look different in two key ways: the reasons for calling such elections, and the competitiveness of these contests by the time they get to the formal voting stage. Predictions of party presidentialization suggest that membership-inclusive leadership votes are more likely to be used for 'coronation' purposes (affirming or reaffirming the personal mandate of a popular leader) as compared to when selection is done by a high-level party assembly (party congress or national executive). If this is true, leadership selection votes that take place at the membership level should tend to have fewer candidates, and higher margins of victory, than selection votes that are delegated to party congresses. This should be particularly true when a leadership contest is triggered by terms set out in party statutes, and hence may be used to strengthen the mandates of sitting leaders, rather than when the contest is held because the position has become vacant.

Findings
We start by examining how far parties provide for plebiscitary votes in their formal statutes. Specifically, we seek evidence that party rules (whether party statutes, or other written procedural rules) recognize that in at least some circumstances members have the right to vote in five types of key intra-party decision procedures: selection of parliamentary candidates, selection of leaders (both political and electoral, in parties that distinguish these roles), selection of presidential candidates in systems with popularly elected presidents who are nominated by political parties, intra-party ballots for policy decisions, and approval of election manifestos. Table 1 sets out the evidence by country.
In Table 1, we see that for most of these procedures, the majority of party statutes and rules do not recognize the option of members having direct voting rights. Only 13.5% of parties give their members a vote on election manifestos, while just under a quarter do so when it comes to the selection of electoral leaders and presidential nominees, and just over a third permit membership votes on the selection of legislative candidates and in ballots on party policies. The most common decision in which party members may get a vote is in the selection of the party leader. In 40.9% of our parties, the rules give members a direct say in the selection decision at least in some circumstances, such as having multiple candidates. Certain countries stand out in particular regards: British parties are especially likely to allow their members to vote on candidate and leadership selection, though policy ballots are less recognized and no parties have rules offering members a vote on election manifestos; all Finnish parties give members parliamentary candidate-selection votes, and most also give presidential Austria 2 (6) 0 (4) 2 (6) 1 (6) 4 (6) 0 (1) Belgium 6 (12) -12 (13) -2 (13) -Bulgaria 0 (1) 0 (1) 1 (2) -1 (2) 0 (2) Croatia 0 (7) 0 (4) 0 (6) -0 (8) 0 (6) Czechia 2 (4) 1 (1) 0 (4) 1 (6) 3 (4) 1 (5) Denmark 4 (7) -2 (6) - -0 (7) 2 (3) 6 (7) 0 (7) Greece 1 (6) -2 (6) -3 (6) 0 (4) Hungary 0 (2) -2 (3) 0 (2) 4 (5) 1 (4) Ireland 5 (5) 0 (6) 3 (6) -0 (6) 0 (1) Italy 0 (4) -2 (5) -2 (5) 1 (5) Lithuania 0 (6) 1 (6) 2 (8) 0 (4) 0 (8) 1 (7) Netherlands 4 (11) -6 (9) -4 (12) 1 (6) Norway 1 (9) -0 (9) -6 (9) 0 (9) Poland 0 (5) 0 (5) 2 (6) - candidate-selection rights, while the German and Irish parties generally allow membership votes on parliamentary candidate-selection. 1 Membership votes also appear to be the norm in political leadership contests in Belgium, Netherlands and Spain. Thus, our first conclusion is to underline that whatever the possible shift towards the adoption of direct membership voting (and PPDB data do not allow conclusions about longitudinal change), in the 2010s these procedures remained the exception rather than the rule among European political parties. Because some scholars have predicted that the expansion of decision rights will extend beyond party members, Table 2 reports on the same procedures as in Table 1, showing the extent to which the rules envisage that these procedures could be open to non-members. This table shows that it 0 (4) 0 (1) 0 (4) 0 (6) 0 (4) Denmark 0 (7) -0 (6) -0 (9) Estonia 0 (1) -0 (3) -0 (6) Finland 0 (8) 2 (8) 0 (8) -0 (8) France 0 (2) 1 (2) 0 (1) -0 (3) Germany 0 (7) -0 (7) 0 (3) 0 (7) Greece 1 (6) -2 (6) -1 (6) Hungary 0 (2) -0 (3) 0 (1) 0 (5) Ireland 0 (5) 0 (5) 0 (5) -0 (6) Italy 0 (4) -1 (5) -0 (5) Lithuania 0 (4) 0 (5) 0 (5) 0 (7) 0 (8) Netherlands 0 (11) -0 (11) -0 (12) Norway 0 (9) -0 (9) -0 (9) Poland 0 (5) 0 (6) 0 (4) - is extremely rare for rules to accord non-members formal voting rights in any of these party decisions. There are a few cases of parties including nonmembers in votes on the selection of parliamentary candidates (one Greek and two Austrian parties), presidential candidates (one French party 2 and two Finnish parties), political leaders (two Greek parties and one each in Italy and the UK), electoral leaders (one each in Austria and Spain) and policy ballots (one each in Bulgaria, Greece and Spain). No parties give non-members the right to vote on election manifestos. Our second conclusion from our data is that the widespread public and scholarly attention given to open 'primaries' in recent years far exceeds the extent to which such rights for non-members have been enshrined in party statutes (Cross & Blais, 2012;Sandri et al., 2015). Some of these widely publicized ballots were held by large parties in large democracies such as France (Parti Socialiste), the UK (Labour Party) and Italy (i Democratici, Lega Nord), which probably accounts for the inflated attention they attracted. That said, it is also worth noting that in about a quarter of the cases for which this information is available (35 of the 61 votes) people could register on the day of the political leadership ballot and take part in the election, meaning that in these cases the boundary between members and non-members is very tenuous. Estonia and Greece were most likely to have parties that ran leadership contests along such lines, but there were examples in Belgium, Italy, the Netherlands and Spain as well. Overall, our data caution against overstating the diagnosis of trends towards the erosion of distinctions between party members and other supporters. Membership-based parties, with rights reserved for party members, still seem to be the dominant organizational mode across European democracies. Formal statutory rights are one thing, but how far have these rights actually been exercised in recent years? The PPDB data reported above relate to the written rules pertaining to a specific decision event, asking whether the rules envisaged that members (and non-members) might have voting rights in some circumstances. Yet the presence of such rules is not necessarily identical with party practices. For instance, rules might envisage a party member vote, but it may be that no vote is held due to only one candidate being nominated. On the other hand, party rules might be silent as to the possibility of a member vote, but one could be organized in an ad hoc fashion if one or many candidates push for this. These factors (plus missing data) explain why Table 3, which reports on ballots that were actually held, shows some slight differences with Tables 1 and 2, which report on rules. Again, we see that member ballots have been most frequently used in political leadership contests, with 67 instances of such votes (i.e. 44.7% of the 150 political leadership contests documented in the PPDB); on only 4 occasions (2.7%) have non-members also voted in such contests. Membership ballots have clearly been the norm in Belgium, Spain and the UK. In about half these instances, those elected have been electoral as well as political leaders. Separate party-wide ballots to decide on electoral leaders have been far rarer, there being sole instances in each of Austria, Czechia, Germany, Greece, Hungary and Romania and three in Spain. In just two cases (one in Austria and one in Spain) have non-members had a say in selecting electoral leaders.
In Table 4 we present data that bears upon the question of whether leadership elections are more likely to be uncontested coronations rather than genuine contests. In 16 cases in which members were balloted about the choice of party leader (29.6%), there was just a single candidate: these were the obvious 'coronations', in which membership plebiscites were little more than symbolic legitimation procedures. There was a single candidate in just one of the four cases in which non-members also had a vote (the  Total  67  4  34  9  2 Note: The data reports the most recent leadership election over this period of time. Source: PPDB R2. Variables in column order: C40PLVT2, C41PLVT2, C88ELVT1, C88ELVT1, C88ELVT2. Sample: see Table 1. Italian case). 3 We can put these figures in perspective by comparing them with what occurred in political leadership selection events in which neither members nor non-members were given a direct say. Of the 54 decisions in which members had a vote, 38 involved two or more candidates (70.3%); in contrast, of the 53 cases in which neither members nor non-members had a vote, only 14 had multiple candidates (26.4%); the remainder were all single-candidate coronations (data not shown in table). This finding works against the argument that the introduction of membership votes necessarily signals a shift towards empowering leaders by increasing the incidence of single-candidate coronations.
In Table 5 we repeat this analysis for parties in which electoral leaders have been elected separately. The former show that Germany, 4 Hungary and Spain  1  2  3  More than 3  Total  1  2  3  Total   Austria  2  0  0  0  2  0  0  0  0  Belgium  3  4  1 Table 3. Reports first round of voting in cases where there were multiple rounds. Source: PPDB R2. each had a case of single-candidate coronations in which members voted: in the Spanish case, non-members also participated. One Austrian party and one Spanish party also had multiple-candidate contests, with non-members voting alongside members in the Austrian case. On the whole, then, these admittedly small numbers tend to suggest that if separate electoral leadership contests occur, they are quite likely to be single-candidate coronations. Again, it is interesting to draw a contrast with electoral leader selections in which members (and non-members) had no right to vote: there are just 11 of these cases in the PPDB, in Austria (3), Germany (3), Hungary (2) and Spain (3). In all but one case (in Spain), these were single-candidate coronations (data not shown in table). Thus, again our evidence shows that selections involving member voting are no more likely than other intra-party procedures to be coronations rather than genuine contests. A further way in which we can gauge the competitiveness of plebiscitary contests for leadership positions is to examine the margins between winning candidates and runners-up. Excluding single-candidate contests, the average winning margin in political leader contests with member ballots is a substantial 33.6%. Again, however, it is instructive to compare this with other political leadership selections for which recorded votes were taken by some party body (for instance, by the party conference). In these instances, the average winning margin was 46.9%. While there is a good deal of variation across countries, and across events, in some cases the winning margins in party ballots are quite narrow. This comparison with other decision modes, as opposed to just looking at membership ballots by themselves, is yet more evidence that the use of membership ballots does not, in itself, indicate that the party voluntary organization is being given a less meaningful choice than would otherwise be the case (Table 6). If anything, the evidence points in the opposite direction.
Another way of diagnosing the meaning of these contests is in terms of participation rates. By this measure, on the whole, it seems that many party members found these contests to be meaningful. The mean turnout for leadership contests was 46.2% of the eligible membership (for the 43 cases for which this was available); the median was 50.8%. 5 Turnout reached or exceeded 50% in 22 (51%) of the elections. Given that voting in these elections is not compulsory, and abstention is by far the easiest route, the fact that so many members use 'voice' options when they are offered suggests that members value such procedures, including when they are more about endorsement than selection.

Conclusion
What have we learned from our data? First, it is now quite common for parties in European democracies to offer their members the right to cast a direct vote in important party decisions, though for no type of decision does a majority of parties provide this opportunity. Such rights are offered by parties in newer as well as more established democracies. In contrast, very few offer such rights to non-members (though some blur the distinction between members and non-members by making it very easy to join). We cannot show a trend with our cross-sectional data but there is ample evidence in the literature that the frequency of membership plebiscites has grown over decades (see Cross & Blais, 2012;Hazan & Rahat, 2010;Kenig, 2009;Kenig, Cross, Pruysers, & Rahat, 2015;Sandri et al., 2015;Scarrow, Webb, & Farrell, 2000). To this extent, our data might be regarded as consistent with the predictions of both the cartel and presidentialized diagnoses, both of which depicted the growing prominence of voting by party members as a shift away from the assembly-based intra-party democracy model of the classical mass parties. Although variation by party ideology does not form part of the cartel or presidentialization theories, we checked all the relationships reported in Tables 1-6 by party family. Parties in certain party families prove to be a little more likely to offer some plebiscitary voting rights to members (e.g. Social Democrats and Greens), which speaks to the occasional importance of ideological commitments to democratic participation, but it is striking that these ideologically-anchored variations are mostly minor, and examples of plebiscitary voting rights for members are found across the spectrum of party families. The plebiscitary model is clearly established as a legitimate and viable way for parties to structure relations between elites and grassroots supporters. However, it is not the dominant one in European parties, and it still remains relatively unused for policy decisions, even where these rights exist. Whether or not these plebiscitary rights represent a gain in democratic quality is not entirely subject to empirical investigation but is, to some degree, a normative question. However, our data offers some clues. We found that direct membership voting rights on leaders are much more frequent than membership votes on manifesto policy. So, recourse to direct membership balloting on personnel issues has not been accompanied to the same extent by an enhanced direct role for members in election programmes, although that does not necessarily imply that members' roles in shaping party policies are weaker than they used to be. And indeed, our data also show that in 60 of the 99 parties for which we have information on manifesto adoption, party conferences were given a vote on the manifesto (33 parties), and in some cases also had (in addition to a vote) a chance for formal input (27 parties). 6 Thus, while the adoption of direct voting rights for personnel selection might represent an individualization of party membership, it does not seem to extend to the policy arena to the same extent. Indeed, it is possible that direct votes for leadership selection may confer an enhanced sense of legitimacy on leaders, which could strengthen their autonomy to steer policy; however, whether that occurs is beyond the scope of our data to resolve.
Elsewhere we have argued that assembly-based and plebiscitary intraparty decision-making represent two distinct types of intra-party democracy (Poguntke et al., 2016;Von dem Berge & Poguntke, 2017). By this logic, an increased involvement of party members through direct balloting does not necessarily represent an increase in intra-party democracy if it undermines the role of party congresses or other party assemblies. Rather, it relies on a different logic of internal decision-making, and both types of decisionmaking can comfortably co-exist. While assembly-based decision-making can allow for discussion, deliberation and amendment, membership plebiscites are more suited to ratifying decisions taken elsewhere. Hence, it is no surprise that parties use membership votes primarily for personnel decisions and that these ballots are frequently decided with a large margin in favour of the winning candidate. That said, it is not clear that such processes are definitely of lower democratic quality, given that leadership contests in which the members (and occasionally non-members) get a vote tend to have more candidates, fewer 'coronations', and smaller winning margins, compared to processes in which party congresses, executives or other bodies choose leaders. Member ballots seem to have at least as much genuine competition as other democratic mechanisms. Our evidence provides no support for generally equating the use of intra-party ballots with populistic trends under which leaders strengthen their position by stifling internal dissent.
Thus, the trend towards a more widespread use of plebiscites in conjunction with some of our findings may be consistent with Peter Mair's contention that parties are increasingly involved in 'ruling the void'in that the new decision-making devices bring increasing attention to the ways that parties operate as arenas for power struggles and platforms for personal careers. It is also consistent with the thesis that parties may increasingly follow a presidentialized logic of leadership. On the other hand, we also want to emphasize that this is not the entirety of our findings. For one thing, while direct member votes are now a well-established procedure among European parties, they have not become the dominant decision-making process in any of our decisional domains. For another thing, contrary to the cartel party thesis, it is by no means the case that decisions are always primarily about personalities rather than policies when membership votes are used. Our data included many hard-fought contests using plebiscitary procedures that focussed on issues and personnel choices that were deeply connected with disputes about policy directions. The German SPD and the British Labour party are but two prominent examples of parties where more plebiscitary leadership elections fundamentally challenged the traditional centrist office-seeking strategy of the party in public office. Thus our conclusion regarding the meaning of these sets of party organizational changes is that none of our three perspectives are correct if taken as indicative of a single developmental trajectory. Of course, this is not to say that the authors originally associated with these perspectives regarded them in this way, but they are nevertheless often associated with statements about widespread trends and common trajectories, when developments themselves are much more nuanced. Perhaps not surprisingly, but still worth stating because it sometimes seems to be forgotten: intra-party voting mechanisms should not be tarred with a single brush, nor should they be viewed either as a symptom or a cause of the hollowing out of party democracy. Indeed, in some cases they may contribute to invigorating itbut in others, they may indeed serve to provide only a veneer of legitimacy for leaders whose main currency is charisma rather than policy offerings. It is not the procedures themselves, but how they are used, that determines which it will be. Notes 1. In the German case, this is regulated by party law which stipulates that constituency candidates for the Bundestag can be selected by either an assembly of delegates or an assembly of members. 2. The case reported in our data refers to the Socialist Party. The Republican Party did select a presidential candidate in 2016 in a ballot open to non-members, but the case reported in PPDB R2 refers to a 2017 selection which was open only to members. 3. The Italian Far Right party Fratelli d'Italia. 4. The Greens had one female and 3 male candidates in 2017. There were separate contests for men and women; hence the contest for the female leader was a coronation. 5. For 31 cases, conservatively excluding 2 contests where the turnout of eligible membership was listed as an implausible 100%. Listed turnout rates ranged from 1% to 80% (calculated from PPDB variables C52PLTO and C53PLCAN1). 6. Calculated from C104MAN5 Role of party congress delegates.