Politics remains a team sport: On the continued relevance of studying party organization

Abstract

In advanced industrial democracies, politics has become heavily personalized. What are the implications of this process for research on, and the practice of, democracy? Some have called for a reconceptualization of the essence of modern democratic politics in a way that moves research away from the so-called party organization paradigm. We argue here for the continued relevance of (studying) party organizations. Shedding the theoretical baggage of the mass party ideal from the party organization paradigm, we maintain that, while political parties may structure elections less than in the past and have recently underperformed in safeguarding democratic norms and rules, personalization has actually increased their capacity in the parliamentary and governmental arenas. The party organization paradigm prompts us to problematize these contrasting developments and articulate normatively where exactly the shortcomings of current political parties lie when it comes to safeguarding liberal democracy.

Introduction

Going into the fourth quarter of the 20^th century, it was widely accepted that modern democratic government was party government (Blondel and Cotta, 1996; Castles and Wildenmann, 1986). There were a few possible exceptions, such as the United States (Fiorina, 1987) or Switzerland (Lehner and Homann, 1987), but even there, parties were recognized as essential elements of the national Herrschaftsorganisation, notwithstanding that they did not conform well to the ideal-types of party (United States) or partyness of government (Switzerland). Over the last several decades, a number of developments have led some to question the continued centrality of parties. Among these developments have been declines in levels of party membership (Scarrow, 2015; Van Biezen et al., 2012; Van Haute and Gauja, 2015) and party identification (Dalton and Wattenberg, 2000; Rahat and Kenig, 2018: tbl. 3.5), increased electoral volatility, increased focus on individuals in media reports about politics (Katz and Mair, 2018: 95; Adam and Maier, 2010; Karvonen, 2010), and the rise of social media (and the internet more generally) as venues for political discussion and the transmission of political messages.

This questioning of the centrality of party has led both to shifts in the research agenda of political scientists, as new topics seem to become more important, and to an interpretive debate within the literature about whether, on the one hand, these trends represent an adaptation of parties to new circumstances without making parties any less central or whether, on the other hand, these trends represent a genuine weakening of parties, perhaps calling for a new conceptualization of the essence of modern democratic politics (for a summary, see Rahat and Kenig, 2018: 19–24).

Some have indeed called for a self-conscious and deliberate abandonment of the “party organization paradigm”. The fear is that adherence to that paradigm makes scholarship conservative, dictating adherence to handed down theoretical axioms and conventional methods and thereby preventing it from adequately appreciating new developments. The suggestion is that a paradigm shift (Kuhn, 1962) is in order or already underway, so that “politicians […] [become] the central basic unit in the analysis of democratic politics” (Friedman and Rahat, 2025: 147; cf. Sikk and Köker, 2023).

Should we heed this call or can the developments that have led to questions about the centrality of the parties be incorporated into the party organization paradigm? Even if currently dominant parties are becoming weaker, does this mean that parties tout court are becoming less important, or only that new, and new types, of parties (but parties nonetheless) are becoming more important? After all, parties were central players in the pre-democratic liberal era and before the “invention” of the mass party, and may well survive its relative demise. Second, can the functioning of contemporary democratic politics be analyzed in the absence of strong attention to political parties? Are there functions that must be performed for democracy to work and are those functions the ones that define party in such a way that any organization or other structure that performs them is by definition a party? ¹

In this paper, we want to advance the position that the party organization paradigm remains a vital perspective for understanding democratic politics. Specifically, we make two main arguments. The first is that democratic politics, especially but not only in parliamentary systems, remains a “team sport” with parties effectively being the teams. This is notwithstanding an apparently greater focus on individuals rather than parties per se, and notwithstanding any apparent weakening of the connection between citizens and civil society organizations, on the one hand, and parties, on the other. We suggest that some forms of personalization may actually enhance team-like behavior. The second is that political parties (but not all of them) have been essential defenders of liberal democracy, with the capacity and willingness of party elites to filter out politicians posing illiberal or demagogic threats to the system serving as important guard rails. We resuscitate the defense of the liberal democratic system as an explicit function of party organizations. This function has been made even more important – but also more problematic – with the rise of personalization and increasing efforts to “democratize” the internal workings of parties.

The legacy of the mass party model

It is important to recognize that the evaluation of parties as being in decline, and the corresponding suggestion that a paradigm shift might be required (or be in process) ultimately rests on an often-unspoken acceptance of the mass party model of party organization as the definition of party and high “partyness of government” ² (Katz, 1987) as the correct characterization of democratic politics, at least in Europe during the second half of the 20th century (e.g., Schmitter, 2001).

And given the importance of these assumptions to the current discussion, it is also important to recognize that they are both serious exaggerations or distortions of reality (Scarrow, 2015). The mass party model was always an ideal type. Mass parties often were dominated by their leaders (Adenauer, Ben Gurion, Churchill, De Gasperi as individuals, or by small cadres of leaders); branch meetings were often little more than social gatherings or sessions in which party leaders instructed their followers what to think (Duverger, 1964), rather than followers instructing their leaders what to do. Even when features of the mass party model (party congresses, election of the party’s leadership by the party’s members) were imposed by law, the reality could be quite different from the idealized “internal party democracy” (see Cross and Katz, 2013).

Similarly, the degree to which the Herrschaftsorganisation of a society is dominated by parties, its “party governmentness” (Katz, 1986), has always been limited by, among other things, the importance/power of non-party civil society organizations, ³ big business, independent media, and/or a strong merit-based civil service. As this suggests, in thinking about party government the distinction between “party governmentness” and the “partyness of government” is crucial (Katz, 1986). In other words, it is possible to have high partyness of government even when there is relatively low party governmentness, because real political power resides outside of the formal institutions of government or other loci dominated by parties.

Given all of this, rather than abandoning the party organization paradigm, we may need to sort out the assumptions defining the paradigm from additional assumptions of the mass party model that have to date dominated research within the paradigm.

The core of the party organization paradigm

To arrive at the core of the party organization paradigm, we must first define party and then consider the role that parties play in democratic government. The literature is of course replete with definitions of party (see Katz, 2017: 209). In this paper, we take the definition proposed in Rudolf Wildenmann’s Future of Party Government project (Katz, 1987): a party is an organization or group exhibiting team-like behavior, aiming at political power, and claiming legitimacy on the basis of success in democratically contested elections. ⁴

It is important to note what this definition does not say. Teams may be organized in many ways, and this definition is consistent with all of them: the traditional (pre-Margaret Thatcher) British Conservative party “constitution” as described by John Ramsden (“autocracy tempered by assassination”; see Hennessy, 2001: 295); leadership based on charisma; decision-making through Quaker-like consensus building; as a representative or direct democracy. Leadership (either by a single leader or a group) can imply the capacity independently to make and impose decisions, or possession of a talent/capacity for consensus building, or simply a talent for articulating (or bending to) the preferences of the followers and then providing strategic guidance for their realization.

The definition says nothing about the particular organizational form, including saying nothing about the role (or even about the existence) of party members, or whether party should be understood as a membership organization rooted in civil society (the mass party model) or simply as an institutionalized alliance of politicians (Aldrich, 2011).

And, although this definition requires team-like behavior in action, it says nothing about the basis for that unity – agreement on ideology or policy (e.g., Socialist parties), ties to a social category (e.g., le Bloc Québécois), loyalty to a person or group (e.g., Berlusconi’s Forza Italia), naked self-interest or ambition – only that the people making up the party act coherently. It is thus indifferent with regard to Rahat’s (2024: 214) question of whether “they [are] mainly teams or mainly platforms for an individual leader … or politicians…?”, asking instead whether the supporters of those on the “platforms” act as members of a team, whatever the reason.

This definition identifies an ideal type, with the “partyness” of parties varying over time, space, and across the venues in which they operates. But again, it is the presence or absence of team-like behavior on the part of those comprising the party, and not the basis for that behavior, that defines partyness. Thus, “the presidentialization of what Poguntke and Webb (2005: 9) call ‘the party face,’ which implies ‘a shift in intra-party power to the benefit of the leader’” (Rahat, 2024: 215), may alter the internal dynamics of the party, but it does not make the party itself any less relevant or party-like.

Given all this, why is personalization so often regarded as a challenge to party and party government? The answer appears to be commitment not just to the mass party model as an organizational form but to a particular conception of how democracy should work within that form. A central part of this conception is the democratic legitimation of government through the establishment of a chain of principal-agent relationships (party as the agent of a segment of the electorate, ministry as the agent of a party or coalition of parties, administration as the agent of the ministry; Katz, 2014: fig. 2). This is, of course, completely compatible with high levels of personalization, provided it does not undermine unity. But there are often a number of additional stipulations. Perhaps the one most incompatible with personalization is that the connections be based on policy (or ideology): that voters are selecting parties as their agents to do particular things (i.e., as delegates; Wahlke et al., 1962: 276–77) rather than as trustees (i.e., authorized to make decisions on the voters’ behalf but independent of the voters’ immediate preferences). The mass party model in this way supports a view of voters giving policy mandates through elections that are then carried out by parties (cf. Klingelhöfer, 2026).

In this case, the choice of agent based on personal characteristics – perhaps especially those that have nothing directly to do with politics, – necessarily diminishes this form of policy-oriented linkage. Related to this is an understanding of authority within the party as being based on contestable internal election, that is rational-legal authority in Weber’s terms, rather than charismatic authority, which would be associated with higher levels of personalism. Further, in thinking about the chain of principal-agent connections, this view sees party as an active linkage between the electorate and the government (Lawson and Merkl, 1988), rather than simply as an object of choice (Sartori, 1962; Schumpeter, 1950). ⁵

Parties’ continued relevance

Even if the policy-oriented mass party model of democratic government and governmental legitimacy seems at times to be increasingly at odds with reality, political parties, without this additional baggage, remain central to Western democracy. For the present discussion, we focus on two functions that are particularly crucial to the democratic system. Parties (a) coordinate parliamentary and governmental activity, and (b) structure elections. ⁶

Recent developments suggest that (a) remains as robust as ever but that in particular (b) has been in decline. This is one of the primary reasons that the question about the utility of the party organization paradigm arises. Given the importance of the mass party as the “legitimizing myth” of Western democracy, not only do the declines themselves, but also ways to arrest these declines, become important research topics. We develop these themes in turn.

Parties coordinate parliamentary and governmental activity

In the stylized account of Aldrich (2011: 32–35), a political party is seen as a “long” legislative coalition: Whenever there are distributive consequences to public policies, legislators can reap higher returns when forming narrow coalitions compared to acting independently or unanimously. Because of the transaction costs involved, forming an enduring party coalition is cheaper compared to assembling new ad hoc majorities for every new piece of legislation. This holds even more in parliamentary systems, where legislative and governmental coalitions between parties largely overlap and coordination into parties and (as often necessary) multi-party coalitions provides not only policy-benefits (Laver and Shepsle, 1999) but also the spoils of governmental office. Even nominally independent MPs, often find that they can only be effective in parliament and government if they either form a party or join one of the previously existing parties. It is party leaders who negotiate coalition agreements, organize legislative majorities, and control the legislative agenda, speaking (not always with total success) for their parties as corporate entities.

In the parliamentary and the governmental arenas, the partyness of parties is classically high, particularly in parliamentary systems (e.g., Sieberer, 2006). Politics here is clearly a team sport and without coordination nothing much can happen. High levels of personalization are completely compatible with high partyness (especially in terms of centralized personalism; see Rahat, 2024 for the distinction between centralized and decentralized personalism). ⁷

The party organization paradigm thus suggests that, rather than assuming a kind of trade-off between the dominance of a leader and team-like behavior, we need to problematize the former. The degree to which a (charismatic) leader can simply command loyalty is something that varies over time and across parties. ⁸ A political party may come to be identified with an individual “star”, but just as a team of stars who do not cooperate with each other is very likely to be defeated by a team of inferior players who do work together, even if a party manages to elect a large number of deputies simply as followers of “the leader”, their team-like behavior in office still must be maintained. Conceptually and empirically, we need to distinguish the importance of personality within the party, from the importance of personality instead of party, from the leaders’ personality defining the party.

Centralized personalism indeed increases partyness. The current Republican Party in the United States is a case in point. As it has become the party of Donald Trump, and is itself increasingly defined by him, it has become more party-like. The loyalty to the leader has become the ultimate measure of what it means to be a “good Republican”. But this increases team-like behavior in Congress rather than undermining it. ⁹

Three additional considerations highlight the need to study party organization, even in contexts of extreme centralized personalism. First, the leaders’ influence over the party’s organizational structures often is crucial to their authority. The capacity to nominate candidates running under the party label exemplifies this. Consider in this regard Geert Wilders’s tight grip on who can be on the list of his party (De Lange and Art, 2011).

Second, even in highly centralized (-personalized) parties, the fact that the leader is never truly “alone” exerts a constraining effect. Sahra Wagenknecht – who leads a left-authoritarian split off from the German socialist party under her name (Wagner et al., 2023) – was forced to enter into a public fight with her local counterpart, Katja Wolf, when the latter publicly demanded “Pragmatism instead of ideology!” in coalition negotiations after state election in Thuringia. ¹⁰ Although Wagenknecht reasserted her authority, even entering direct negotiations with the Thuringian leader of the Christian Democrats, ¹¹ the internal disagreement carried into intra-party ratification of the coalition agreement and beyond. ¹² Indeed, an attempt of Wagenknecht to sack Wolf as state party leader at the subsequent convention of the Thuringian BSW was unsuccessful. ¹³ Even where a leader is firmly in control organizationally, they have to fear infighting for electoral reasons (Bolleyer and Kölln, 2024; Greene and Haber, 2015; Klingelhöfer and Müller, 2024) and for the maintenance of their authority. Indeed, the episode highlighted to many that the Bündnis Sahra Wagenknecht is more than just Sahra Wagenknecht.

Third, while leaders sometimes unilaterally shape internal rules and exert high influence over the organization, ¹⁴ the party organizational set-up also constrains who can become the leader in the first place. Ansell and Fish have suggested that, if personalism emerges within a party, the kind of personalism reflects the share and confluence of internal ideological and territorial divisions (Ansell and Fish, 1999: 292). Similarly, Dilling’s (2024) study of the relationship between factionalism and party adaptation suggests that who comes to power and the kind of leadership they can exert depend on the level of factionalism within the party, the latter itself a remnant of the initial way that a party’s leadership board was constituted. Absent complete ownership of a party by the leader (the personal party of Rahat, 2024), there is as good a theoretical rationale to expect the leader to reflect the party as to expect the party to reflect the leader.

Parties structure elections

Parties also solve the coordination problems associated with democratic elections (Aldrich, 2011: 44), which became particularly acute with the advent of mass suffrage (Blyth and Katz, 2005: 35–36). Party organizations’ roles here concern different but related aspects.

First, parties are crucial to the recruitment (or selection) of candidates. Although contemporary standards require that independent (non-partisan) candidacies be allowed, ¹⁵ parties provide the overwhelming number of candidates and independents continue to be a tiny minority of elected officials. In electoral systems with closed lists, the party is explicitly the object of choice. And even where citizens formally vote for candidates, they usually will be visibly associated with parties on the ballot.

More than that, parties give coherence to the set of candidates who run with them. Parties publish manifestos that, in the mass party vision, bind each candidate should they enter office. They also coordinate the election campaign both in substantive and tactical terms. Less formally, the party label generally comes with a reputation that candidates can attempt to harness (or less commonly, from which they can attempt to distance themselves). The coherence in program and campaigning is then supposed to have reverberations at the voter level. Because parties provide not only momentary coherence but also continuity over time, they can serve as objects of identification for voters in a way that individual (and mortal) politicians cannot.

In a sense, partyness has always been less pronounced in interactions with the electorate compared to the parliamentary and governmental realm, particularly in electoral systems that allow intraparty preference voting, and even more in systems with panachage. Such systems undermine the team-like behavior of candidates during campaigns (Katz, 1980), giving them incentives to cultivate personal brands (Carey and Shugart, 1995). While there is no clear trend of increasing personalization of election campaigns (Kriesi, 2012), parties are less and less objects of identification for citizens (Dalton, 2020: 192) and vote choice among parties depends more and more on who the party leaders are (Garzia et al., 2022). Most critically, this relates to how politics is communicated to citizens via the media. Here, evidence of personalization is often strongest (Friedman and Rahat, 2025: 132).

From the perspective of the party organization paradigm, the decrease in the degree to which parties structure elections requires us again to problematize the resulting variation in partyness rather than move to the candidate level (Sikk and Köker, 2023). Electoral choice is only likely to be meaningful if it takes into account who is allied with whom. Only the continuity of the parliamentary “long coalition” (Aldrich, 2011) allows voters to reward or punish based on performance. This is because the “slate of candidates” (Sikk and Köker, 2023: 8) indeed changes frequently, with incumbents retiring and dying, and people moving in and out of politics and potentially also between parties (Sikk and Köker, 2023; Volpi, 2019). And while when the slate of candidates has changed, the party likely has also changed in other important dimensions (Sikk and Köker, 2023), treating the list of candidates as analytically primary might foreclose avenues for our understanding. Specifically, akin to how certain kinds of party organization bring forth specific leaders, characteristics of, and developments within, a party organization as a whole also attract (and repulse) potential candidates. What is more, the party is never completely malleable. Even if 100% of its candidates would change from one election to the next, organizational habits and electoral reputations likely leave their mark.

Parties’ (need to) defend democracy

There is another function that political parties should fulfill – defending the democratic system itself. It might seem unusual to add the defense of democracy as a function of political parties, but we argue that it is indeed a central one, which has been highlighted in the past by theorists such as Sartori and Schumpeter, but has been ignored as long as democracy was working reasonably well.

Much of democratic theory has been concerned with protecting the majority from exploitation by a minority. While there have been some concerns about the dangers of demagogues in general (e.g., Hamilton et al., 1982[1788]) and concerns raised by the erosion of democratic norms under the fascist experience or the rise of new populists to power (e.g., Levitsky and Ziblatt, 2018), the idea that, democracy may itself need to be protected from “the people” is often not addressed. Instead, the electorate has been assumed to be liberal and democracy-supporting.

Sartori in this respect notes that

[d]emolatry notwithstanding, quite a number of democracies have been overthrown by popular insurrection, and many dictatorships have been legitimized by plebiscite. Conversely, all existing democracies have been founded and established by ad hoc elites and minorities. What we have to fear, then, is that democracy–as in the myth of Saturn–may destroy its own leaders, thereby creating the conditions for their replacement by undemocratic counter-elites. (Sartori, 1962: 119; emphasis in original)

So, when (significant parts of) the electorate support actors that actively seek to undermine the liberal democratic system, mainstream parties need to step up in its defense. While the defense of the liberal democratic system is not a function that an organization must fulfill to be a party in a minimal sense (i.e., illiberal organizations can exhibit team-like behavior in the quest for political power), its abrogation endangers the liberal democratic system and, in consequence, also the political party as such. Just as we can define a fish without making reference to the sea, a fish cannot live without water.

Historical experience suggests that the function of safeguarding the democratic character of the system cannot be sufficiently performed solely by non-party actors. A central turning point of the Weimar Republic was when the Nazi party was included in government. (Minority) opposition (from within and) outside parliament was not able to hold off further descent into fascism. What is more, this function must play a vital role even in hypothetical alternative visions of democracy, such as lottocracy: As long as democracy entails freedoms of thought and association (Vandamme and Lucardie, 2025: 7), forces seeking to undermine these components will have to be kept in check.

The external and the internal defense of democracy

Defending liberal democracy has two behavioral manifestations. The first is to coordinate a cordon sanitaire that holds illiberal forces from power (cf. Vandamme and Lucardie, 2025: 12). Coalescing in order to “block” an extreme party from gaining office or influencing policy-making (Downs, 2001: 27) has indeed been found to be an effective strategy in holding challenges to democracy at bay (Bartels, 2023; De Jonge, 2021; Heinze, 2018). It can be seen as a “militant” defense of democracy (Capoccia, 2005). ¹⁶

The problem is that cordons sanitaires, which function like social norms (Axelsen, 2024), are hard to sustain. With electoral breakthroughs of extreme right parties, and given their lip-service to democracy in the abstract, elements of mainstream parties begin to question whether to uphold the cordon or to move toward engagement. This has often led to considerable intra-party strain, undermining partyness. Particularly for center-right parties, the process has proven difficult with many opening the gates (Bale, 2003) but also wavering back and forth in their approach. In this regard, consider the Austrian ÖVP which has governed together with the FPÖ from 2000 to 2005 and 2017–2019, and in early 2025 appeared ready for a while to enter a joint government again, this time as a junior partner. Rather than a clear and full embrace, the party reoriented whenever it became obvious that the FPÖ threatened the democratic system (e.g., the Ibiza scandal of 2019). ¹⁷ Importantly, however, once steps to engage parties threatening the democratic systems are taken, it is hard to walk them back and to reestablish firm disengagement (Heinze, 2018: 303) and to re-delegitimize those parties in the eyes of the voters (cf. Daur, 2025; Valentim, 2024).

How the “mainstreaming” of extremist actors unfolded highlights not only decreasing partyness but also decreasing party governmentness. In particular, the media play an independent and highly consequential role in how visible extremist parties are and whether they are portrayed as legitimate competitors or not (De Jonge, 2021; Bartels, 2023: 197; Völker, 2026; Völker and Saldivia Gonzatti, 2024). When the media do portray these challengers as legitimate contenders, the pressure for the established parties to follow suit increases. And of course, once (some) parties do, the media also follow.

The second way in which parties defend democracy is internal. Party elites can use organizational channels to screen those who would join their ranks, eliminating candidates who would undermine democracy before they can make demagogic appeals to the electorate. Party organizations essentially “vet” would-be political leaders for their “responsibility”. This, of course, assumes that the current elites are themselves committed to liberal democratic values.

Parties here also arguably have under-performed. In an attempt to revitalize the supposed direct impact of members and supporters on party affairs envisioned in the mass party, or perhaps simply in an attempt to reverse declining membership numbers, many parties have increasingly “democratized” (e.g., Kenig, 2009). Whereas with centralized procedures of leadership and candidate selection, party elites were firmly in control and able to screen contenders, open procedures invite greater unpredictability and allow only for informal steering by party elites. An example would be the US primary system, in which elites can push their preferred candidate only indirectly (Cohen et al., 2009), resulting in a more severe coordination problem and the possibility of the elite being overruled. This was seen in Donald Trump’s initial rise to the Republican nomination in 2016 (Bartels, 2023: 226–28). Another example was Jeremy Corbyn’s election as leader of the British Labour Party over the overwhelming opposition of members of the parliamentary party, which had been accomplished by “recruiting a new party to outvote members of the old one” (Rentoul, 2017). ¹⁸ Similarly, an open system potentially allows more influence of outside actors (such as economic elites) who might not have an intense commitment to democracy, undermining party governmentness “from the inside”.

Implications of the party organization paradigm

That parties are underperforming in the defense of democracy underscores the importance of considering reforms. The task is especially acute given the almost unrestrained circulation of social media ads and fake news, undermining the gate-keeping capability of the traditional media.

An initial step might be to abandon the mass party assumption of parties as associations of citizens – which, as we described above, is not inherent to the party organization paradigm itself – and move more forcefully to the Sartorian/Schumpeterian view of parties as alliances of politicians engaging in a competitive struggle for power (Schumpeter, 1950: 283).

With this comes the need for a thorough reevaluation of the concept of leadership (Klingelhöfer, 2026: 230–231). Leadership is often falsely taken as antithetical to democracy (Sartori, 1962: Chap. 6). The very existence of elites in nominally democratic systems consequently proves to critics our inability to realize democracy. The mass party model concurs with this standard view in the sense that leaders are seen as agents of their parties and, in the last step, their social segment. For Schumpeter and Sartori, instead, nothing in the aggregation of power in the hands of the few is per se undemocratic. According to Schumpeter “[p]arty and machine politicians are simply the response to the fact that the electoral mass is incapable of action other than a stampede [. …] The psycho-technics of party management and party advertising, slogans and marching tunes, are not accessories. They are the essence of politics. So is the political boss” (Schumpeter, 1950: 283).

The democratic formula need “neither to eliminate nor submit passively to power, but to make it a function; to control the leaders in the exercise of this function; and to put in office responsible, accountable, and capable leaders” (Sartori, 1962: 120). The issue thus is not how to minimize leadership (patently, the rule of a minority) for the supposedly more democratic rule of the majority. Rather, how can we improve the quality of leadership given that the functioning and the very sustenance of the democratic system itself depends on it?

“[T]he first condition [for the success of the democratic method] is that the human material of politics–the people who man the party machines, are elected to serve in parliament, rise to cabinet office–should be of sufficiently high quality” (Schumpeter, 1950: 290). Schumpeter discusses two characteristics that the “human material” of politics requires. First, politicians “must be on an intellectual and moral level high enough to be proof against the offerings of the crook and the crank, or else men who are neither will be driven into the ways of both” (Schumpeter, 1950: 294). Second, they need to have a large degree of tolerance for different opinions and be “resolved to abide by the rules of the democratic game” (Schumpeter, 1950: 301, also 295–96; cf. also Bartels, 2023: 232–33).

Because, “[p]ro-democratic elites are […] an essential guarantee of the system” (Sartori, 1962: 118), we need to look for mechanisms that ensure their existence. Schumpeter and Sartori place great emphasis on the competitive forces of election to weed out the “the crook and the crank” but the way in which parties organize might be even more important. We need institutional mechanisms that provide elites room to maneuver and incentivize them to forgo short-term benefits for the long-term benefit of sustaining the democratic system (Bermeo, 2003: 237–52; Bartels, 2023: 262): The recent experience with decentralized leadership and candidate selection methods recounted above suggests that the “democratization” of internal party affairs may be counterproductive because it allows politicians to rise to the top without being sufficiently “vetted” by existing party leaders for their commitments to cordons sanitaires and the democratic rules of the game more generally.

While the decentralization of organizational power in parties undermines the capability to vet candidates, excessive centralization around a single leader might be equally counterproductive. If a single person vets the candidates, the partyness of the party may well be enhanced, but the chances increase that personal loyalty trumps democratic commitment. Consider how central Victor Orbán’s tight control of party organization was to democratic backsliding in Hungary (Bartels, 2023: 229; Metz and Várnagy, 2021: 323–25). More generally, parties with heavily centralized personalism seem to favor illiberal and excessively majoritarian views of democracy (Klingelhöfer and Rahat, 2026).

Taken together this suggests what to retain from the mass party model: the view that party organization should vest vetting power in collegial institutions that bring together elites from different parts of the party (Rahat, 2024). When evaluating proposals for party organizational reforms the primary normative yardstick should not be whether the reform enhances the realization of a hypothetical “will of the voters” or “will of the party members” but rather whether the reform enhances the recruitment of leaders committed to liberal democracy.

Conclusion

Political parties remain essential players in democratic politics. And because parties are collective actors, the ways they organize to make decisions and choose leaders remain important to understanding democratic politics. Increased personalization reflects changes in the way parties organize as well as changes in their environment.

To conclude that personalization undermines the party as such – and that therefore we should adopt a more personally-oriented paradigm – would be premature because the effect of personalization varies among the venues in which parties operate. Personalization decreases the capability of parties to structure elections and hold anti-democratic forces at bay, but it may actually increase partyness in parliament and government. Even so, this may still undermine the principal-agent policy linkage between the citizens and the government envisioned by the mass party model, possibly resulting in a kind of cult-of-personality reinterpretation of democratic norms.

As was recognized by Schumpeter and Sartori, democracy critically depends on the liberal democratic commitment of elites (particularly party elites) far more than it does on the dubious liberal commitment of ordinary citizens. Therefore, a particularly important set of questions concerns the conditions, including the organizational conditions, that favor elites with such commitments (and filter out would-be elites without them). Two manifestations of this would be with regard to elite recruitment and election within parties, and the willingness of elites to sacrifice possible short-run advantages of breaching a cordon sanitaire that excludes anti-democratic forces from power.

One hypothesis that our discussion suggests is that excessive intra-party democracy, in particular the use of primary elections, undercuts the capacity of party elites to protect democracy and may ultimately lead to exactly the kind of people who are dangerous for democracy joining party elites. This hypothesis is in need of further research. Rather than studying individual politicians and personalization instead of party organization, we need to focus on the normative commitments of party (would-be-) elites and how they can be directed in ways that support, rather than undermine, liberal democracy.

Both the politics of accommodation (Lijphart, 1975), which emerged against the background of the collapse of the Weimar Republic, and the cartel party model (Katz and Mair, 2018) – for all of their flaws – highlight the possibility of reliably institutionalizing ways that makes parties discount short-term gains. We should draw lessons from this for the defense of democracy. Otherwise, we might come to regret the passing of the cartel party more than we did the passing of the mass party (cf. Kirchheimer, 1966: 200).