The study of democracy and democratization, and in recent years satisfaction with democracy (SWD), is at the centre of scholarly work in political science but is of importance to other disciplines as well. At the heart of contemporary democratic politics we find the institution of political parties. Modern democracy cannot work without political parties and it does not perform well if these parties are not stable or if they are weak. This has been a strong consensus among scholars of political parties and democracy for more than half a century; arguably, it still holds, as many surveys (e.g. V-DEM) still measure democratization in terms of party existence and free and unobstructed functioning. This is not without reason. Political parties emerged and evolved as institutions that helped legislators, candidates and voters overcome a variety of challenges associated with democracy (Hicken Reference Hicken2020: 1). Parties pioneered the democratization of the political process in the twentieth century, bringing new social classes and strata into the political system (Katz Reference Katz, Katz and Crotty2006; Katz and Crotty Reference Katz, Crotty, Katz and Crotty2006; Mozaffar Reference Mozaffar, Katz and Crotty2006).
However, for many years now, political parties have taken the lead in the list of the least trusted institutions worldwide, and their influence has been constantly on the decline (Dalton and Wattenberg Reference Dalton and Wattenberg2000). Moreover, they are often blamed for most of the flaws of contemporary democratic politics (Webb et al. Reference Webb, Scarrow and Poguntke2022: 151). This process raises a series of challenges not only for political parties but – more crucially – to liberal, representative democracies. One such challenge is the waning of democratic accountability, which, in turn, affects voters and citizens' SWD. Yet, this has not changed the fact that political parties remain the key actors of politics.
Satisfaction with democracy is considered a crucial measure of the political health of any given political system (Singh and Mayne Reference Singh and Mayne2023). In media and public discourse, lack of satisfaction with democratic politics and processes, political institutions and the political elites has been cited as a cause and an outcome of the rise of challenger parties, the bewilderment of institutions that symbolize the status quo, and the catalyst of events such as ‘Brexit’, the Trump presidency, the rise of far-right populist parties, and so on (Eatwell and Goodwin Reference Eatwell and Goodwin2018; Kaltwasser and Van Hauwaert Reference Kaltwasser and Van Hauwaert2020; McCall Reference McCall2020). It also links with wider debates on such issues as the ‘crisis of party’ or a ‘legitimacy crisis’ of liberal, representative democracies (Merkel Reference Merkel2014).
Resulting from the above, research on SWD has been growing over the years. This research examined a number of variables and their effect on SWD and on many levels. Different angles were adopted and have provided various explanations for levels of SWD. On the other hand, scholarly work on the relationship between intra-party politics and workings and their impact on SWD are significantly fewer. Although we know a lot of things about the workings of political parties, ‘we know very little about how parties help to shape citizens’ attitudes towards the democracies in which they compete’ (Webb et al. Reference Webb, Scarrow and Poguntke2022: 152). Moreover, although scholarship has advanced on issues related to the context and results of intra-party leadership contests, it has progressed little regarding their consequences on ordinary party voters. This is the gap this article aims to fill in by examining how intra-party workings, and particularly leadership contestations, may affect SWD among voters. Our main task is not to explain the levels of SWD per se, but to examine whether specific intra-party factors – in this case intra-party leadership contests – might affect such attitudes.
Previous research has shown that intra-party democracy can positively influence SWD by improving citizens' evaluation of democratic processes (Farrell and McAllister Reference Farrell and McAllister2006; Put et al. Reference Put, Lavy and Shomer2014; Webb et al. Reference Webb, Scarrow and Poguntke2022). Given that intra-party leadership elections are arguably the most significant intra-party events, we anticipate that their competitive nature will have a measurable impact on SWD among party voters. Specifically, we expect that highly contested intra-party leadership elections will improve voters' perceptions of democracy by signalling the party's commitment to accountability and democratic norms. However, our main finding is that intra-party politics at its probably most competitive form – that is, elections for the head of the party – does not seem to exercise any significant influence on voters' SWD.
The focus on intra-party leadership politics is timely and important not only because political parties continue to dominate politics in most countries of the globe and particularly in liberal European democracies, thus exercising a significant influence on citizens' quality of life. More crucially, the way politics unfolds in contemporary times, largely focusing on personalities and individual leadership characteristics, is an additional, vital factor that invites attention to party leadership contests. A notable illustration of this phenomenon occurred during the intra-party leadership contest within the Greek radical left party SYRIZA, a party which has been the subject of considerable scholarly scrutiny in recent years. In an unexpected turn of events, Stefanos Kasselakis, a complete outsider hailing from the USA, won the elections. His candidacy garnered significant media coverage and captured the attention of citizens, adding a unique dimension to the campaign.
The definitional part of social sciences is usually a minefield. Our endeavour in this article is not theoretical: it is not to discuss and debate the problems in conceptualizing democracy and SWD (for this debate see Linde and Ekman Reference Linde and Ekman2003). Nonetheless, in order to proceed we need to establish a minimum definitional understanding of the basic concepts we utilize in this article. The first such concept is SWD. Although being far from a perfect indicator of support for the performance of democratic regimes, since it is highly sensitive to different institutional contexts (Linde and Ekman Reference Linde and Ekman2003), political scientists usually define this as the support citizens have for their political institutions, which in turn indicates the political health of a democratic society (Valgarðsson and Devine Reference Valgarðsson and Devine2021). Most scholars treat SWD as an item that taps into the level of support for the way a democratic regime works in practice and not as an indicator of support for the principles of democracy. This is the definition we adopt too, as it fits well with our unit of analysis: political parties, a key institution for contemporary liberal democracies.
Competitive intra-party elections is a second concept. There is a widely held belief, dating back at least to Schumpeter (Reference Schumpeter1942), that competitive elections are necessary for a well-functioning democracy. Nonetheless, there have been surprisingly few attempts to define precisely what requirements must be followed to make an election competitive (Klingelhöfer Reference Klingelhöfer2021: 1). Thomas Brunell and Harold Clarke (Reference Brunell and Clarke2012: 125) give a good description of a widespread understanding, which equates competitiveness with small margins of victory. Justin Buchler (Reference Buchler2007) suggests defining competitive elections as elections in which all sides involved have more or less equal chances to win. The number of competing candidates is also important, as well as the preconditions provided for candidacies (i.e. are they really equal and uniform to all?) and the body that ultimate votes for the new party leader in terms of inclusiveness and/or exclusiveness. Although not all these criteria relate directly to intra-party elections, they can apply in intra-party contests too. Therefore, we acknowledge the above criteria as basic components of competitive intra-party elections too.
The article draws on empirical data for a cross-sectional analysis from three different datasets: one providing information on a significant number of intra-party leadership contestations among parties of different ideological outlooks in 11 different countries over the period 1965–2019 (COmparative Study or PArty Leaders – COSPAL) (Pilet et al. Reference Pilet, Cross and Pruysers2021); and two measuring SWD (Mannheim Eurobarometer – ΜΕ – and the European Social Survey – ESS) (see section on methodology). The 11 countries examined are: Norway, Austria, Belgium, Germany, Denmark, UK, Italy, Hungary, Israel, Portugal and Spain. For a longitudinal investigation of the relationship under examination, we use data from the German Longitudinal Election Study (GLES). The article is structured as follows. In the next section, we review existing literature to which this study contributes, and in the third section we discuss our cases selection, data and methods. The fourth part is empirical, showing the results of our research, which we also discuss. The final part concludes the discussion.
Party politics and satisfaction with democracy
Satisfaction with democracy (SWD) is a common measure of public opinion in social science surveys (Singh and Mayne Reference Singh and Mayne2023; Valgarðsson and Devine Reference Valgarðsson and Devine2021: 1), often used to gauge political support and the health of a democratic society (Aarts and Thomassen Reference Aarts and Thomassen2008; Bernauer and Vatter Reference Bernauer and Vatter2012; Dassonneville and McAllister Reference Dassonneville and McAllister2020). It serves as both an explanatory and outcome variable in research on public support for political institutions, elites and democratic performance (Valgarðsson and Devine Reference Valgarðsson and Devine2021: 1). For stable democracy, it is required that citizens believe in democratic principles, and for consolidated democracies a majority must share this belief (Linde and Ekman Reference Linde and Ekman2003: 392). However, there are complexities in measuring SWD (see Linde and Ekman Reference Linde and Ekman2003).
SWD is seen as a mid-level indicator of political support, lying between regime-type preferences and support for specific actors and institutions (Easton Reference Easton1975; Norris Reference Norris and Norris1999), crucial in debates about democratic legitimacy crises (Singh and Mayne Reference Singh and Mayne2023: 194). Research on SWD focuses on a number of things: its relation to vote choice, citizen participation, governance support and institutional reforms (Singh and Mayne Reference Singh and Mayne2023: 166–199). It also examines factors affecting SWD and democratic politics, such as ideology (Lisi et al. Reference Lisi, Freire, Barberà, Cross and Pilet2015; Poguntke et al. Reference Poguntke, Scarrow, Webb, Allern, Aylott, Van Biezen and Enyedi2016), election outcomes (Anderson et al. Reference Anderson, Blais, Bowler, Donovan and Listhaug2005), democratic representation (Reher Reference Reher2015), accountability (Aarts and Thomassen Reference Aarts and Thomassen2008), electoral systems (Aarts and Thomassen Reference Aarts and Thomassen2008), participation (Blais and Gélineau Reference Blais and Gélineau2007), quality of governance (Webb et al. Reference Webb, Scarrow and Poguntke2022: 152–153) and economic performance (Quaranta and Martini Reference Quaranta and Martini2016).
Political parties play a central role in democracy, influencing voter satisfaction through their functioning and organization. Well-functioning party systems, representing population preferences and maintaining accountability, are associated with higher SWD (Hicken Reference Hicken2020: 3) and increased democracy legitimization (Mair Reference Mair2013). Strong parties are linked to more durable democracies and reduced success of populists (Self and Hicken Reference Self and Hicken2018). Parties also function as democracy's transmission belt – that is, connecting voters to government by representing their interests in the policymaking process and holding policymakers accountable (Hicken Reference Hicken2020: 3). Effective party organizations influence voters' attitudes towards democracy (Webb et al. Reference Webb, Scarrow and Poguntke2022). Intra-party workings also affect citizens' perceptions of democracy. Parties are often seen as democratic systems en miniature and their internal functioning influences citizens' perception of and satisfaction with democracy too. Perceived internal divisions, corruption, unfulfilled promises and unresponsiveness can undermine public trust and lead to dissatisfaction with democracy.
Intra-party leadership politics, competition and democracy
The study of party leadership, neglected until a few years ago – albeit with some exceptions mostly related to intra-party democracy and selection of candidates (Cross and Blais Reference Cross and Blais2012; Cross and Katz Reference Cross and Katz2013; Rahat and Hazan Reference Rahat, Hazan, Zittel and Fuchs2007) – has gained prominence largely due to such processes as the presidentialization (Poguntke and Webb Reference Poguntke and Webb2005) and personalization of political parties (Karvonen Reference Karvonen2010). This is not without reason. Party leaders control ministerial selections (Dowding and Dumont Reference Dowding and Dumont2009), decisively influence candidate selections (Hazan and Rahat Reference Hazan and Rahat2010), define party policies and bring new issues in the agenda (Scarrow et al. Reference Scarrow, Farrell, Webb, Dalton and Wattenberg2000). Crucially, party leaders are often the public face of the party conveying its message to voters and the media both between elections and during campaigns (Pilet and Cross Reference Pilet, Cross, Cross and Pilet2015: 2).
The emphasis on intra-party elections is timely for another reason too. The personalization thesis of Thomas Poguntke and Paul Webb (Reference Poguntke and Webb2005) highlights the growing importance of party leaders and the consequent attention that must be paid to the politics of party leadership. In highly individualized societies with political landscapes largely centred on personalized parties and presidentialized governments (Poguntke and Webb Reference Poguntke and Webb2005), political elites not only attach their message to latent or full-fledged lines of division, but often become lines of division themselves (Bordignon Reference Bordignon2020). The role of party leaders has grown in importance, impacting voter decisions and party image (Aarts et al. Reference Aarts, Blais and Schmitt2011). As such, we expect that intra-party leadership elections will have an effect beyond merely intra-party affairs. Given the enhanced role of party leaders, it is a logical assumption to infer that party voters and not only the membership would be interested in party leadership electoral contests.
Most publications on party leadership examine specific contests and in specific countries, selection criteria and mechanisms, institutional rules, characteristics of party leaders and so on (Cross and Pilet Reference Cross and Pilet2015: 34). Very few studies have examined how party leadership politics impacts on voters' SWD. This is where the present article aims to make a contribution. Party leadership politics involves a number of things. In this article we examine intra-party election measured by the degree of competitiveness and ask how this impacts SWD. We expect intra-party competition as it is expressed in intra-party leadership contests to exercise a positive influence on SWD among party voters.
Previous research focused on other aspects and dimensions of party workings and impacts on voters' SWD. For example, David Farrell and Ian McAllister (Reference Farrell and McAllister2006) found that preferential voting in candidate-centred systems promotes a greater sense of fairness about election outcomes, which in turn increases satisfaction with the democratic system. Miroslav Nemčok (Reference Nemčok2020) further emphasized the role of parties, noting that voters who support the winning party are generally more satisfied with democracy. Gert-Jan Put et al. (Reference Put, Lavy and Shomer2014) highlighted the potential positive effects of democratizing candidate selection procedures on citizens' views on democracy, suggesting that more open and democratized processes could lead to greater political participation and SWD. André Blais et al. (Reference Blais, Morin-Chassé and Singh2017) found that election outcomes, particularly voting for winning parties, and deficits in representation, can also impact SWD. Given the lack of literature and empirical data on the specific issue of intra-party leadership contests and SWD beyond the ‘winners and losers’ framework, we believe that the answer lies with the fundamental premises of democratic theory in general and its application in party workings more particularly and mainly the effect of competition (see below).
The selectorate – those eligible to vote – is another crucial parameter in the entire process. The open and participatory nature of (American) primaries, for example, means that many more members and supporters will be engaged directly in the intra-party contest as opposed to a closed, parliamentarian caucus voting for the next party leader. The more open and inclusive the body that elects the party leader, the more the process is externalized beyond party elites to the society (Cross and Pruysers Reference Cross and Pruysers2019: 484). At the same time, the entire process is not kept behind closed doors but is open for everyone to see and judge.
Looking at the competitiveness of leadership races, Ofer Kenig et al. (Reference Kenig, Rahat, Tuttnauer, Cross and Pilet2015) show that these result in different outcomes: some contests typically look like coronations, others are actual, and sometimes contests are really close, thus enhancing competitiveness. Indicators of competitiveness include the likelihood of a coronation, the number of competing candidates, the margin between the votes won by the top two candidates and the likelihood of the incumbent's success (Kenig et al. Reference Kenig, Rahat, Tuttnauer, Cross and Pilet2015). The degree of competitiveness in leadership contests is affected by various factors that include, inter alia: the political system in which the party operates; the time (year) the selection process took place; the type of selectorate; the ideological position of the party; and the party's governing position.
Intra-party leadership contests are a significant aspect of intra-party politics and can impact on SWD, as they can provide citizens with an opportunity to participate in the democratic process and have a say in the direction of their party and, potentially, the government. In fact, intra-party leadership contests can be seen as a healthy and necessary aspect of democratic politics, as they promote competition, accountability and legitimacy within political parties. They can also serve as a mechanism for parties to renew and refresh their leadership, which can be important for staying responsive to the changing needs and concerns of the electorate. On the other hand, if leadership contests are particularly virulent or divisive, it can contribute to perceptions of instability and gridlock, which can further erode confidence in the democratic process both within the party and in the broader political system. The repercussions of the aforementioned developments are anticipated to extend far beyond the confines of the party membership or the selectorate involved in a particular intra-party competition. These consequences are expected to have a broader impact, influencing a much larger population. If parties are indeed a transition belt from and to society, then we should expect that it would affect the way citizens view not only one party's workings but also how they view the political system overall, which crucially includes SWD.
The way that leadership contests are conducted within a political party can signal to its voters and the wider public whether the party is committed to democratic values and processes, which can impact voters' perceptions of the party's legitimacy and trustworthiness. When a party holds a leadership contest, it can generate media attention and public discourse around its internal workings. For example, if the leadership contest is conducted in a fair and transparent manner, it can demonstrate the party's commitment to democratic values and principles, which can increase public trust in the party and in the democratic system more broadly. This can, in turn, increase satisfaction with the overall functioning of democracy and thus indirectly affect party voters' SWD.
It is obvious that research on the effect of party organizational resources on electoral politics and how they affect citizens perceptions about their political systems has progressed. The same cannot be said about scholarly work on the link between intra-party developments and politics and their relationship with SWD (Webb et al. Reference Webb, Scarrow and Poguntke2022). However, there are assumptions embedded in the study of democratic and party politics that link intra-party politics with a better working of democracy. For example, campaigning by party candidates produces positive externalities or spillovers for other members of the same party ticket (Aldrich Reference Aldrich1995). In the context of intra-party elections, we contend that competitive intra-party elections can enhance satisfaction among voters. Even those not eligible to cast a vote often harbour a consolidated preference for one candidate. Competition is crucial in making the process more engaging and arguably more democratized.
Liberal democracies rest on several core principles and features, which collectively ensure the functioning of a democratic system. Crucial among these is competition. The concept of competition in liberal democracies has been explored from various angles. Jonathan Hearn (Reference Hearn2021) argues that the decline of aristocratic and religious authority, and the national subordination of martial power, opened the way for more liberal forms of society in which authority is routinely contested through competition across economy, politics, culture and beliefs. Nicholas Gane (Reference Gane2020) provides a historical analysis of competition, tracing its evolution and its role in neoliberal economic thought. Kenneth Schultz and Barry Weingast (Reference Schultz and Weingast1996) argue that liberal democracies have a long-term advantage in international competition due to their ability to establish credible limited government. Moreover, liberal democracies rely on free-market economies where businesses and individuals can compete in the economic sphere, fostering innovation and efficiency. Stefano Bartolini (Reference Bartolini1999, Reference Bartolini2000) discusses the relationship between competition and democracy, highlighting the dimensions of contestability, availability, decidability and vulnerability. Terence Qualter (Reference Qualter and Qualter1991) emphasizes the centrality of electoral competition in liberal democracy, viewing free elections as a defining feature. These studies collectively underscore the complex and multifaceted nature of competition within liberal democracies.
Competition is of central importance in both democratic theory and empirical studies of party behaviour (Bartolini Reference Bartolini, Gunther, Montero and Linz2002). Free and fair elections are the primary mechanism for political competition, allowing different parties and candidates to compete for public office. Multiparty systems and plurality of choice are a usual outcome. A healthy liberal democracy also presupposes and encourages competition among various civil society organizations and interest groups which advocate for different causes and represent diverse interests. All the above aim to ensure people's satisfaction with the system in which they live in, as competition in its various forms is vital for ensuring that the system remains dynamic, responsive and true to its core principles of individual liberty and democratic governance.
Given that political parties represent the basic pillar of liberal democracy, which is fundamentally based and structured upon competition in all its functions including the political arena, this should equally apply for intra-party workings. Therefore, we expect intra-party competition and leadership elections to positively influence voters' satisfaction with democracy. When multiple candidates within a party compete for leadership positions, it signals to voters that the party is open to diverse viewpoints and encourages debate. This can enhance the perception that the party, and by extension the democratic system, is more responsive to different interests within the electorate and committed to democratic principles. Transparent and competitive leadership elections can make voters feel that leaders are chosen based on merit and popular support rather than backroom deals, enhancing the legitimacy of the leadership.
Our focus is with party voters. Although not necessarily directly engaged in intra-party leadership contests, and most often they are not, we expect that party voters will be interested and to some extent affected by what happens internally to the party they support and vote in. This, in turn, might affect their perceptions and SWD, given the central role parties have in contemporary liberal democracies. Party voters who do not partake in intra-party elections, not only often have a preference for the leader of the party they vote for, but frequently acknowledge that party leaders influence national politics, sometimes more than parties as organizations. When there are contested elections, it unavoidably affects not only the membership, but also the voters due to the spillover effects of the process. To put it differently, in the same way that we expect norms and behaviours that appear at the system level to affect the party as an open subsystem that reacts to changes in its environment (Barnea and Rahat Reference Barnea and Rahat2007), we also expect the opposite direction of influence. Based on the discussion above we expect that:
Hypothesis: The more contested the intra-party leadership election, the greater the levels of SWD among party voters.
Case selection and methodology
Our case selection for the cross-sectional analysis includes 11 of the countries included in the Comparative Study or Party Leaders (COSPAL). The countries examined are Norway, Austria, Belgium, Germany, Denmark, UK, Hungary, Israel, Portugal, Spain and Italy. These countries share a number of common characteristics that allow a meaningful comparison: they are parliamentary democracies, and they have multiparty systems and membership-based party organizations. As such, the findings presented here are likely to be more generalizable for countries with similar institutional characteristics.
The main objective of this research is to examine whether the degree of competitiveness in intra-party elections for the party's leadership potentially impacts party voters' SWD.Footnote 1 The primary hypothesis is that there is an increase in mean SWD among party voters whose parties had competitive intra-party elections for the party leadership. To arrive at tangible results, we need two sources of data: (a) data on the timing when each party held intra-party elections for party leadership; and (b) surveys of SWD among party voters before and after the intra-party elections.
With regard to the inventory of intra-party elections for the party leadership, we use the COSPAL, which contains information on intra-party elections from 11 countries. For data on SWD, we relied on the Mannheim Eurobarometer (ΜΕ) Trend File 1970–2002 and all nine waves of the European Social Survey (ESS) (2002–2018). In a sequential format, the ME contains the Eurobarometers conducted from 1970 to 2002. From around the mid 1980s to the mid 1990s, the survey asked European voters how satisfied they were with democracy (on a scale of 1 to 4, where 1 corresponds to ‘very unsatisfied’ and 4 to ‘very satisfied’) and which party they voted for in the last national election.Footnote 2 The ESS includes the same question, but the time gap between its waves is two years. Our analysis acknowledges this difference and controls for any implications (see below). Due to the combination of these datasets, the analysis only employs data from countries where the Eurobarometer and the ESS were conducted on a regular basis.
To examine whether competitiveness in intra-party elections affects SWD, we estimated the mean value of SWD of party voters before and after the intra-party elections. The outcome variable is the difference between the two mean values of SWD (before and after t). The two main explanatory variables capture the level of competition in the intra-party elections. The first explanatory variable is a categorical variable with three categories: ‘Coronation’ (baseline level), ‘Duel’ and ‘Multi-competitive’. Coronations are intra-party elections with only one candidate (regardless of whether this is the existing party leader or a new party leader), duels are intra-party elections with only two candidates, and multi-competitive elections are intra-party leadership races with more than two candidates. Alongside the categorical assessment of competition, the establishment of a systematic method for quantifying the extent of intra-party competition for party leadership in various electoral formulas is essential. For this purpose, we employed Markku Laakso and Rein Taagepera's (Reference Laakso and Taagepera1979) effective number of parties (hereafter ENC)Footnote 3 equation:
where n is the number of candidates with at least one vote and $C_i^2$
the square of each candidate's vote share. The equation yields higher values in intra-party elections with many candidates competing for the party's leadership, and each acquires a substantial proportion of the votes. The lower values come from intra-party elections where the winning candidate won by a significant margin from the rest. Thus, higher values of ENC indicate more significant intra-party competition.
In many cases, intra-party elections for the party's leadership are foreseen by the party statutes at specific intervals. They do not occur solely due to an electoral failure or an unanticipated development. Hence, our data contain intra-party elections where the incumbent party leadership faced no opposition (which are the intra-party events that we classify as coronations). We chose not to exclude these cases as the hypothesized mechanism, which may link intra-party contestations and change in SWD, can be evident. In these cases, the ENC value is 1 (the lowest possible).
The statistical models included relevant control variables to control for any latent source of SWD variation unrelated to intra-party elections for party leadership. A principal finding in the literature is that voters of parties that won elections are more satisfied with democracy (Anderson et al. Reference Anderson, Blais, Bowler, Donovan and Listhaug2005; Blais and Gélineau Reference Blais and Gélineau2007; Singh et al. Reference Singh, Karakoç and Blais2012). Therefore, two essential control variables we included are whether the party is in government and the difference in the party's electoral results in the last two national elections. We have added the ideological parameter as one can presume that parties' ideological affiliations directly influence how their voters perceive intra-party procedures.
To control for possible effects of events that took place between the two survey waves, we first added a variable measuring the length of time, in calendar days, between the two waves and a variable measuring the length of time between the leadership race and the second wave. We also added a variable in the models that measure the difference between the last national legislative elections and the intra-party leadership race (also in calendar days). Furthermore, a dichotomous variable was added that takes values when there were national legislative elections between the two waves. Finally, we added a variable that takes value if there was a change in the party leader or if the incumbent remained in office.
Table 1 summarizes the number of observations per country and time period. It should be noted that there are no observations between 1996 and 2002. The reason is that although the two mass surveys we used (ME and ESS) cover the last four decades, not all Eurobarometer waves combine questions on SWD and voting in the last national elections. Unfortunately, between 1996 and 2002, no Eurobarometer wave included both questions of interest. In addition, out of the 186 observations of the independent variable (holding intra-party elections or not), only 56 elections were contested by more than one candidate, and we have data for the vote shares of all candidates. The remaining 130 intra-party elections had one candidate; therefore, the ENC value in these cases is 1.
Table 1. Summary of the Countries and Years of Observations
Furthermore, we ran an individual-level analysis, using panel data from the German Longitudinal Election Study (GLES). More specifically, waves 14 and 15 of GLES were conducted before and after the January 2021 Christian Democratic Union (CDU) leadership election. The ruling centre-right party was in the midst of a crisis as its new leader, Annegret Kramp-Karrenbauer, was challenged and eventually forced to resign after a series of negative results in regional elections. The new intra-party election was held on 15 January 2021. None of the three candidates secured 50% of the votes, so the first two candidates advanced to the second round. There, Armin Laschet defeated Friedrich Merz by 5.58%. In addition to the question of SWD, the GLES provides participants' evaluations of the two candidates, thus enabling us to identify which CDU voters preferred the winning or losing candidate in the intra-party election. The two waves conducted between the intra-party election allow us to examine whether CDU voters were more likely to state that they were less satisfied with democracy than before the intra-party election. This individual-level analysis complements the first and permits us to reach more conclusive findings on the main research question, as it allows us to examine whether there was any effect on SWD for voters of a party having an arguably very competitive intra-party election compared to voters of other parties who did not experience an equivalent situation.
Results and analysis
Intra-party elections are often, and at least in some European countries, a noteworthy political development. However, there are significant differences in the frequency of intra-party elections between countries, types of parties and time periods. Based on the descriptive evidence from the sample, competitive intra-party elections are much more usual in some countries than in others. This suggests that the particular features of each party system may drive different intra-party dynamics and developments.
For example, in our sample, the countries with the highest numbers of competitive intra-party elections are Belgium (14) and Portugal (12). In Belgium, intra-party elections seem quite widespread. Given that the Belgian party system is one of the most fragmented in Europe, the high number of competitive intra-party elections is not so unexpected. In Norway, in all 22 sample cases, the intra-party elections had only one candidate. However, a leadership change was observed in 7 of these 22 cases. The above suggests that there are significant differences between countries regarding the frequency with which parties hold competitive intra-party elections.
The total number of intra-party elections included in our sample is 186; 93 of them took place after 2000 and 93 before 2000. Among the 56 elections that were competitive (i.e. had at least two candidates), 33 were held after 2000 and only 23 before the start of the current millennium. This may suggest that the continuing decline in the public's contempt for political parties, which has intensified over the last two decades, has created deeper tensions within parties, thus resulting in increased intra-party activity and division.
Figure 1 presents the variance of Δ SWD based on the three groups of the independent variable that reports the type of the leadership race – that is, whether or not the intra-party elections had only one candidate, only two candidates or more than two candidates. The range of values for the parties that did not have contested intra-party elections appears to be wider. However, the mean values (represented by the large white dot for each of the three groups) are nearly identical. The mean value for parties that did not have contested intra-party elections is only slightly higher than the other two categories (differences 0.064 and 0.088 respectively). Therefore, prima facie, there does not seem to be a significant decrease in SWD for parties' voters that had contested internal elections compared to parties that did not.
Figure 1. Distribution of Δ SWD and Leadership Race Categories.
Figure 2 shows the association between Δ SWD and ENC. First, we observe that most of the ENC values are 1, as most intra-party elections of our sample had only one candidate – that is, they were coronations. Beyond that, we observe that many ENC values are around 2. This may arise because, in most intra-party elections with more than one candidate in our sample, the total number of candidates is two. Based on the equation we use to impute competition in intra-party elections for party leadership, the value of ENC cannot exceed the value of 2 when there are only two candidates. Based on Pearson's correlation result, the relationship between the two variables is negative but very weak (R = −0.1). The correlation is not statistically significant. Hence, it does not seem that the magnitude of the intra-party competition affects the SWD.
Figure 2. Scatterplot of Δ SWD and Effective Number of Candidates – Regression Line.
Table 2 presents the results of country/survey wave fixed-effects OLS models.Footnote 4 We chose to control for any differences that may arise between countries and time periods, since, as we explained above, intra-party leadership elections are not equally frequent throughout our sample. The dependent variable (DV) in all models is ‘Δ SWD’.
Table 2. OLS Country/Year Fixed-Effect Models
Dependent variable: Δ SWD.
Notes: Models include fixed effects for country and survey wave. +p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001. GE = general election; LR = leadership race.
In Models 1–3, we examine whether the type of the leadership race affects voters' SWD. The reference level of the main independent variable in these models is ‘Coronations’. In Model 1, the coefficient of Duels and Multi-competitive leadership races are negative but non-significant. This does not alter when we add in Model 2 the interaction between the type of the leadership race and the government/opposition status of the party (i.e. whether the party is in government). The coefficients of Duels and Multi-competitive races remain negative and non-significant. The interaction coefficients are also non-significant. Adding the interaction between selectorate type and the type of the leadership race makes the coefficient of Multi-competitive races statistically significant. However, the interaction coefficient is of similar magnitude, positive and barely not statistically significant at the 95% level (p-value: 0.071), indicating a possible ‘cross-cancellation’ of the two coefficients of this interaction. Moreover, the interaction coefficient between Duels and selectorate type is not statistically significant.
In Models 4–6, we examine whether the level of competition in intra-party elections (i.e. the level of ENC) affects voters' SWD.Footnote 5 In Model 4, the coefficient of ENC is negative but non-significant. This does not alter when we add in Model 5 the interaction between ENC and the government/opposition status of the party (i.e. whether the party is in government). The coefficient of ENC remains negative and non-significant. The interaction coefficient is non-significant and negative, showing that highly competitive intra-party elections of parties in government are associated with lower Δ SWD. Adding the interaction between selectorate type and ENC in Model 6 does not change the previous results. Once again, the interaction coefficient between ENC and selectorate type is not statistically significant.Footnote 6
Turning to individual-level analysis, Table 3 presents the results of the individual-level analysis we performed using data from GLES waves 14 and 15. The two survey waves were conducted immediately before and after the CDU's intra-party election for the party leadership on 15 January 2021.Footnote 7 In this election, crucial for the political trajectory of the CDU, the primary contenders were Armin Laschet and Friedrich Merz. Our analysis focuses on the impact of this leadership battle on the SWD of voters for the Christian Democratic Union/Christian Social Union (CDU/CSU) alliance, always compared to the voters of the other parties.
Table 3. OLS Models Using GLES Panel Data
Dependent variables: Δ SWD and Vote-switching.
Notes: Models include fixed effects for country and survey wave. +p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001. GE = general election; LR = leadership race.
The main explanatory variable is ‘CDU/CSU voter’, which denotes if the respondent intends to vote for CDU/CSU (=1) or not (=0). The primary interest is in the interaction between ‘Survey wave (post LR)’ and ‘CDU/CSU voter’ as this will show if CDU/CSU voters are more satisfied with democracy after a very competitive leadership race. Model 7 includes the main effects of the survey wave, CDU/CSU voter status, and control variables such as support for Friedrich Merz (over Armin Laschet), left–right position, gender and political interest. Model 8 adds the interaction term between the survey wave and CDU/CSU voter status to examine how an arguably competitive leadership race influences SWD among CDU/CSU voters compared to other party voters. Model 9 further includes a three-way interaction term among the survey wave, CDU/CSU voter status and support for Merz to explore if the interaction effect varies by the degree of support for the two main candidates. Moreover, including individual fixed effects in these models is crucial as it controls for unobserved individual-specific factors that could influence SWD, thereby isolating the impact of the variables of interest. This approach helps in accounting for potential biases due to time-invariant characteristics of the respondents, leading to more accurate estimates of the causal effects.
The results in Model 7 show a significant negative effect of the survey wave (post-leadership race) on SWD, indicating a general decrease in satisfaction after the leadership race across all voters. However, the coefficient for ‘CDU/CSU voter’ is not significant, suggesting no overall difference in SWD between CDU/CSU voters and other party voters. In Model 8, the interaction term between the survey wave and CDU/CSU voter status is negative but not statistically significant, implying that the competitive leadership race did not significantly change the SWD for CDU/CSU voters relative to other party voters. Model 9 includes additional interaction terms, but none is statistically significant, indicating that neither support for Merz nor its interaction with the survey wave and CDU/CSU voter status significantly affected SWD.
Furthermore, we examine whether the running of the leadership race was likely to have other similar effects on party voters. Expanding on our main hypothesis, we expect that CDU/CSU voters are less likely, in the post-leadership race survey wave, to switch to other parties or to abstain, as the highly competitive internal party process enhanced their perception of their party's levels of internal democracy and, by extension, their perception of the party in general. To test this hypothesis, we compare CDU/CSU voters to the rest of the respondents who stated a voting intention in the pre-leadership race survey wave to see if they stuck to their party or deflected in the post-race survey wave. The logistic regression model presented in Table 4 assesses the likelihood of voter deflection, with a particular focus on the level of support for Friedrich Merz over Armin Laschet to check whether a possible positive effect is homogenous among all CDU/CSU voters.
Table 4. Logistic Regression Predicting Voter Deflection by CDU/CSU Pre-Leadership Race Status and Support for Friedrich Merz (over Armin Laschet)
Notes: +p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001.
In Model 10, the dependent variable is ‘Deflect’, which indicates whether a voter changed their voting intention from the pre-race survey to the post-race survey wave, coded as 1 if deflected and 0 if not. The main explanatory variable is ‘CDU/CSU voter (lagged)’, which indicates whether the respondent was a CDU/CSU voter in the pre-leadership race survey (coded as 1 if yes, 0 if no). Another key variable is ‘Merz support’, which measures the respondent's support for Friedrich Merz over Armin Laschet.
The coefficients for CDU/CSU voter status, Left–right position, Gender and Political interest are statistically significant, indicating their relevance in predicting vote-switching. The coefficient for CDU/CSU voter status (lagged) is negative and statistically significant, suggesting that CDU/CSU voters are less likely to deflect compared to non-CDU/CSU voters. The coefficient for Merz support is not statistically significant, implying that the support for Merz over Laschet does not have a direct effect on voter deflection. However, the interaction term between CDU/CSU voter status (lagged) and support for Merz is positive and marginally significant, suggesting that among CDU/CSU voters, higher support for Merz may slightly increase the likelihood of deflection.
To better understand the above interaction, we illustrate the relationship graphically, as interpreting the interaction term from the regression table can be challenging. Figure 3 provides the visual representation of the interaction. The x-axis represents the preference among the two main CDU candidates, with negative values indicating a preference for Laschet and positive values indicating a preference for Merz. The y-axis shows the predicted probability of deflection, the likelihood that a voter will change their voting intention from the pre-leadership race survey wave to the post-race survey wave. Overall CDU/CSU voters are less likely to deflect compared to the other participants. However, the predicted probability of deflection slightly increases as support for Merz increases, suggesting that those who preferred Merz were marginally more likely to change their voting intention. This finding suggests that indeed CDU/CSU voters are more likely to stick to their party but if this is the impact of holding a competitive intra-party election, this does not have homogeneous effects as Merz supporters are more likely to deflect.
Figure 3. Predicted Probability of Deflection by Merz Support, Differentiating between CDU/CSU and Non-CDU/CSU Voters.
Overall, the coefficients of the primary explanatory variable – the type and level of competitiveness of intra-party elections – are negative but non-significant. This common pattern leads us to conclude that intra-party competitiveness and SWD may be negatively correlated, but we cannot prove a causal link between them.
The individual-level analysis results confirm no causal relationship between competitive intra-party elections and levels of SWD. The magnitude of the respective coefficients is minimal and also not statistically significant, indicating that there is no direct impact on voters of parties holding intra-party elections characterized by high levels of contestation.
Conclusions
The present article examined the effect of intra-party elections for the party leadership on SWD. The results partly contradict the existing, limited literature. Οur hypothesis is refuted. There is a negative (not statistically significant) correlation between levels of contestation in intra-party elections and SWD. Our findings also suggest that the type of the selectorate has no effect on the association between contestation in intra-party elections and SWD. Taken together, the findings of the cross-sectional analysis provide evidence that intra-party leadership contestations do not affect party voters' SWD levels.
This outcome may be attributed to the limitations of our study. One possible limitation is the aggregated nature of the analysis, which cannot capture how much voters are engaged in intra-party processes. This aggregation can obscure voters' attitudes, leading to an incomplete understanding of the effects of intra-party elections. Furthermore, the existing datasets lack comprehensive data on voter engagement and attitudes toward intra-party contests, which are crucial for a more detailed analysis. These limitations underline the need for more detailed data on voters' attitudes towards intra-party democratic procedures.
Possibly, our results indicate that the continuous disdain for parties has ultimately led to a significant distance between parties and voters, a distance capable of mitigating or nullifying any possible effects on voters stemming from intra-party processes. It also confirms the evidence suggesting that the level of SWD is mainly susceptible to macro-factors (such as the government's overall performance and the country's economic situation) rather than micro-developments primarily related to the configuration of the party system and parties' internal politics.
Additionally, the panel analysis, emphasizing the January 2021 CDU intra-party election, mostly confirms the results of the initial aggregated study. The findings from GLES waves 14 and 15 confirm that intra-party election outcomes do not significantly impact overall SWD. They may well suggest a strengthening of voters' proximity to their party, but this does not seem to be a homogeneous effect as it was mainly observed among Armin Laschet's supporters. However, both cross-sectional and longitudinal analyses align in demonstrating the lack of a statistically significant relationship between intra-party contestation and SWD.
However, the absence of a causal link must not detract from the fact that, despite the small number of observations, the association between the two variables is barely non-statistically significant. Contrary to what we anticipated, we believe that the results of this study lead us to infer a possible negative correlation between intra-party elections and SWD. The increased intra-party competition tends to be weakly associated with a decrease in voters' SWD. Future research should examine what factors cause these two variables to coincide, as this study has shown that one does not cause the other, but rather both emerge based on common determinants. One possible explanation for this negative correlation is the ‘winners and losers’ mechanism, where supporters of losing candidates in competitive leadership contests may become disillusioned, impacting their SWD. However, this effect may be more pronounced at a lower level, particularly among party members who are more directly involved in the intra-party competition, and as such, it could not be fully explored within the scope of this research.
The results of this study indicate a necessity for further research that delves into individual-level analyses. Such investigations could offer a more comprehensive insight into how intra-party dynamics, such as leadership contests, influence voters. It is crucial to explore the psychological and perceptual mechanisms at play in these phenomena. By so doing, future research can shed light on the more nuanced ways in which intra-party politics mould democratic engagement and voter satisfaction.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/gov.2024.27.
Acknowledgements
We would like to thank the three anonymous referees for their constructive comments.
Open access
