Introduction and theoretical framework

Parties in established democracies have recently faced three significant trends that have altered their relationships with the grassroots: declining voter loyalty, declining party membership, and the declining importance of cleavage politics. The three constitutive organizational elements of party politics – namely, the ‘party in central office’, the ‘party in public office’, and ‘the party on the ground’ – are developing in very different directions and coping with general declining political trust and participation and growing political discontent in different ways. Several authors have pointed out that the ‘party on the ground’ dimension is facing a real crisis (Cross and Katz, Reference Cross and Katz2013: 65). The most significant aspect of ‘party decline’ could be interpreted as a crisis of participation within parties.

The two main organizational responses that parties have recently adopted to cope with new challenges (e.g. anti-party attitudes, eroding electorates) include the expansion of intra-party democracy and the introduction of new forms of party membership. On the one hand, parties have been prompted to develop new strategies to broaden their boundaries and reach out to non-member supporters. On the other hand, parties have adopted a wide range of internal organizational reforms that, at least formally, give members more say over outcomes. Direct democracy is now used in a diverse range of intra-party decision-making procedures, such as candidate selection, leadership selection, and policy position formulation. These two responses – and party organizational change in general – have triggered significant modifications in the role and behaviour of grassroots activists – namely, party members. The case of the Italian centre-left parties, and particularly the Democratic Party (PD, Partito Democratico), is in this sense particularly relevant, as for over a decade these parties have been reaching out to supporters in order to integrate them into decision-making processes, such as the selection of party leaders and candidates to legislative and executive offices. Primary elections seem to be an adaptive reaction used by Italian parties of the left in order to reactivate the link with their members and outline new relationships with their supporters.

Primary elections are a recurrent theme in the debate about parties and their organizational change (Farrell and Webb, Reference Farrell and Webb2000; Wattemberg and Dalton, 2000) and the personalization of politics (Calise, Reference Calise2000; Pogunkte and Webb, Reference Pogunkte and Webb2007; Blondel and Thiebault, Reference Blondel and Thiebault2010). Literature on party politics generally argues that open primaries for selecting party candidates or leaders – that is, direct elections open to all party voters – represent a further step in the organizational evolution of political parties. In fact, in open primaries, both enrolled members and simple party supporters can vote and mobilize internally, for selecting either the party leader or candidates for elections.

Katz and Mair (Reference Katz and Mair1994, Reference Katz and Mair1995, Reference Katz and Mair2009) argue that parties have progressively and strategically reduced the size of the ‘party on the ground’. The party in public office has taken over the organizational role of mass membership. Political parties find new legitimacy in the participation in government rather than in social integration and encapsulation; the result is a shift in the mobilizing dynamics of intra-party politics. From this perspective, the model of parties as instruments of social integration has been transformed through the adoption of new mobilizing strategies that go beyond the traditional ideological boundaries and that increase parties’ organizational permeability (Sandri and Pauwels, Reference Sandri and Pauwels2010).

Western European parties are nowadays more electorally oriented than in the past. This has led to parties adopting innovative tools for mobilizing new voters and supporters. These new supporters are not as sensitive to the traditional party discourse as loyal, formally enrolled, and ideologically identified members. The growing professionalization of political communication, influenced by political marketing, allows political parties to use mobilizing strategies for bypassing the traditional but costly resource represented by enrolled members. They are hardly needed to convey the party’s political discourse and party manifesto; they are less needed for campaigning and for mobilizing voters. Members end up playing a marginal role. IT and new media resources allow parties to contact a wider public, providing a more general message, which is less focused on belonging and ideological rhetoric, and aims at attracting new electoral support rather than new members (Katz and Mair, Reference Katz and Mair1995).

Parties have replaced their old mass-party structures, which were rooted in grassroots membership and were ideologically distinctive, by greater organizational permeability and by the mobilization of party supporters through new instruments of internal democracy, such as open primaries. Although this new tool for mobilizing voters may attract new quotas of the electoral market, it does not guarantee loyal and consistent electoral support (Dalton and Wattenberg, Reference Dalton and Wattenberg2000; Cross and Katz, Reference Cross and Katz2013), thus affecting negatively the transformation of voters into affiliated members (Raniolo, Reference Raniolo2004, Reference Raniolo2006). Looking at declining membership figures and election turnout, it seems that this new approach to electoral and internal party mobilization is less effective in terms of consistency in the long term (Scarrow, Reference Scarrow2000; van Biezen et al., Reference van Biezen, Mair and Poguntke2011).

A closer look reveals that the high degree of inclusiveness of open primaries is aimed at mobilizing supporters rather than regularly affiliated members. In fact, primary elections promote a new image of political parties – namely, that they are more democratic and transparent and open to all members who are finally admitted into the smoke-filled rooms (Hopkin, Reference Hopkin2001: 344). Yet, the real targets of this mobilizing strategy are voters (Cross and Katz, Reference Cross and Katz2013: 10). Inclusiveness of open primaries means that both members and supporters without any formal affiliation take part in core decision-making processes, such as the selection of candidates and party leaders. This is bound to affect party organization. However, this also means that these two categories of selectors are very different from a political standpoint. Although enrolled members might be easily considered to be strongly involved and interested in internal party life, this is not necessarily true for supporters, who remain external to party structures. Incentives for intra-party mobilization are very different and could lead to different outcomes from the viewpoint of the internal organizational relationship (Sandri and Seddone, Reference Sandri and Seddone2012; Seddone and Venturino, Reference Seddone and Venturino2013b; Sandri et al., Reference Sandri, Seddone and Venturino2015b).

Over the past decades, the diversity in possible types of party membership has increased with the introduction of new participatory opportunities that challenge the very notion of formal party membership; this has widened the possibility for individuals to interact and participate in very different ways with the party (Young, Reference Young2013; Gauja, Reference Gauja2014; Scarrow, Reference Scarrow2014). One of the most prominent new categories is the so-called ‘supporter’ category, which allows the participation of non-members in electoral campaigns, policy development, leadership, and candidate selection. Other parties have opened up their organizational boundaries in order to interact directly with ‘supporters’ during primary elections or online, for example, through Twitter, Facebook, and party websites (Gibson et al., Reference Gibson, Gillan, Greffet, Lee and Ward2013). Italian parties have been quite innovative with regard to new types of internal involvement, as Italian parties have used primaries for more than a decade, and they involve both party supporters and enrolled members in internal decision-making processes. Also, several Italian parties have restructured their organizational setting so that new typologies of members, supporters, volunteers, or online members could be integrated into party activities.

Internally, democratic political organizations provide crucial instruments for political integration by giving opportunities to members and ordinary citizens to influence the choices voters are offered (Scarrow et al., Reference Scarrow, Webb and Farrell2000: 130). When using direct elections that are open to all members, parties do not distinguish anymore between active and passive members (Katz and Mair, Reference Katz and Mair1995: 20; Kenig Reference Kenig2009; Hazan and Rahat, Reference Hazan and Rahat2010). In addition, open primaries only require prior registration or registration at the moment of the vote. Therefore, the unmediated nature of these participatory instruments is thought to be associated with the enhanced individualism of new types of party members’ role and profiles (Russell, Reference Russell2005: 267; Bolleyer, Reference Bolleyer2009: 563). This atomistic conception of party membership is also considered to be related to a weakened distinction between the functions, identity, and role of party affiliates and supporters.

As a consequence, the introduction and diffusion of open primaries weakens the distinction between members, supporters, and ‘external voters’ in terms of activities and power. Parties adopting new intra-party democratic instruments and primary elections in particular have increasingly blurred the member/non-member distinction by inviting supporters and primary voters to join in their activities, such as campaign actions, whether formally enrolled or not. However, the distinction is not (yet) blurred to the point of rendering the different categories of partisan affiliation identical in terms of role, activities, and power. The traditional difference in terms of socio-demographic and ideological profile and party identification among the three groups thus needs to be empirically documented. Also, parties that involve non-members in their internal activities claim organizational openness to be an expression of their democratic values. While openness is unproblematic for the working of internal processes as long as members have little say over party decisions, the involvement of non-members within the party can create tensions, especially if a parallel process of membership empowerment has taken place and/or access to decision making does not require any indication of organizational commitment (Bolleyer, Reference Bolleyer2009; Sandri, Reference Sandri2011). Thus, the research questions explored in this article address the relationship between parties, members, and new forms of party involvement.

First, in order to explore this relationship, the article assesses the differences between the three groups of primary voters – namely, party members, supporters, and external voters – in terms of interest in politics and ideological profiles. Second, the article explores the potential consequences of the variations in partisan affiliation by focusing on voting behaviour and party attachment of three different groups of primary voters. It assesses the overall degree of differentiation between the various internal mobilization strata. We explore the main reasons that have shaped the selectors’ voting choices. We also explore the extent to which voting intentions in general elections are related to feelings of party belonging or party identification, or on the contrary can be better explained by other factors, such as the degree of personalization of politics or electability of the candidate. Thus, we address the following research questions: to what extent do members, supporters, and external voters differ with regard to political attitudes? To what extent do members, supporters, and external voters differ in terms of voting behaviour? More specifically, to what extent does the different profiles of the three groups contribute to explaining their voting motivations in primary elections and voting intentions in general elections?

We explore these issues using an original data set on the profiles, political attitudes, and behaviours of party members, voters, and supporters of the main Italian centre-left party, the PD. In the first part of the article, we examine the political profiles of PD’s enrolled party members and non-enrolled supporters and voters. This exploratory analysis, which aims to identify the main variations among the three groups, will be developed on the basis of survey data collected through the exit polls carried out during November–December 2012 during the centre-left coalition’s primary elections of the candidate to the prime ministerial position. In the second part of the article, we explore the consequences of the differences in the political profiles of the three groups in terms of voting intentions and behaviour. This second step aims to analyse the variations in the relationship that each group develops with the party.

New forms of mobilization, members, supporters, and internal elections in Italy

Among Italian parties, the PD is the only political organization that has adopted primaries not only as an instrument to enhance intra-party democracy but also as its main distinctive feature, transforming internal elections into a symbolic and political identity factor. In fact, primary elections have been organized to select the first leader and to celebrate the foundation of the new party, which was created in 2007 with the merger of the leftist, post-communist party DS (Democrats of the Left) and the centre-left party ‘La Margherita’ (The Daisy). The adoption of such an inclusive procedure of leadership and candidate selection as open primaries intended to symbolize a break with the past so that the traditional mass-party structures could be improved by offering new participatory instruments to both members (thus strengthening the party–members relations) and unaffiliated supporters (namely, loyal voters).

The PD provides a relevant case study because it has been reaching out to supporters since 2005 in order to include them into various types of internal elections to choose the party leader, candidates for national, regional, and local offices, and the candidate to the prime ministerial position. Since 2005, centre-left parties and the PD in particular have organized open primaries for selecting candidates of mayoral and regional elections (including regional governors) in several cities. More than 900 primaries have been organized since 2005 at the local level to choose the mayor candidate in more than 70 cities (Seddone and Venturino, Reference Seddone and Venturino2013a). Moreover, the PD has also organized open primary elections to select the party leader in 2007, 2009, and 2013, the prime ministerial candidate in 2005 and 2012, and candidates at national elections in 2012.

Other political organizations such as SEL (‘Sinistra Ecologia e Libertà/Left Ecology and Freedom’, a small leftist party) also use open primaries to select candidates (in fact, SEL has organized several open primaries together with PD to select the common candidates for their electoral coalitions at the local, regional, or national level); thus, they formally recognize different forms of partisan affiliation. However, for the PD, primary elections do not constitute merely a new instrument for mobilizing electoral support during electoral campaigns. The adoption of primaries has significantly affected the party’s organizational features. The peculiar relationship between enrolled members and non-enrolled supporters constitutes the main organizational specificity of the PD. As is the case for many other Western European parties, enrolled members of the PD have been consistently declining over time. The two founder parties of the PD, the DS and the Margherita, already experienced a significant decline in their membership in the years that led to the creation of the new party; also, recent data confirm that the PD has experienced a similar decline. The broadening of the PD’s organizational boundaries was thus thought to provide a significant boost to partisan engagement.

In fact, the number of PD’s enrolled members has been declining subsequent to the mobilization ‘momentum’ created by the foundation of the party and the first primary elections held in 2007. Still, the overall membership is well beyond half a million affiliates. In 2008, the PD counted 820,000 officially enrolled members, whereas the official data released by party central organs declared only 618,768 members in 2010. Although membership figures briefly increased to a total of 763,783 in 2011, they dropped back to 500,163 in 2013. This means that the PD ‘party on the ground’ represents around 6% of the party’s voters, which corresponds with the European average in terms of encapsulation ratio (Scarrow and Gezgor, Reference Scarrow and Gezgor2010; van Biezen et al., Reference van Biezen, Mair and Poguntke2011). However, the rapid decline of aggregate membership in the year following the foundation (2007) is rather striking, even if we take into account that in 2009 and 2013 primary elections selected only a leader. It seems that several members joined because they were attracted by this type of event (instant members) and then left shortly after (Rahat and Hazan, Reference Rahat and Hazan2007). Furthermore, this downward trend was already characterizing the membership structures of the two parties that merged in 2007 – namely, DS and Margherita (see Tables A1 and A2).

Since its creation, the PD has introduced a new type of involvement in the party – namely, the supporter: a voter who is not formally enrolled in the party but who can participate in some internal activities, such as canvassing, campaigning, and selecting candidates for elections through primaries. The PD is thus characterized by a relatively broad variety of partisan affiliation. The distinction between formally enrolled members and non-enrolled party supporters is enshrined in the PD’s statutes and internal rules. The first article of the party constitution states that the internal life is based around two different units: members and (registered) supporters. Italian citizensFootnote ¹ and citizens of an EU country residing in Italy as well as those from a third country with a valid working visa who accept to be integrated in the ‘public register of PD’s supporters’ (articles 2 and 3 of the party statutes) have the right to participate in the internal life of the party, including decision-making processes. They shall be older than 16, and they shall formally declare to accept the PD’s ideological stance and programme and pledge to support it.

The PD is not the only Italian party entailing different types of partisan affiliation. The organizational permeability of Italian parties seems relatively high. Three other parties (LN, PDL, and IDV – Italia dei Valori) also recognize the category of ‘party supporter/sympathizer’ in their internal statutes. However, in the case of LN, this category simply represents the first step in the complex procedure of becoming a member, whereas in the cases of IDV and PDL this formal category does not actually correspond to a specific role or function within the party (Sandri et al., Reference Sandri, Seddone and Bulli2015a). Conversely, in the case of the PD, party supporters are integrated within the internal life of the party.

PD’s ‘registered supporters’ are in fact primary voters who accept to be listed in the party register when casting their vote for the party leader or candidates. They have a wide array of formal rights: not only can they participate in primary elections, either for the leader, the prime ministerial candidate, the gubernatorial candidate, or the mayoral candidate, but they can also participate in internal policy forums and referenda and can be informed about every aspect of the party’s internal life (article 1.3 of the statute).Footnote ² They can also participate in local branch meetings (but without voting rights). Nonetheless, they have limited obligations and crucial membership rights; for example, selecting delegates to party congress is still reserved for formally enrolled members. The turnout at nationwide primary elections varies: between 4,300,000 voters registered in 2005 (selection of the prime ministerial candidate of the centre-left coalition) and around 3,100,000 voters registered in 2012 at the first round of the elections for the same position (around 2,800,000 voters participated in the second round). The number of voters who participated in the primary elections for selecting the national party leader was 3,550,000 in 2007, 3,102,000 in 2009, and 2,815,000 in 2013. The introduction of ‘party supporter’ category represents a crucial step in the process of internal democratization and affects the working of PD’s internal processes.

However, there are substantial differences in costs and benefits regarding access to membership compared with primary election eligibility.Footnote ³ In terms of costs, the difference is both monetary and procedural. Supporters pay only 2 Euros when they register (usually on the primary election day), whereas members must pay fees varying from 15 to 50 Euros, depending on age and income, in order to obtain their membership card. PD allows online recruitment of members; however, at some point even the online recruits need to enrol at the local party branch. When joining, members need to not only support the PD’s manifesto and pledge to vote for the party in the same manner as registered supporters, but they also have an obligation to respect the statutes and become involved in the party. Also, the distinction between the two categories is at the same time formal and behavioural. Primaries represent the main venue in which both members and supporters are involved at the same time and share the same rights and obligations. During primary election procedures, members and supporters thus share the same collective and selective incentives even though their respective overall involvement in party activities differ. Registered supporters enjoy similar rights as party members. Yet, only members can elect their delegates to party congress and vote in local sections meetings.

Given that significant differences still exist in terms of the benefits and costs of obtaining the two positions, we expect members of the two groups to be different in terms of demography, political profile, voting behaviour, and motivations. More specifically, in fourth section we analyse the differences in terms of political profiles and attitudes and in fifth section we analyse the differences in terms of voting behaviour and linkage with the party of the different categories of partisan affiliation. Also, we assess whether and to what extent these differences could be explained by the specific profiles of each group. In this study, the profiles and attitudes of the different categories of partisan affiliation constitute the main independent variable, whereas members, voters, and supporters’ voting behaviour and motivations represent our dependent variables.

Data, methods, and the case study

The main goal of this article is to assess the differences in terms of political attitudes and behaviours between party members and supporters. We explore the variation in our dependent variable (voting motivations in primaries) on the basis of an original data set. The data have been collected by the Candidate & Leader Selection research group of the Italian Association of Political Science on the basis of the 2012 exit poll survey conducted during the centre-left coalition’s primary elections. The electoral coalition called ‘Italia Bene Comune’ (Italy Common Good) was formed by PD, SEL, CD-Democratic Centre (a small centre party), and the Italian Socialist Party. Primaries were organized in November and December 2012 to select the leader of the coalition who would stand as common candidate for the office of Prime Minister in the subsequent general election, which took place on 24–25 February 2013. Five candidates ran in the primaries: Pier Luigi Bersani, Matteo Renzi, Nichi Vendola, Laura Puppato, and Bruno Tabacci. Bersani won 61% of the votes, defeating Matteo Renzi in the run-off.

Although the results of the internal election were not surprising, this ballot was quite innovative from an organizational point of view with regard to the voting rules and the candidates. Due to relevant power shifts in the dominant coalition within the main party of the electoral cartel – namely, the PD – the rules concerning the voting system and the registration of voters were changed a few months before the election. The party leadership adopted a two-round, run-off voting system as well as a new rule of previous (or election day) compulsory registration of primary voters for both rounds. Before 2012, the two-round system has been used only in a handful of local primaries of mayoral candidates.

More importantly, historically, primary voters were not required to pre-register in the register of PD primary selectors (and to formally pledge to vote for the centre-left coalition in the 2013 general elections) in order to be allowed to vote. This new rule has been specifically designed to monitor internal participation and avoid crossover voting – namely, the participation of voters in centre-left primaries who are affiliated or loyal to other parties (Fracchiolla and Venturino, Reference Fracchiolla and Venturino2013). The fear of crossover voting was related to the exceptional candidacy of Matteo Renzi, the main competitor to the front runner and party leader, Pier Luigi Bersani.Footnote ⁴ In fact, Renzi focused his primary election campaign on party renewal and broadening the party’s societal reach, by trying to win the electoral support of centrist (or traditionally non-PD) voters.

The data collection process at the individual level through exit poll surveys presents some major methodological challenges, especially in the case of primary elections. The main difficulty concerns the elaboration of a probability sample that could be considered representative of the target population. In the case of open primaries, it is rather complex to define the sampling frame in order to elaborate a simple or stratified random sample, given that the target population corresponds to the whole coalition electorate (and more generally the population residing in the country aged 16 and over). Even the formal requirement of pre-registration that was introduced in 2012 could not prevent the participation to the ballot of voters or members of parties not belonging to the centre-left coalition. Thus, using the party’s electorate in the previous election as the target population was not the best option, given that voters of other parties could also take part in the ballot (at least theoretically).

For all these reasons, the Candidate & Leader Selection research group has chosen to use the whole resident population aged 16 and over as the target population. This was applied by elaborating a non-probability sample, taking into account population size (at regional and then municipal level) and past voting history, of 102 voting precincts nationwide.Footnote ⁵ The sample of 3500 interviews has thus been elaborated by allocating a fixed number of interviews per polling station on the basis of PD’s results in previous elections in the given constituency, including both stations where the party won and others in which it was defeatedFootnote ⁶ (see Table A3). Primary voters were interviewed as they exited the voting station. At each sampling location, an interviewer approached voters as he or she exited the polling place.Footnote ⁷ Participation was voluntary and anonymous.Footnote ⁸ The high number of interviews carried out, the quality of the data collected (in terms of prediction of the actual results of the primaries), and the similarity of the socio-demographic characteristics of previous exit poll samples allow us to use probabilistic statistics.

To analyse the variations in the forms of partisan affiliation and their consequences, we have classified primary voters into three different categories (Table 1). The following coding has been used to place respondents into three categories of participants:

1. 1. The ‘external voters’: they are primary voters not formally enrolled in the party and they did not vote for the party in previous general elections. They are citizens with varying degrees of party identification, who are willing to mobilize politically due to the low costs of participation of primary elections. Presumably, there are some loyal voters for whom this primary represents the first time they have participated in a party activity.

2. 2. The ‘supporters’: they are primary voters not formally enrolled in the party but are loyal party voters, in the sense that they declare to have voted for the party in previous general elections. Supporters include those who have consistently voted for the party and occasionally participate in other intra-party activities.

3. 3. The ‘members’: they are primary voters who are also formally enrolled members and loyal voters (in the sense that they declare to have voted for the party in previous general elections). They represent the category of partisan affiliation most involved in intra-party activities. There are some members who are not active in intra-party activities as the distinction is based on formal criteria (formal membership, previously voting for the party) rather than on behavioural criteria (intensity of participation in internal activities).

Table 1 Typologies of PD primary voters

The three categories are thus ordered on the basis of their respective level of party attachment. The fourth category is that of ‘disloyal members’: although they are formally affiliated to the party, they voted for other parties during previous elections. Due to the limited numbers of cases that fall into this last category, they are not taken into account in our analysis.

On the basis of the three categories of participation, we explored our two research questions through: (a) descriptive analyses of the differences in the profiles and political attitudes of the three groups and (b) inferential analyses of the extent to which voting motivations in primary elections and voting intentions in general elections can be explained by the specific profiles of each group. Fourth section presents the descriptive analyses on the political profiles and attitudes of the three groups by exploring their level of reported interest in politics and ideological self-positioning over a left-right scale (Table 2). To further study the differences in primary voters’ profiles (and particularly their relationship with the party), we also provide in Tables 3 and 4 a descriptive account of (a) the influence of the primary electoral campaign on their voting choice, (b) the reported main voting motivations, and (c) their voting intentions in general elections.

Table 2 The profiles of primary voters (percentages)

*χ ² test, p<0.01.

Table 3 The reported motivations for the vote choice in primary elections (percentages)

*χ ² test, p<0.01.

Table 4 The voting intentions of primary voters at the subsequent general election (percentages)

*χ ² test, p<0.01.

These preliminary analyses contribute to clarifying the nature of the relationship between the three groups of primary voters and their party, and provide useful elements for interpreting the results of the multivariate analysis. Given that literature on new types of membership still remains at an embryonic stage (Fisher et al., Reference Fisher, Fieldhouse and Cutts2014; Gauja, Reference Gauja2014; Scarrow, Reference Scarrow2014), it is rather challenging to formulate theoretically grounded and deductive hypotheses. We provide more limited and empirically grounded hypotheses based on the results of the descriptive analyses. The main argument guiding our analyses is that voting motivations in primary elections are explained by the different relationship that the three groups have with the party:

In order to explore the impact of the type of relationship with the party on voting motivations, we recoded the dependent variable into a dichotomous one. The variable measuring the reported motivations for the voting choice in the primary is presented in Table 3. We recoded the 12 response categories for explaining the choice for a given candidate in primary elections by collapsing them into two categories: the first one merges all the response categories dealing with ‘strategic’ motivations for choosing a candidate and the second one merges all the response categories dealing with voting motivations based on the ‘feeling of belonging’ to the party. The recoded dichotomous variable thus distinguishes between ‘strategic motivations’ (0) and ‘feeling of belonging motivations’ (1). The ‘strategic motivations’ category contains the following response categories: (a) she/he represents the renewal of party elites; (b) she/he is the most fit to lead Italy; (c) I like her/his political programme; (d) she/he is the most fit to win against the centre-right coalition; (e) she/he is the least bad choice; (f) I like the candidate’s personal profile; (g) she/he has been recommended to me by friends/family. The ‘feeling of belonging motivations’ category contains the following response categories: (a) she/he represents my ideological values; (b) she/he is formally supported by my party; (c) I like the outcomes of her/his past political activities; (d) I like primaries and participation in general.

Moreover, besides the main independent variable (i.e. type of partisan affiliation as summarized by the three groups presented in Table 1), we also considered the degree of interest in politics,Footnote ⁹ the ideological self-placement on the right-left spectrum,Footnote ¹⁰ the voting intentions in general elections (recoded as a dummy: intention to vote or not for the PD), and the perception of the electability of each candidate (measured here through a proxy – namely, which candidate the respondent thinks would win the primary ballot). Results are presented in Table 5.

Table 5 Explaining voting motivations

Logistic regression. Dependent variable: voting motivation (0=strategic motivation; 1=feeling of belonging motivation; SEs in brackets). Reference categories are relationship with the party: members; interest in politics: high; ideological profile: right; predicted winner: Renzi; voting intentions for PD 2013: yes.

***p<0.01, **p<0.05.

Different types of membership and profiles

In this section, we explore empirically whether significant differences exist between members, voters, and supporters with regard to their political profile and their political behaviour in primary elections. Primary elections trigger the development of new forms of political mobilization within parties by opening up internal decision making to those citizens who are not interested in making a strong commitment through formal party enrolment but are willing to mobilize politically (Sandri et al., Reference Sandri, Seddone and Venturino2015b).

Table 2 presents data on the political profiles of the three different groups – namely, party members, supporters, and external voters.Footnote ¹¹ Not surprisingly, the level of interest in politics is higher among members (91.9% are interested or strongly interested in politics) than among supporters (85.3%). Moreover, the test of significance shows that the differences between the two groups are highly significant.Footnote ¹²

We also explored the self-placement of the respondents on the right-left spectrum. The χ ² tests also have substantive implications concerning the differences in terms of ideological positioning. They show that the three groups are significantly different from each other with regard to political attitudes. Given the declining levels of party identification in Italy and the fact that voting choices are nowadays less based on ideological orientations and cultural belonging, it is not surprising to see that in terms of ideological positioning there are significant differences between the three categories. External voters are more ideologically moderate than members and supporters. Primary elections attract selectors from different ideological traditions and do not always share the same political orientations of the party organizing the ballot. In terms of internal mobilization, the involvement of citizens less rooted in the centre-left subculture could represent an electoral added value because this means that the party reaches out to new voters. Nevertheless, the question of the relationship with the party of these ‘external voters’ once primaries are over remains empirically unexplored. The data reported in Tables 3 and 4 contribute to answering this question.

Table 3 summarizes the reported motivations for the respondents’ voting choice in primary elections and the perceived relevance of the electoral campaign. Given the overall high level of interest in politics detailed in Table 2, it is not surprising that the electoral campaign only marginally affected the voting choices of most respondents. The influence of the campaign is even less relevant among members and supporters. In terms of self-reported reasons for choosing a chief executive candidate, members are more interested in their capacity of leading the country if elected (26.7% of members reported this as the main reason for their voting choice), whereas for external voters it is more important that the chosen candidate represent political and party renewal (23.3% of the ‘external voters’ reported this as the main reason for their voting choice). Considering that only 7.5% of members declared that they took into account political and party renewal when elaborating their voting choice, this seems to empirically support the expected organizational distance between external voters and the party. External voters more often select their candidate based on their own ideological values (19.7% said so) and on the candidates’ political programmes rather than because she/he is supported by party elites. In this case too, the χ ² tests show that the three groups are significantly different from each other with regard to voting motivations.

Primary voters’ political profiles vary significantly on the basis of their relationship with the party. Members are ideologically closer to the social-democratic party organizing the ballot, and their profiles correspond to a traditional pattern of partisan mobilization based on the sense of belonging and party internal discipline. Supporters are less involved in intra-party ordinary activities but are highly interested in politics and in primary elections as an instrument of electoral mobilization. The most interesting category is represented by external voters, which remain outside the party’s societal reach and organizational boundaries. They are quite distinctive both in terms of ideological positioning and in terms of voting motivations. The differences between the three groups are even greater when looking at the voting intentions in subsequent general elections (Table 4). Besides the obvious loyalty displayed by members (93.7% of them would vote for the PD), Table 4 shows that supporters and external voters are similarly undecided and may choose not to vote for the party in general elections. Data reported in Table 4 (and the results of the related χ ² tests) clearly show that the variation in voting intentions in general elections is associated with the different relationships that the three groups have with the party.

These data (and the results of the χ ² tests reported in Tables 3 and 4) show that the political profiles of the three categories of primary voters differ significantly: the first two categories are more similar, whereas the third is clearly distinct. This could lead to the identification of three different attitudes towards political mobilization within parties: members correspond to traditional forms of party membership, whereas supporters’ attitudes and profiles seem to suggest the emergence of new forms of internal mobilization closely linked to the electoral dimension of political organizations. In terms of socio-demographicFootnote ¹³ and political profiles, they are closer to members than to external voters; however, their occasional and less-intensive involvement in intra-party life highlights a pattern of cognitive political mobilization. Supporters are highly interested in politics and mobilize in low cost activities, such as participating in internal elections (e.g. campaigning), as they are able to contribute to crucial party decisions without having to make any formal commitments. The last type of internal mobilization is represented by external voters. Their profiles and political attitudes are rather distinctive; they only marginally engage in intra-party activities. They are interested in primary elections as an opportunity to participate in politics, but they do not develop any links with the organizing party. This is clearly shown by the reported voting intentions and the rationales of the voting choices for this category of respondents.

Exploration of the survey data concerning the second round of voting leads to similar results to those presented above.Footnote ¹⁴ Given that the political profiles of party members and supporters are relatively similar, and primary elections offer rights and power to voters and supporters, other than to affiliated members, it is relevant to explore how the latter would or plan to behave in such inclusive decision-making procedures as primaries, which distribute collective and selective incentives to this distinctive mobilization group regardless of their real involvement in party life or general elections.

Voting motivations and relationship with the party

In this section we explore the consequences of the differences in political profiles and political attitudes of the three groups in terms of voting behaviour. We have assessed that there are significant differences between the three groups not only in terms of political profiles and attitudes but also in terms of voting motivations and intentions in general elections. This second step in the examination of the different types of party membership of the PD analyses the variations in the relationship that each group develops with the party. In particular, we develop multivariate analyses for assessing the extent to which voting motivations in primary elections can be explained by the specific profiles of each group. Following the two hypotheses formulated in third section, we examine the extent to which their respective voting motivations in primary elections are related to feelings of party belonging or party identification, or to other factors, such as the electability of the candidate.

More specifically, we perform a logistic regression on the impact of a set of independent variables (the main independent variable is represented by the type of partisan affiliation, plus a set of control variables) on the type of motivations that determine the voting choice in primary elections; consequently, this is taken as the main dependent variable for measuring the political consequences of the different profiles of members and supporters.Footnote ¹⁵

The results of the logistic regression are reported in Table 5. On the basis of the log-likelihood statistic, the model provides a satisfactory fit to the data. In this analysis, the key dependent variable is constituted by the reported motivation for the respondents’ voting choices in the primary election. The variable distinguishes between ‘strategic motivations’, mainly related to the electability of each candidate or to their political programme, and ‘feeling of belonging motivations’, linked to party or candidate identification. The key independent variable is a categorical variable for the type of partisan affiliation, broken down into the three categories discussed in third section: enrolled member, supporter, external voter; they are ordered according to the intensity of attachment to the party (ranging from the lower category – ‘external voter’ – to the higher – ‘member’).

The other control variables included in the model are those describing their political profile – namely, the degree of interest in politics and the ideological self-positioning. We also included a control variable that measures the perceived electability of the candidates. This is assessed by using a proxy variable (predicted winner). In the questionnaire, we included an item asking the respondents to state which candidate they thought would win the primary ballot (1=Bersani; 2=Tabacci; 3=Puppato; 4=Vendola; 5=Renzi). Finally, the model also includes a variable measuring the voting intentions in the upcoming general elections (1=will vote for PD; 0=will not vote for PD).

An overview of the results suggests that the different degrees of attachment with the party of the respondents determine their voting motivations in primary elections. We can see that the logit coefficients associated with the type of membership are significant and that, taking ‘members’ as a reference category, a change in our independent variable considerably affects our dependent variable. Being a supporter decreases the odds of voting following ‘feeling of belonging’ motivations by a factor of 0.7, whereas external voters are 0.8 times less likely to vote according to party identification. This means that members are more likely to be motivated by feelings of belonging to the party than are supporters and external voters. Although this is unsurprising, what is rather interesting is that the difference in the odds ratios associated with supporters and external voters is quite small.

However, if the size of the effect of our other independent variables on voting motivations is remarkable, the coefficients of interest in politics and ideological positioning are not significant. Even so, they are quite high and show, in particular, that ideology positioning matters: leftist and centre-left selectors are 1.7 and 1.2 times, respectively, more likely to be motivated by feelings of belonging to the party than strategic considerations. A variation of one unit in the degree of interest in politics (from ‘high’ to ‘average’) decreases the odds of being motivated by feelings of belonging to the party by a factor of 0.98. The size of the effect on our dependent variable of the last control variable – namely, the perceived electability of the candidates – is also quite important. Selectors who believe that the more ideologically or party labelled candidates will win the primary ballot (such as Vendola or Bersani) are more likely to be motivated by feelings of belonging to their party than by strategic incentives. We can see that the odds of scoring 1 (feeling of belonging motivations) increase by a factor of 1.8 when the respondents believe that Bersani, the party leader (and former communist), will win the election, compared with those who predict the victory of the centrist newcomer, Renzi. Predicting the victory of the SEL leader, Vendola, increases the odds of voting following ‘feeling of belonging’ motivations by a factor of 3.0.

The perception of the electability of the candidate has a significant impact on respondents’ voting motivations. This result may be related to the high level of competitiveness of the 2012 primary elections, which counted a relatively high number of candidates and small differences in the shares of votes of the first two candidates, especially in the first round. Moreover, the primary electoral campaign has been quite divisive and the competition among the candidates has been based on intergenerational issues and party elites renewal rather than ideological conflict (albeit Renzi openly tried to mobilize support from centre and even centre-right voters). Finally, and unsurprisingly, the multivariate analysis shows that a change in one unit in the voting intentions for the PD in general elections increases the odds of voting following ‘feeling of belonging’ motivations by a factor of 1.3.

All in all, based on the results reported in Table 5, we can see that the three types of partisan attachment – namely, members, supporters, and voters – are significantly differentiated with regard to the type of motivations for voting choices in primary elections. This is particularly relevant from our perspective, and – not surprisingly – members are more likely to be motivated by their tighter relationship with the party when choosing their candidate, whereas less involved categories, such as supporters and external voters, are mainly driven by strategic considerations. This seems to suggest that our data support the first hypothesis formulated in third section. We also argued (Hypothesis 2) that the voting motivations of supporters and voters are more ‘strategic’, and this too appears to be supported by empirical evidence (with a slight but significant increase in the strength of the effect of partisan affiliation on the likelihood of following strategic motivations when the independent variable changes from supporters to external voters).