Varieties of capital and gender party office in Italy

This article advances a new approach based on ‘varieties of capital’ to explain gendered upward mobility in political parties. Research on gender political advancement unduly neglects women delegates to national party congresses. Our work seeks to redress the imbalance by drawing on data gathered from 5,122 questionnaires issued to national party delegates at 20 national conventions that took place between 2004 and 2013 in Italy. To analyse the data we develop a new framework based on ‘varieties of capital’. Our approach builds on Bourdieu's three types of personal capital – economic, social and cultural – and interprets the findings borrowing analytical tools from recent feminist institutional theory, especially the concept of homosocial capital. Comparisons of male and female party delegates in terms of background and their political trajectories reveal the persistence of an uneven playing field, with gendered hierarchies in Italian political parties confirming an international pattern.


Introduction
In Italy, concern for gender political advancement has recently been addressed by a small but growing literature (Belluati, Piccio and Sampugnaro 2020;Guadagnini 2005; Palici di Suni 2012; Pansardi and Vercesi 2017;Papavero and Zucchini 2018). This literature investigates gender legislative behaviour as well as gender equality in Italian parliamentary representation. However, gender ascendant career within party organisations in Italy has somehow been neglected, creating a situation in which we have a gap in our knowledge about career paths. This is regrettable, since modern party organisations are undergoing dramatic changes. Their structure has become more hierarchical and connected to elites in allied professions ('cadre parties') (Bardi, Bartolini and Trechsel 2014). While party resources are becoming more and more concentrated at the top, party membership is increasingly multi-faceted, multi-speed and requires multi-stream activation (Achury et al. 2020;Power and Dommett 2020). The increasing concentration of power at the top may adversely affect women's opportunity structure in light of the traditional female under-representation in the party's upper echelons.
Political parties are still the central mechanism by which the democratic processes of delegation and accountability work in practice (Müller 2000, 309), so that party change raises issues regarding gender representation and the quality of democracy.
Bourdieu's view, social actors have different endowments of economic, social and cultural capital that determine their position. For Bourdieu, the most significant types of personal capital to acquire are 'economic ', 'cultural' and 'social'. In his work the meaning of economic capital is straightforwardit simply means money. Economic capital offers several advantages, including access to more exclusive forms of occupation. Cultural capital is more complex. It accumulates through a mix of formal and informal education. Education provides the cognitive skills necessary for acquiring, mastering and controlling information vital for evaluating executive and policy options. Social capital is different. It depends on skill in establishing relations with others in social networks. 1 These are 'more or less institutionalised relationships of mutual acquaintance and recognition' (Bourdieu 1986, 21). By exploring the distribution of Bourdieu's personal capitals, scholars reach conclusions regarding power positions of party activists and leaders. Though innovative and interesting, this research neglects gender upward mobility within political parties.
To explore the gender distribution of party office, we follow the lead of a number of authors who argue for positive engagement between Bourdieu's social theory and gender studies (Krais 2006;McCall 1992;Reddy, Sharma and Jha 2019). In Bourdieu's framework, gender is not an analytical category or personal capital (Huppatz and Goodwin 2013); gender is considered a 'secondary' constituent of social divisions. It is therefore liable to the criticism that this 'secondary dimension' reproduces simplistic sexist dichotomies of male domination and female subordination (McLeod 2005). McCall (1992), however, warns against a superficial reading of Bourdieu's social theory. She clarifies that Bourdieu 'variously refers to secondary criteria as "hidden", "unofficial", and "real"'. She explains that a 'secondary criterion', though 'unofficial', is pervasive in social space. In this way Bourdieu exposes how real principles of selection and exclusion are hidden, justifying the usefulness of this analysis to gender studies (McCall 1992, 24). In Bourdieu's social theory, 'secondary' refers to a mediating dimension of position, but then he fails to capture the logical extension of this dimension in political advancement (McCall 1992, 842).
A focus on this mediating dimension is central to recent work on the gender dynamics of political recruitment (Bjarnegård and Kenny 2017;Piscopo and Kenny 2020;Waylen 2017). This new wave of research offers a nuanced understanding of how gender shapes upward mobility by shedding light on informal institutions and hidden practices. Waylen's edited collection (2017) addresses informal norms and codes that sustain gender bias and discrimination, indicating that removing formal obstacles to gender political advancement is insufficient. These works tackle the conceptual and methodological challenge of examining the intangible and implicit yet resilient and tenacious nature of informal institutions. They disentangle informal and informality rules to examine gender inequalities in executive bodies (Francheschet 2017). They suggest that the dynamics of gendered power inequalities are ingrained in many dimensions of political institutions, from party office to ministerial appointments, and cut across different political systems.
Most notably, these scholars shift the research agenda away from explaining gender gaps in ambition, resources and motivation towards understanding the informal institutions that reinforce male dominance. They argue for an 'informal' turn in institutional analysis that addresses the discriminatory effect of gendered norms which pervade the candidate selection process (Bjarnegård and Kenny 2017, 203). These authors propose a new research agenda for the field of 'gender and political recruitment' centred on informal institutions.
Though early feminist work acknowledged the importance of formal and informal institutions, nevertheless the dominant framework to explore party recruitment has been the supply and demand metaphor developed by Norris and Lovenduski (1995). This first generation of research emphasised obstacles to the recruitment of women at individual level, including resources, motivation and ambition that create a bias against women (Fox and Lawless 2011).
Doubts about this emphasis on supply-side factors in accounting for women descriptive underrepresentation are raised by the second generation of researchers. Evidence suggests that individual characteristics, such as political ambition, cannot be explained by the psychological traits of the candidate (Allen and Cutts 2020). Women don't 'lean in' like men, not because of their lack of ambition or motivation or resources; rather, female under-representation stems from lack of demand by party gatekeepers (Kittilson 2013). Demand-side factors involve the political elites' concept of the 'ideal candidate'. For example, Niven (1998) finds that party elites describe the 'ideal candidate' with stereotypical masculine characteristics.
However, Krook (2010b) contends that the metaphor of supply and demand cannot explain the pervasiveness of female under-representation in political recruitment in every country of the world. Stalled progress in explaining gender political recruitment is not the result of a lopsided focus on one set of sources and therefore cannot readily be improved simply by theorising 'the other side'. Since there are more women in the world than men, what accounts for the distortion of female under-representation? Supply and demand dynamics do not operate in a vacuum but unfold within political institutions where formal and informal rules and procedures may generate distortions in political recruitment.
Building on the work that has been done using the supply and demand framework, the third generation of feminist scholars moves beyond this approach to focus on mechanisms that interact with individual supply and demand resources. It does so by combining gender as an analytical category with the insights of institutionalism, exploring the potential gender distortions rooted in formal and informal institutions.
Many factors justify an emphasis on institutions. Institutions are weapons of coercion and redistribution but they also help mitigate problems of collective action; institutions resist redesign, ultimately because they incorporate a vested interest in their standard operating procedures. The term institution is used to refer to informal codes of behaviour, written contracts or complex organisations (Lowndes 1996). Institutions are associated with conventions and customs embedded in the organisational structure that provide frames of meaning guiding human action. Since actions are inserted in institutional structures of norms, expectations and traditions (Marche and Olsen 1989, 23), institutions should not be left out of any serious analysis of gender political recruitment.
For this reason, feminist institutional theory moves beyond the supply and demand metaphor. It is not just a question of supply and demand: matters may be more complex. For instance, formal and informal mechanisms may reinforce the power of dominant players in politics, traditionally men (Cheng and Tavits 2011). Josefsson (2020) finds that male party leaders believe women are less trustworthy and loyal than men, which makes them feel threatened by female candidates. Thus, political recruitment is gendered because the constructions of masculinity and femininity are intertwined in the 'logic' of the recruitment process, shaping 'ways of valuing things' (Bjarnegård and Kenny 2017, 209).
By the same token, O'Brien (2015) demonstrates that female candidates play by a different set of rules than their male counterparts. For example, male leaders can be praised for appointing women to their cabinets, while female leaders appointing 'too many' female ministers may not be interpreted as progressive, but rather accused of favouritism or identity politics . Along the same line, Müller-Rommel and Vercesi (2017) explore gender differences in the paths to prime ministerial posts. They find that female prime ministers have more political experience in both parliament and cabinet than their male counterparts, but a shorter duration in office once they become prime minister, indicating that their political resources are less valued than men's. This discussion is lively and useful because it goes to the heart of key theoretical and empirical questions concerned with gender political recruitment: Is there a gender logic in political recruitment? Are institutions gender-neutral? Gender-neutral formal selection criteria may also have unintended gendered consequences. Bjarnegård and Zetterberg (2019) carry out a cross-party comparative analysis that demonstrates how selection criteria may have 'unintended gendered consequences of seemingly gender-neutral formal selection criteria on the representation of men and women ' (p. 325). This third generation of research on gender political recruitment examines the informal 'gendered institutional logic'. 'Gender is understood here as a conservative element of social relations based on perceived (socially constructed and culturally variable) differences between women and men, and primary ways of signifying (and naturalising) relationships of power and hierarchy' (Bjarnegård and Kenny 2017, 209). Gendered institutional logic interacts with individual supply and demand resources with potentially significant gender effects.
This new wave of research points to male homosocial capital as an important informal mechanism that helps understand the complex gender institutional logic of political recruitment (Verge and Claveria 2017;Bjarnegård 2013). Bjarnegård (2013) initially developed the concept of homosocial capital to shed light on political advancement in party organisations. Borrowing from the literature on recruitment processes in other organisations, Bjarnegård defines homosocial capital as an 'uncertainty reduction resource' in an otherwise unpredictable political environment. She argues that similarity between individuals in terms of perceived characteristics is seen to ensure predictable behaviour and, in the long run, generate trust. In Bjarnegård's approach, homosocial capital has two main components: a perceived pragmatic necessity to build links with those having access to important resources, and a more psychological desire to cooperate with individuals whose behaviour can be understood, predicted and trusted. Thus, homosocial capital is an interpersonal mechanism that interacts with individual supply and demand dynamics.
Building on this new wave of feminist institutionalism, we argue that Bourdieu's social theory of personal capital and his emphasis on the 'hidden' nature of principles of candidate selection dovetail with the recent feminist-institutional focus on the interpersonal quality of homosocial capital. This cross-fertilisation may enliven the debate because one camp can fruitfully borrow insights from the other to engage in a deeper treatment of gender political advancement. Granted, this article cannot do justice to the complexities of Bourdieu's social theory nor to those of feminist institutional theory. We limit our analysis to assessing whether differential endowment of personal capitals sustains individual career paths and whether the interpersonal homosocial capital may help understand any distorted outcome.

Unit of analysis, case selection and data
Unit of analysis National party delegates are not just simply party members, but activists with long political militancy. They are pivotal members of the party communities thanks to organisational positions in politics and local associations. This is why they are known as 'middle level elite' (Reif, Cayrol and Niedermayer 1980;Niedermayer 1986). In most West European parties, delegates have traditionally been perceived as the privileged pool from which party leaders and candidates are selected (Baras et al 2015;Bordandini, Di Virgilio and Mulè 2011). Delegates act as a participatory linkage in party organisational chains because they build bridges between the rank-and-file and the upper echelons (Bordandini 2013;Rohrscheneier 1994).
This participatory linkage of party delegates between the 'party on the ground' and the 'party in the central office' is especially important today with the advent of new forms of party affiliation, such as multi-speed membership (Achury et al. 2020;Baras et al. 2015;Gomez et al. 2019;Hooghe and Kölln 2020;Kölln 2016;Scarrow 2015). New forms of party membership make responsiveness by national party delegates all the more important. In the motions submitted at party congresses, these delegates use responsiveness both as a means of appeasement of party activists and to reward valuable intra-party groups (Kaltenegger, Heugl and Müller 2019). Their speeches at party national congresses provide useful information about the relative location of actors' preferences (Greene and Haber 2016). As part of an internal representative and deliberative process, national party delegates articulate different viewpoints over the party's electoral strategy and policy programmes. The party congress is therefore a deliberative body where factional deliberation is filtered through leader mediation and conciliation. Ultimately, party congresses serve as venues for decision-making, shaping policy statements and party manifestos (Ceron and Greene 2019).
In such circumstances, the under-representation of women at national party congresses affects party policies, programmes and priorities. In her book The Politics of Presence, Anne Phillips (1995) argues that different life experiences and personal characteristics of representatives influence their viewpoints and policy priorities. Moreover, the representation of women at national party congresses is important because being nominated or elected a delegate is often the steppingstone to professional politics.

Case selection
In Italy a long line of scholarship addresses the key role of party activists and middle-level elites in promoting and preserving the quality of democracy (Cotta and Verzichelli 2000;Morlino 1991). Historically the two most powerful parties, the Christian Democratic Party and the Communist Party, enjoyed large membership and active middle-level elites. As in most west European democracies, high degrees of party institutionalisation helped these parties exert strong control on society and on the operation of institutions.
In the last ten years many western European democracies have experienced sequences of party system deinstitutionalisation, with the gradual weakening of the stability and predictability of interparty competition within a given country. Italy is an interesting case study because the process of party deinstitutionalisation was sudden and more dramatic. In the early 1990s, the 'Tangentopoli' inquiry marked a sharp turn in Italian politics, propelling the implosion of traditional party organisations. The advent of new parties, the fruit of a sequence of extinctions, new entries, splits and mergers, began in 1993-4 and continued for about 20 years (Bartolini, Chiaramonte and D'Alimonte 2004;Calise 2010;Cotta and Verzichelli 2000;Di Virgilio 2010;Ignazi, Bardi and Massari 2010;Pasquino 2002). This political upheaval set in motion a relentless process of party deinstitutionalisation that is still underway, with unpredictable consequences for electoral outcomes (Chiaramonte and Emanuele 2015).
Moreover, Italy is an interesting case insofar as it is a highly gendered society that reflects the persistence of traditional norms of gender-appropriate behaviour (Sartori, Tuorto and Ghigi 2017). More generally, in parliamentary systems such as Italy, holding party office may be more relevant for candidate selection because ministers typically have broad political party backgrounds (Siaroff 2000). Finally, the availability of a unique dataset on national party delegates spanning about ten years reinforces the meaningfulness of the Italian case.

Data
Although not on an ongoing basis, party delegates have been a tested research subject in the literature on political parties. The first comparative research based on this unit of analysis dates back to the 1970s, namely, the EPPMLE Project (European Political Parties Middle Level Elites). The project involved 12 European countries (Belgium, Denmark, Germany, Spain, France, Ireland, Northern Ireland, Italy, Luxembourg, the Netherlands, Portugal and the United Kingdom) and 68 parties; surveys on party delegates were conducted between 1978 and 1981, mainly with the same research design. The EPPMLE investigation marked a milestone in comparative middle-level party elite research, but unfortunately there has not been a comparative follow-up. Case studies on party delegates are still conducted in single countries, such as France, Italy, Spain, Portugal and Germany, but the datasets have not been harmonised for comparative analyses (Baras et al. 2015;Lisi 2015).
Our dataset is based on the Italian delegates research project, an inter-university research programme co-funded by the Ministry of Education, University, and Research and four universities (Florence, Bologna, Cosenza and Trieste). 2 Data was collected by means of a structured selfadministered questionnaire distributed during national party congresses (except for the Verdi [Green] party, for which a postal survey was organised immediately after their national congress). The questionnaire consisted of two partsa general section for all parties and a second, specific section for party characteristics and the political context in which their congress took place. The questionnaire was distributed to all delegates to the national party congress. Table 1 illustrates the distribution of delegates to the national party congress by party and gender. 3 It shows that 5,122 delegates returned the questionnaire, and that the percentage of respondents varied between a maximum of 48.5% and a minimum of 8%. The share of women respondents was fairly similar to the share of women in the party delegate universe (9 out of 20 political parties provided no information on the total number of their delegates). Interestingly, 59 delegates did not reply to the question regarding their gender.
The data indicate that irrespective of party ideology, there was about one woman to every two male delegates; women delegates are on average 48% fewer than men. As expected, the data revealed higher woman quotas in left-wing parties, which is consistent with the literature that detects more gender 'friendliness' in left-wing parties (Ewig and Marx 2013;Lovenduski and Norris 1993). 4 The parties involved in the surveys differed in size, electoral appeal and geographical distribution of votes. In most cases, delegates were elected by party members, directly or indirectly, except for the Radicals, where any party member can attend the national congress becoming a de facto delegate.

The gender distribution of personal capital
This section traces the distribution of personal capital between male and female party delegates. Broadly speaking, the step from concept to variable is always challenging and liable to criticism, while empirical analysis is often limited by the availability of data. Our study is no exception. Here we followed mainstream operationalisation of personal capital. Economic capital was measured by taking the delegate's profession and productive sector as proxies. These indicators are only apparently gender-neutral, as horizontal and vertical gender segregation affect labour markets in every country.
Cultural capital was operationalised with the variable 'education' of the party delegate and her/ his father (unfortunately no information was gathered for 'mother's education'). Our data did not allow us to differentiate the fields of delegates' qualifications (medicine, engineering, arts etc.) but only the level of education (primary, secondary, university).
Social capital was operationalised with the variable 'membership of social groups and social organisations' and 'position in associations'. These indicators are appropriate because social networks are extremely useful in political recruitment and mobility (Bourdieu 1986). Joining associations fosters social capital because it strengthens social networks. Economic capital: profession Job status may well buttress upward mobility by developing skills, know-how and support networks. It affects the time delegates are able to devote to political activities to strengthen their social networks and political circuits. Our data captures major gender gaps in the quantity and quality of job position (Table 2). About 35% of party delegates were employed in high status jobs, such as professions, management and entrepreneurship, yet there are remarkable gender differences. Women delegates holding executive positions are 8.6 percentage points behind men, though the gap narrows slightly in younger generations. This means that women control fewer economic resources than men. They consequently reach the intermediate level of party office with weaker spill-over from their job status. By contrast, women delegates employed in the public sector were 11.5 percentage points ahead of men, and again this figure drops in younger generations.

Cultural capital: education
Cultural capital is a scarce resource that equips individuals with knowledge, practical skills and a sense of 'the rules of the game' in the educational system. These attributes are recognised and rewarded by institutional gatekeepers and peers (Bourdieu and Passeron 1977). We operationalised cultural capital with the gender distribution of education in party delegates and their fathers (there was no data on mothers' education). Father's education is an indicator of a delegate's cultural background. As mentioned above, education provides the cognitive skills necessary for acquiring and mastering the flow of information required to evaluate executive and policy options. Table 3 indicates that among national party delegates, women were more educated than men. The gender difference is 10.5 percentage points higher for women (50.4% men against 60.9% women), with a similar gap in the younger generation. 5 The figures show that women's cultural background was from families with higher levels of education than families of male delegates, as measured by the father's education. Family cultural gaps disappear in the younger generation, probably because expansion of higher education in the 1970s increased the number of graduates. All in all, these results suggest that in Italy, women reach an intermediate level of party office with higher levels of cultural capital than men.

Social capital: participation in associations
In Bourdieu's view, social capital is related to an individual's social networks. 6 Personal social capital is what individuals gain from their position in organisational networks. It yields better knowledge, power and promotion within and between organisations. Importantly, for Bourdieu social capital is the most effective form of capital for facilitating upward mobility (Bourdieu 1986). Bourdieu's personal capitals are individual resources that help explain why a party activist advances to party delegate. We operationalised the delegate's social capital with 'participation in associations' and 'positions in associations'. Table 4 shows that women delegates participate more in secular volunteer associations than men (57.6% against 45.3%), and are more active in environmental groups (39.2% against 33.8%). The data does not show any significant differences in gender participation in cultural, religious and recreational associations. We measured the gender participation gap by calculating a participation index based on the number of associations the delegate belonged to. The index ranges from 0 to 5: 0 if the delegate is not a member of any association and 5 if s/he is a member of all the associations. A delegate is deemed active in social networks if s/he is a member of more than two associations (high participation >2), because 2 is the median value. The participation index in Table 4 shows that women delegates have higher degrees of involvement (36% against 33%) irrespective of whether the association is professional, recreational, secular or religious. The figures indicate that women join the party middle-level elite endowed with more social capital than men, as measured by the participation index. This result is important because, as mentioned in section 2, social capital is the most effective form of capital in facilitating upward mobility (Bourdieu 1986). Since women delegates are endowed with higher levels of social capital, we expect that they should hold at least as many, if not more, executive positions than men in social associations. However, Table 4 tells a different story: fewer women hold executive positions than men in associations (52.5% against 59.5%), in spite of women's higher participation index.
Overall, our data reveal that fewer women hold executive positions than men (about 11 percentage points less), despite being endowed with more social and cultural capital ( Table 5). The most important implication is that personal capital is valued differently in the ascending party careers of women and men. Hence personal capital is not gender neutral: women's resources are valued less than men's.

Towards understanding party-office gender dynamics
Our results on the distributions of personal capital suggest an uneven political playing field. They confirm the contention that the game is 'organised to serve the aspirations, abilities and opportunities of elite men' (Piscopo and Kenny 2020, 6). They tally with Bourdieu's conception of gender as a 'hidden', unofficial criterion, a mediating dimension that interacts with individual resources and is pervasive in social space.
To help understand the dynamics of uneven playing fields, we borrow from recent advances in feminist institutional theory that point to a key informal mechanism: homosocial capital. One can argue that homosocial capital cannot be seen as a stand-alone factor, simply added to other forms of capital. We agree with this view. Bjarnegård (2013) contends that homosocial capital acts as an Position in associations (manager or ex manager) 52.5 59.5 −7.0 informal interpersonal mechanism that has gendered consequences. Homosocial capital generates expressive resources, such as trust, likeability, predictability and familiarity. In particular, male homosocial capital is a Janus-faced concept. On one hand, it highlights the fact that interpersonal capital needs to be built up before an individual is included in a political network; on the other hand, it points to gendered interpersonal capital: it is predominantly accessible to men and more valuable when built up between men (Bjarnegård 2013). Homosocial capital describes a propensity to consider background characteristics as a measure of predictability. It can be interpreted as an 'information short cut' in which candidates are evaluated on the basis of characteristics associated with their group (Krook 2010a). The unavoidable uncertainty in candidate selection may induce selectors to look for 'background characteristics as a proxy measure of abilities and character' (Norris and Lovenduski 1995, 14).
Yet operationalising the concept of homosocial capital is a challenging task, chiefly because it involves the measurement of expressive resources. Here we examine 'predictability' as a crucial element of homosocial capital. Predictability is a key feature of homosocial capitalit is the desire to cooperate with individuals whose behaviour and attitude can be understood and trusted (Bjarnegård 2013). One way of operationalising predictability is to inspect gender gaps in values. It is reasonable to assume that individuals with similar values form a subset, where behaviour is more predictable than for individuals who espouse different values. Similarity in values can thus be used as a proxy for homosocial capital. In the empirical analysis, we follow the political sociology literature that explores values by looking at post-materialist issues (Inglehart 1990;Inglehart and Welzen 2005;Norris 1999). Table 6 reports similarities and differences between male and female party delegates in attitudes towards post-materialist issues (abortion, gay rights, environment). These figures reveal striking differences in the responses of male and female party delegates to questions regarding civil rights, such as abortion and gay rights. Women appear more progressive than men. They are less inclined to restrict abortion (the gender gap is more than 12 percentage points) and less likely to believe a foetus is an unborn citizen (13 percentage points less than men); women are more liberal with respect to drug use (7 percentage points gender difference) and more favourable to gay rights (about 20 points difference).
However, women's often more progressive attitudes can also be traced beyond the borders of political families. A recent paper on Italian party middle elites indicates that with regard to the ideological positions of parties, women delegates tend to be more post-materialist than men (Bordandini and Mulè 2018). Others find that increasing female descriptive representation in parliament affects the party policy agenda in a number of ways (Espírito-Santo, Freire and Serra-Silver, 2020; Kittilson 2006Kittilson , 2011Greene and O'Brien 2016). It increases the diversity of issues in the election campaign and is associated with more left-leaning manifestos, even when parties' prior ideological positions are taken into account. This gender gap in attitudes may adversely affect women's careers in political parties because it suggests that women are less predictable than men, if the expectation is a male norm. Candidate selection is not a 'hard science' but is strongly influenced by perceptions of trustworthiness, ability and qualifications. These perceptions are shaped by the preferences and opinions of party gatekeepers who are usually men. Scholarly work on gender political advancement indicates biases in the recruitment practices of political elites which lead to selection of fewer female candidates than males (Hinojosa 2012;Kenny 2013;Norris and Lovenduski 1995). These biases may be linked to a psychological desire to cooperate with individuals whose behaviour can be predicted and trusted, i.e. male homosocial capital.
Male homosocial capital can thus help to understand why Italian women delegates hold fewer executive positions, despite enjoying more cultural and social capital than men. Our results lend weight to comparative literature in the field, which indicates that parties are not gender neutral but based on a complex web of formal and informal norms and practices, the impact of which is different for women and men (Kittilson 2013;Lovenduski and Norris 1989;Norris and Lovenduski 1995;Waylen 2017).
Before concluding, it is worth exploring whether the gender gap in post-materialist attitudes is narrowing or broadening among younger Italian delegates. Table 7 shows the differences in attitudes to post-materialist issues between male and female delegates by age (40 years or under, over 40s). The data indicates that the gender gap narrows among younger delegates. For every item, the gender difference in attitudes towards post-materialist issues is systematically smaller. In some respects, these results are reassuring because they suggest that re-gendering party organisations may be possible in the not too distant future.

Conclusion
This article contributes to the literature on gender recruitment for party office in Italy by weaving Bourdieu's theory of personal capital with insights gained from advances in feminist institutional theory that address the role of informal institutional mechanisms. The latter suggest that what matters in recruitment patterns are informal encouragement and gatekeeping practices, 'leaving large room for leeway, interpretations and loopholes "on the ground"' (Bjarnegård and Kenny 2015, 380). Our work lends weight to this claim because it indicates that gender is pervasive in the recruitment process, shaping ways of valuing personal capital, so much so that women with more social and cultural capital than men hold fewer party office positions in Italy. Clearly, party gatekeepers value personal capital differently in women and men, suggesting that party office is a gendered political resource and gendered power dynamics are deeply entrenched in political parties.
The under-representation of women in the higher echelons of political parties raises questions of democratic justice, legitimacy, responsiveness and effectiveness. Since party delegates are involved in the approval of party programmes, the under-representation of women could substantially impact on policies, programmes and priorities of elected governments. This is a legitimate matter of democratic concern.
Like any research design, our study has limitations. It is a single-country study and focuses on party office holding, which could limit the generalisability of our results. Another limit of our dataset is that it does not allow us to explore a dynamic profile of party delegates because it is a snapshot of the delegates' situation. We are therefore cautious with our claims and do not attempt to end the debate on causality between party office, personal capital and informal mechanisms, using a gender perspective. However, we believe that suggesting a theoretical framework that employs 'varieties of capital', both personal and interpersonal, to shed light on gender political advancement, fosters progress in this area.
Our findings highlight the need for more and deeper research on gender political advancement, some of which depends on expanding data capacity beyond what is available today. One important avenue for future research is to tackle the complex measurement issues arising in the empirical analysis of homosocial capital, including identifying relevant indicators. Future research could also benefit from panel studies to inspect the dynamic profile of party delegates.
The varieties of capital approach can be applied in other contexts to further our understanding of gender patterns in the opportunity structure of political parties. It opens up promising research strategies in a growing body of work concerned with gender political advancement and with the quality of democratic representation. Notes 1. Bourdieu's use of the term is distinct from that of others such as Coleman (1990) and Putnam (1993). 2. For detailed information regarding the history of the Italian national party delegate surveys see the website: https://centri.unibo.it/osservatorio-sui-partiti-aldo-di-virgilio/it/ricerca/la-storia. 3. Questionnaires were administered at the following national party conferences: Third PDCI (Party of Italian