Abstract
Throughout the world, the number of women elected to legislatures has risen dramatically. Most of the quantitative research explaining party, district, or national differences has focused on the aggregate rather than the candidate level thereby overlooking women’s access to party ballots. In examining both the election and selection stages, we focus on Spain, a closed-list proportional representation system where parties have tight control over their ballots and the election of candidates is largely a function of rank orders on the ballot. In this South European democracy women’s representation in the national parliament has experienced an incremental track, reaching 39 percent in 2016. Party differences in gender outcomes and policies promoting equal gender representation did not vanish once a legislated quota was introduced in 2007. The empirical analysis builds on an original set of candidate longitudinal data covering nine elections held between 1986 and 2016. Specifically, we test how party and candidate factors differentially affect the selection of men and women to party ballots and their likelihood of getting elected. We show that strategic discrimination against female candidates affects all parties and it happens irrespective of candidates’ political experience, which explains why male overrepresentation has been significantly reduced but not overturned.
Keywords
Introduction
Research on women’s representation in parliaments has examined whether selectorates—voters and/or political parties—discriminate against female candidates and whether such discrimination drives men’s overrepresentation across the globe. Extant research reports very little voter discrimination (Golder et al. 2017; McElroy and Marsh 2010; Murray, Krook, and Opello 2012; Schwindt-Bayer, Malecki, and Crisp 2010). Only in societies where traditional cultural gender norms prevail have electoral systems been found to produce some voter discrimination against women, particularly those that permit intraparty voting, such as single-member plurality systems, single-transferable vote systems and open-list or ordered proportional representation (PR) systems (Valdini 2012).
Conversely, a common factor that stands out in the literature is the way in which political parties favor male candidates when drafting their ballots. This so-called “strategic discrimination” against female candidates (Murray 2008; Verge and Troupel 2011) limits their chances to be elected to office. Studies have shown that each electoral system provides parties with specific opportunities to practice such gendered discrimination (Fortin-Rittberger and Rittberger 2014; Luhiste 2015; Matland and Studlar 2002; Norris 2004). This also applies to PR electoral systems, which have been found to generally elect more women into legislative office than plurality systems (McAllister and Studlar 2002; Norris 1985; Paxton 1997; Reynolds 1999; Thames and Williams 2010), but see, Moser 2001; Roberts, Seawright, and Cyr 2013).
PR systems make it easier for parties to present more gender-balanced tickets, as higher district magnitude allows women to be nominated without deposing male candidates (Matland and Taylor 1997). In such party-centered systems, party leaders cannot “blame” the electorate or women candidates for not campaigning hard enough, as it is they themselves who are clearly accountable for the levels of women’s representation on party ballots and in parliaments. Furthermore, in multiparty systems, losing a few votes might entail fewer seats in parliament, which may set in motion a competition or “contagion” across parties to increase their appeal to female voters by presenting more feminized party lists (Matland and Studlar 1996). Finally, PR systems, especially those with no preference vote option, also allow for a more effective implementation of gender quotas, including placement mandates when they are established (Jones 2009; Schmidt 2009; Schwindt-Bayer 2009).
This notwithstanding, parity remains an elusive goal in most parliaments elected by PR systems, and the question is why. In studying the extent to which party biases shape the election of women as members of parliament (MPs), closed-list proportional representation (CLPR) systems provide an ideal setting, since voters can only choose parties (Carey and Shugart 1995). In this case, women’s chances of being selected and ultimately elected lie exclusively in the hands of party gatekeepers. Thus, to broaden our understanding of the barriers political parties impose on women in CLPR systems, research should not just focus on the aggregate level and thus on the election stage; rather, it is essential to examine the gendered structure of ballots and pay attention to the selection stage.
Recent studies, mainly using process-tracing and single-case studies approaches, have shown that neglecting the candidate level means that the most crucial stage of the ladder of recruitment is overlooked, meaning that the ways in which gendered discrimination operates may go unnoticed (see, Kenny and Verge 2016; Lovenduski 2016). Yet, qualitative approaches do not allow us to measure the scope of such discrimination, that is, the magnitude of its effect. Only a few scholars, focusing on a single election or on a few of them (see Fortin-Rittberger and Rittberger 2014; Górecki and Kukołowicz 2014; Jankowski and Marcinkiewicz 2017; Luhiste 2015), have applied quantitative methods to unravel the contextual and institutional factors (including the type of PR system in use) that shape gender biases at the selection stage. A static picture, though, does not allow us to study how parties’ strategic discrimination evolves over time and the extent to which gender quotas can eliminate or mitigate such practice.
This article takes stock of previous research by addressing remaining gaps in the study of the selection stage under CLPR. Specifically, the goals of the article are twofold: to measure the scope of the strategic discrimination applied against women candidates across parties and to assess its pervasiveness over time. To do so, we focus on Spain, a third-wave democracy which uses CLPR, and where women’s representation in the national parliament has experienced an “incremental track”—that is, a steady increase over time brought on by the use of positive action (Dahlerup and Freidenvall 2005). Party quotas introduced in the late 1980s boosted women’s representation from levels of around 5 percent in the late 1970s to 36 percent in 2004. Despite the introduction of a legislated gender quota in 2007, women’s representation has not made further gains, and has hovered around the same percentage since then, while gender gaps have persisted.
To explore why this is the case, we use a unique data set of candidates who appear on party ballots for elections to the Spanish lower house during the 1986–2016 period, covering nine elections. 1 The combination of a longitudinal party-based and candidate-based approach enables us to simultaneously factor in party characteristics and candidate traits and to assess the impact of both party and legislated quotas in gradually closing the gender gap over time. We focus on how the demand for certain types of candidates and their placement on party ballots shape their likelihood of getting elected, and explore the question of whether individual candidate attributes may mitigate the gender biases underpinning candidate selection processes.
The results of the empirical analyses suggest that strategic discrimination against women candidates in the allocation of list positions affects all parties, irrespective of candidates’ political experience, although left-wing parties have gradually reduced the prevalence of such practices thanks to the effective implementation of party quotas. The results also show that statutory gender quotas with a placement mandate are still vulnerable to strategic discrimination, which explains why male overrepresentation has been significantly eroded but not overturned.
Explaining Women’s Representation in CLPR Systems
Political parties shape a candidate’s chance of winning a seat, and they do so in different ways depending on the electoral system in use. Whereas in plurality systems candidates’ likelihood of getting elected depends on the constituency in which they have been assigned to run (safe vs. hopeless districts), in PR-list systems the chances are determined by rank orders on party ballots, with winnability largely depending on having been placed in a viable position (Norris 2004).
While CLPR increases women’s chances of being selected to viable positions on party tickets and, thus, of being elected to parliament, variation is still found within countries across parties competing under these electoral rules (Luhiste 2015). How is the gender gap sustained in CLPR systems? From a party-based approach, cross-party differences stem fundamentally from dissimilar strategies for equal gender representation. The literature has established that left-wing parties produce higher levels of both women candidates and elected female MPs (Caul 1999; Paxton and Kunovich 2003), mainly thanks to the adoption of equality guarantees that seek to redress men’s overrepresentation (Freidenvall 2013; Kittilson 2006; Lovenduski 1993). Under CLPR, voluntarily adopted gender quotas—codified in party rules or assumed as nonwritten targets or recommendations—engender candidate selection processes by establishing a minimum share of women candidates (Krook, Lovenduski, and Squires 2009, 784).
While the introduction of equality guarantees restricts party gatekeepers’ margin of maneuver when drafting candidate tickets, quota rules may only have a positive effect on the share of women candidates, but not on the gender distribution of viable positions (Dahlerup and Freidenvall 2005; Jones 2009; Murray 2008). As noted by Luhiste (2015, 91), “while being a woman may increase an aspirant’s chance of being selected as a candidate, it may decrease her chances of being granted a viable candidacy.” Female candidates may still be perceived as “political underdogs” by party gatekeepers and be assigned to hopeless places on the ballots (lower positions on the list) or head party ballots in hopeless districts (Gallagher and Marsh 1988), both strongly diminishing women’s chances of getting elected to parliament.
Closing the mismatch between the proportion mandated by quotas and gender outcomes is a gradual process through which the consideration of equal gender representation criteria in list-building is incorporated as a routine-based performance (Inhetveen 1999, 405). Women’s agency is crucial in this institutionalization process in that they hold their parties accountable (Kittilson 2006; Lovenduski and Norris 1993; Verge and Espírito-Santo 2016). The fact that candidate selection processes tend to be rather centralized under CLPR systems (Matland and Studlar 1996) allows female party members to “hold party gatekeepers responsible for their dismal commitment to fielding female candidates” (Luhiste 2015, 95), facilitating the supervision of the effective implementation of quotas to viable list positions, namely those positions in party ballots that are most likely to secure winning a seat. This especially affects “safe seats,” that is, the seats a party can expect to win if it obtains the same results as in the previous election (Simón and Verge 2017).
Provided that party quotas are effectively implemented, pioneer parties might lose their comparative advantage once legislated gender quotas are imposed upon all party competitors, particularly when strong sanctions for noncompliance are established. Yet, studies indicate that this is not necessarily the case. Meier (2004) shows that the introduction of statutory quotas incentivizes parties that were already using quotas to strengthen their equality guarantees by surpassing the legally imposed requirements, stimulating them to further champion gender equal representation in their party system. Likewise, Verge and Espírito-Santo (2016, 418–19) pinpoint that an effective compliance with legislated quotas is greater in parties which already apply voluntary quotas successfully, whereas they still need to be “accommodated” to parties where this representational norm is not a taken-for-granted act or where positive action is vocally opposed (see also Cheng and Tavits 2011). While the latter group of parties may comply with the “letter” of quota rules (required minimum proportions of female candidates on party ballots), they are likely to fail to respect their “spirit” (substantially increasing women’s presence in viable positions), thereby electing fewer women MPs than parties with longstanding voluntary quotas. Therefore, we predict the following:
In contrast, party differences may be rather small as regards the gender allocation of heads of list—that is, position number one in party ballots. Rank orders reflect candidates’ party positional power (Childs and Murray 2014; Franceschet and Piscopo 2014; Threlfall 2005), with the lion’s share of top positions on party ballots still being largely allocated to male candidates, who tend to be the district-level party leaders (Verge and Espírito-Santo 2016, 427). Not only do men still hold most of the leading positions in the party apparatus (Wauters and Pilet 2015, 82) but they also benefit more than women from this political resource in their ascendant political careers (see Verge and Claveria 2018). Accordingly, we expect the following:
From a candidate-level approach, we can also identify relevant factors affecting women’s chances of getting elected to parliament. Both district magnitude (the number of seats contested in a given district) and party magnitude (the number of seats obtained by a party in the previous election) may provide gendered opportunities to candidates. The underlying logic is that the more seats there are in internal competition as embodied by party magnitude or in external competition through district magnitude, the easier it is for parties to apply a “positive sum,” with ballots reflecting more diversity in candidates’ profiles, including gender (Golder et al. 2017, 111). In practice, party magnitude has proven to be more crucial to women’s representation than district magnitude (Matland 1993; see also Wylie and dos Santos 2016, 435). Where district magnitude is large but many parties compete for seats party magnitude is reduced, so obtaining a viable candidacy might be just as hard for women as in districts with fewer parties and smaller magnitudes (Darcy, Welch, and Clark 1994). All in all, we predict the following:
Lastly, individual candidate qualities such as previous candidate experience and incumbency may be highly valued by party selectors when distributing the higher-ranked positions on party ballots (Górecki and Kukołowicz 2014). Yet, several scholars have shown that political experience is not equally attainable for men and women. For one thing, women are discouraged from pursuing long-lasting careers by the male-oriented and family unfriendly culture of political institutions (Dolan, Deckman, and Swers 2010). Simultaneously, research on political recruitment has provided evidence indicating that men are much more likely to be encouraged to run or to be reselected by party selectors—who are typically men (Dittmar 2015; Lawless and Fox 2010)—and that “women have to be exceptional to overcome social, structural, and political barriers to office” (Murray 2014, 530). Female candidates also have less chance of being placed in viable positions on parties’ ballots, regardless of their political experience (Cheng and Tavits 2011; Luhiste 2015; Murray 2014), although incumbency has a larger positive effect for female than for male candidates (Chiru and Popescu 2017). These findings point out the double standards imposed on female politicians. Extant research has also shown that men land in viable positions directly from party office without having previously held an elected seat more often than women (Luhiste and Kenny 2016). Likewise, men with experience in party office are more likely to be fasttracked to viable positions on party ballots than women with similar seniority in holding party office (Verge and Claveria 2018). In this vein, our last theoretical expectation posits the following:
Data and Methods
Among the ten European countries using CLPR, Spain is the country that has presented higher levels of female representation over time, due to the early introduction of gender party quotas. Six of these ten countries, including Spain, currently have legislated electoral gender quotas. The Spanish statutory quota is the only one establishing thresholds for gender balance that fit into the European Union’s definition of “parity democracy” (i.e., close or equivalent participation of women and men, within a 40–60 range of representation), with the other five legislated gender quotas fixing a minimum proportion for women of around a third of candidacies in party ballots (Institute for Democracy and Electoral Assistance 2013). The Spanish statutory quota is also one of the few presenting the strongest sanctions for noncompliance, namely the withdrawal of party ballots which do not meet the prescribed minimum percentage of candidacies for the underrepresented sex. Therefore, Spain offers an ideal setting to examine the extent to which parties’ strategic discrimination against women still occurs (i.e., the fact that parties favor male candidates when drafting their ballots) despite having the most advantageous conditions for producing equal gender representation.
Our empirical analysis concentrates on the nine national elections held in Spain between 1986 and 2016. As we have mentioned, throughout this period some political parties adopted and gradually expanded gender party quotas. Such parties mostly come from the left, namely the left of center Partido Socialista Obrero Español/Spanish Socialist Workers’ Party (PSOE) and Izquierda Unida/United Left (IU), the successor of the Partido Comunista de España/Communist Party of Spain (PCE), along with their respective sister parties or stable coalitional partners in Catalonia, respectively, the Partit dels Socialistes de Catalunya/Party of the Catalan Socialists (PSC) and Iniciativa per Catalunya Verds/Initiative for Catalonia Greens (ICV) (Verge 2012). In the most recent elections all these parties, in addition to the new left-populist party Podemos/We Can (Ps), applied zipping to their party ballots through which women and men strictly alternate throughout the list. Nonstatewide parties have also adopted quotas or targets for women’s representation, including left-wing parties (Esquerra Republicana de Catalunya/Republican Left of Catalonia, ERC; Bloque Nacionalista Galego/Galician Nationalist Block, BNG; Eusko Alkartasuna/Basque Solidarity, EA) and right-wing parties (Partido Nacionalista Vasco/Basque Nationalist Party; Coalición Canaria/Canarian Coalition, CC; Convergència Democràtica de Catalunya/Democratic Convergence of Catalonia, CDC). In sharp contrast, the main statewide conservative party, the Partido Popular/Popular Party (PP), has long stood firm in rejecting quotas on ideological grounds, a position shared by the recently created statewide liberal party Ciudadanos/Citizens (Cs). Table 1 summarizes the timeline of party quota adoption in Spain.
Timeline of Party Quota Adoption in Spain.
Left and left-of-centre parties: PSC = Partit dels Socialistes de Catalunya/Party of the Catalan Socialists; IU = Izquierda Unida/United Left; PCE = Partido Comunista de España/Communist Party of Spain; PSOE = Partido Socialista Obrero Español/Spanish Socialist Workers’ Party; ICV = Iniciativa per Catalunya Verds/Initiative for Catalonia Greens; BNG = Bloque Nacionalista Galego/Galician Nationalist Block; ERC = Esquerra Republicana de Catalunya/Republican Left of Catalonia; EA = Eusko Alkartasuna/Basque Solidarity; Ps = Podemos/We Can. Right-of-center and liberal parties: CC = Coalición Canaria/Canarian Coalition; PNV = Partido Nacionalista Vasco/Basque Nationalist Party; CDC = Convergència Democràtica de Catalunya/Democratic Convergence of Catalonia.
Female membership share at the time.
A minimum proportion of 40 percent for either sex.
While party quotas significantly boosted women’s representation in the lower house, left-wing parties sought to pass a legislated quota in the late 1990s and early 2000s. After several previous attempts were blocked by the conservative PP, the 2007 Equality Law (Act 3/2007) introduced a statutory quota. Drafted in gender-neutral terms, it requires all political parties to ensure that no sex obtains less than 40 percent and no more than 60 percent of candidacies both throughout the party ballot and across every five positions on the ballot. 2 Given that electoral authorities shall withdraw noncompliant party ballots, the statutory quota has been effectively implemented by all political parties as regards the minimum share of candidates of either sex.
The use of gender quotas and the degree to which they are effectively implemented determines the gender outcomes observed in parliament, reason why the time span of our empirical analysis covers three distinct periods, each of which includes three elections: (1) 1986–1993, where party quotas were very few and poorly implemented; (2) 1996–2004, where various stronger and more institutionalized party quotas were used; and (3) 2008–2016, where a legislated quota has been applied following its introduction in 2007.
We conduct our analysis in two parts. In the first part, we provide a descriptive survey of the magnitude of the gender gap and its evolution over time as regards both levels of women elected as well as the share of female candidates and their placement in safe seats and heads of ballot. The coding of safe seats follows the way in which extant research has accounted for a candidate’s party-determined viability, which codes the candidate’s list position in relation to the number of seats won by her party in the previous election (Luhiste 2015, 99; see also Simón and Verge 2017). This variable has been calculated for all party ballots at the district level for all the elections under examination. In the case of heads of ballot a value of 1 is given when a woman occupies position number one on the party list and a value of 0 when the ballot is headed by a man. Our data set includes a total of 18,485 candidates who ran in the ballots of those parties which had won at least one seat in the national parliament ever.
In the second part, we test our theoretical expectations more directly. We restrict the original data set to the candidates fielded by competitive parties—that is, those which always had or almost always had parliamentary groups in the Congreso de los Diputados across the 1986–2016 period. 3 We can then ensure that the results are not confounded by small parties with a marginal presence in the lower house and for which viable list positions are not available to either men or women. This yields a total of 11,289 candidates from between seven and nine parties in each election. 4
At the party level, we run ordinary least square regressions to explore the selection of women using as our dependent variables the share of female candidacies, the share of female candidates in safe seats and the share of female candidates as heads of ballot. To test H1 and H2, we include dummies to account for the different time periods pointed out above; this allows us to examine the effect of the institutionalization of voluntary gender quotas and to contrast it with the impact of legislated quotas. In addition we add an indicator for party quotas, a continuous variable that measures the minimum allocation of list positions for women that political parties have voluntarily assumed. It varies across parties as well as over time. Due to the strong overlap of voluntary quotas and left-wing parties, we refrain from including any control for party ideology but we do run fixed-effect models to control for party idiosyncrasies. Finally, we control for electoral swings—difference in votes obtained by parties—between consecutive elections, the share of seats won by women in the previous election, as well as the share of male incumbents.
As noted by previous scholarship, the more centralized the candidate selection processes are, the larger the capacity of a party to adopt gender quota reforms and the easier it is for party leaders to ensure that all party levels comply with them (Kenny and Verge 2013; Kittilson 2006; Murray 2010). We do not include this variable in the empirical analysis because all Spanish parties use rather centralized candidate selection methods (Fabre and Méndez-Lago 2009). Party lists are drafted by district-level (provincial) branches and are then submitted for approval and/or ratification to central party bodies, which hold veto power and have the capacity to alter the order of candidates in the lists (Pallarés, Astudillo, and Verge 2015). Candidate selection follows detailed and standardized written rules, which facilitates the effective implementation of (party and/or legislated) quotas once the party leadership is committed to gender equal representation (Verge and Espírito-Santo 2006). This prevents party selectors from interpreting selection rules at their discretion and restricts their capacity to distribute personal patronage (Norris 1993, 322).
At the candidate level, to test H3 and H4, we analyze the selection of women to different positions on party ballots with two different dependent variables. On one hand, we use a categorical variable that classifies candidates as being placed as the head of ballot, whether the candidate is in a safe seat on the ballot, or whether the candidate is placed in a lower ranking position. Of our 11,289 candidates, 14.1 percent are heads of ballot, 17.5 percent are placed in safe seats, and 68.4 are placed lower down. Given the nature of this variable, an ordered logistic regression model would be the adequate estimation procedure but a Brant test reveals that the parallel regression assumption is not met. For this reason, we use multinominal logistic models as an acceptable alternative. On the other hand, we take advantage of our time series and model what it takes for a candidate to be placed for the first time in a safe seat or as head of ballot, focusing on whether male and female candidates are promoted in the same way by parties. Then we only consider the first occurrence, setting subsequent occurrences to missing values, and use logistic regressions to estimate these models. To make sure that we capture a candidate’s upward move through the ballot ranking, we exclude those few instances in which a candidate had started as head of ballot and was then moved further down the list positions.
Using individual data allows us to examine key characteristics of candidates that may explain women’s placement on ballots and their subsequent election as MPs, in conjunction with institutional and contextual factors. As Luhiste (2015) points out, men tend to be politically more experienced than women, which may explain their placement in viable positions (see also Górecki and Kukołowicz 2014). We exploit the information contained in our time series by including a variable indicating the number of consecutive candidacies at the national level as a proxy for candidate experience. We acknowledge that this measurement is limited, insofar as it does not include the experience that candidates may have gained at other levels of office (local and regional level, executive office or party office). Nonetheless, it allows us to trace the fate of the same candidate over a prolonged period of time. In addition, we include a dummy for incumbency measured as 1 when the candidate was elected in the previous election and 0 otherwise. 5 To test for institutional and contextual factors, we again include a variable for party quotas, a control for our three different time periods and capture a party’s viability at the district level by measuring the proportion of contested seats that a party has won in a given district in the previous election. 6 Lastly, we add a dummy variable controlling for densely populated districts that include major cities to account for a potential cleavage capable of influencing the gendered recruitment practice of parties. 7
Results and Discussion
In Spain, the transition to democracy in the post-Franco period began in 1976. The first democratic parliament was elected in 1977 and drafted the new constitution, which was ratified by the Spanish people in 1978. Women comprised about 6 percent of the lower house up to 1986, and since then, women’s representation increased steadily until 2004, when it leveled off at around 36 percent. The progress made by women in the Spanish case has been attributed to feminists working inside the main left-wing political parties to improve gender equality socially, economically, and politically, and also to the willingness of party elites to comply with gender quotas (Threlfall 2005; Valiente 2005; Verge 2012). Still, parity has not been attained in parliament. Throughout the period under examination, gender gaps are observed in both the composition of party ballots and the pool of elected candidates. Figure 1 visualizes this gap across the 1982–2016 period for those parties which had won at least one seat in the national parliament, as well as separately for left-wing and right-wing parties. Vertical dashed lines indicate the different time periods, underscoring the impact of the expansion of voluntary party quotas (1996–2004) and the introduction of the legislated quota (2008 onward). As can be seen, the gap for women’s representation as candidates has closed comparatively faster than their representation as MPs, and particularly before the introduction of the statutory quota. In both instances, this narrowing was driven by parties on the left. For right-wing parties, while the legislated quota significantly increased the level of female candidacies, in terms of elected female MPs the gains were relatively small. In the three elections held after the introduction of the legislated quota (2008, 2011, and 2016), women’s presence on party ballots neared 50 percent but women’s presence among elected representatives still lagged behind at closer to 36 percent in 2008 and 2011. In 2016, women MPs obtained 39 percent of seats. Despite becoming the most feminized parliament so far, a gender gap of 11 percentage points remains. Even for left-wing parties, which fielded more women than men candidates in this election, a gap of 5 percent persists among elected MPs.
Gender gap in candidates and elected members, 1982–2016.
Figure 2 provides a closer examination of party ballots and highlights a disparate allocation of viable list positions that may help to shed light on the patterns uncovered above. A clear tendency to reward men candidates in the distribution of safe seats and as head of ballot is visible, although the gender gap has narrowed over time. In 1986, only 8 percent of women were placed in safe seat positions, a figure that steadily increased to reach 41 percent in 2016. However, the trajectory for left-wing and right-wing parties differed substantially due to the implementation of voluntary party quotas. While in 1989, right-wing parties had placed about 7 percent of women in safe seats, left-wing parties allocated 19 percent of their women candidates in such positions, a percentage that jumped to 36 percent in 1996, increasing to 44 percent in 2016. For right-wing parties the largest gain happened from about 20 percent of women in safe seats in 2000 to 29 percent in 2004, reaching 39 percent after the introduction of the legislated quota. Progress regarding the top position in the ticket has been much slower over time and all parties are guilty of underrepresenting women as heads of ballot. In 1982, men led 98 percent of the party lists, a figure that slowly dropped to 77 percent in 2008. So neither voluntary party quotas nor the statutory quota have had a discernible effect in this case. In fact, up to the 2011 elections, the differences across ideological groups in the allocation of head of ballot positions across sex were rather marginal and much smaller than the gender gap found in the case of safe seats.
Gender gap in viable positions, 1982–2016.
Our descriptive examination of candidate data has revealed that political parties continue to discriminate against female candidates. Parties’ enactment in practice of the 40–60 range as a 40 percent threshold for women with regards to the allocation of top positions means that the share of elected female MPs has not yet reached the minimum of 40 percent, thereby failing to respect the spirit of the Equality Law.
We start exploring our explanations for such gendered discrimination by focusing on the party level and, in particular, on which factors determine the levels of the following variables: share of female candidates, share of women in safe seats, and share of women as heads of ballot. Table 2 shows the results for ordinary least squared regressions at the party-election-year level for each of these variables, with and without party fixed effect to account for party idiosyncrasies. As Models 1a and 1b indicate, the pool of female candidates is strongly determined by the targeted percentage established by voluntary party quotas. The share of female candidates also increases with the more widespread adoption, gradual enlargement, and eventual institutionalization of voluntary party quotas, as well as with the implementation of legislated quotas, as indicated by our two period dummies—taking the first three elections (1986, 1989, and 1993) as a baseline. It should be noted that, since the legislated quota followed exactly the same quota rules used by extant party quotas, for parties which used positive action to draw up their own gender-balanced tickets, this procedure had already become a taken-for-granted act. Nevertheless, it forced the remaining parties to adjust their recruitment strategy. In other words, quotas are decisive for the overall increase in female candidates.
Party-Level Analysis.
Standard errors in parentheses.
p < .1. **p < .05. ***p < .01.
Does this effect also hold for safe seats? Models 2a and 2b in Table 2 confirm our first hypothesis (H1), showing that the institutionalization of party quotas in Spain (during the period 1996–2004), combined with the level of party quota adopted, was of particular importance for the share of women in safe seats. Conversely, as indicated by our dummy for the period 2008–2016, legislated quotas do not affect the share of female candidates allocated to safe seats. The strong sanction established for failure to comply with the required proportions—that is, withdrawal of noncompliant district-level party lists by electoral authorities—have instilled a significant increase in the share of female candidates among parties which did not previously use voluntary quotas, but the placement mandate fails to guarantee an equal distribution of safe seats. While the share of seats won by female candidates in the previous election appears to drive the allocation of female candidates to safe seats, the party fixed effects included in Model 2b hint at the importance of women’s agency for pushing parties to include more women in such positions, as is well documented for example in the case of the PSOE. By leaking to the press that the party risked not reaching parity in the parliamentary group, female MPs used the party’s strong public commitment to gender equality to make party leaders enact the quota proportions to safe seats (Verge and Espírito-Santo 2016, 433).
Our second hypothesis (H2) regarding the most prominent list position is also confirmed. Models 3a and 3b in Table 2 indicate that head of ballot positions tend to be secured against possible female contenders. The share of male incumbents significantly lowers the share of female candidates placed at the top position of the ballot, confirming the bias in favor of male candidates already shown in our descriptive analysis above. Despite its persistence, it is worth highlighting that the gender gap in head-of-ballot positions has narrowed over time, especially in the most recent period. The highly significant coefficient for the post-statutory quota elections—the 2008–2016 period—reflects the “compensatory” strategy applied by conservative parties, which consists in placing female candidates in the most visible position to cover up for the fact that women remain underrepresented in safe seats. Such a strategy can be understood as “defensive” behavior, with party laggards, typically on the right, seeking to prevent vote losses, especially once the gender-balance norm in political representation has gained significant international legitimacy. For their part, to avoid losing their comparative advantage, left-wing parties responded with an “offensive behavior,” firstly with a large increase in the share of women heads of ballot and subsequently with the application of horizontal zipping whereby half the ballots are headed by women (Simón and Verge 2017, 187).
The question that arises is whether the tight control of the candidate selection process granted by the CLPR system and the strategic discrimination visible in the party-level results can be compensated at the candidate level. The results shown in Table 3 clearly indicate that women are more likely to populate the lower ranks of party ballots, as shown by the negative and significant coefficient of the sex variable. Experience in the form of candidate experience or incumbency is particularly important to rise above the lower ranks, whereas being an incumbent does not appear to matter for being placed as head of ballot vis-à-vis a safe seat position.
Modeling the Selection of Candidates in Spain.
Standard errors in parentheses.
p < .1. **p < .05. ***p < .01.
To better grasp the scope of gendered discrimination we need to explore the impact of our explanatory variables particularly on women candidates. Since we can expect the variances to differ across sex, we run separate models and compare the differences of the coefficients with Wald tests (see Luhiste 2015). 8 While the left part of Table 4 compares access to safe seats vis-à-vis being ranked at a lower position separately for female and male candidates (Models 1 and 2), the right part of the table examines what it takes to rise above the ranks of safe seats to be placed at the head of a party list (Models 3 and 4).
Modeling the Selection of Female and Male Candidates in Spain.
Standard errors in parentheses.
p < .1. **p < .05. ***p < .01.
The results indicate that districts in which parties are more viable improve both male and female candidates’ chances of being placed in a safe seat. Nevertheless, the equality of coefficients based on the Wald tests is rejected, meaning that in districts in which seats previously won by a party exceed two-thirds of available seats, female candidates’ chances surpass that of their male counterparts. For instance, if a party holds 75 percent of the seats, a woman candidate’s average probability of being placed in such a position is 0.63 (compared to 0.60 for men) and 0.89 if the party has won all seats available (compared to 0.85 for men). Models 3 and 4, on the other hand, reveal that women candidates are in general disadvantaged regarding head of ballot positions. In nonviable districts, a female candidate’s average predicted probability amounts to 0.11 vis-à-vis 0.24 for a male candidate, a substantial gap that persists even if a party obtains half the district seats (0.07 vs. 0.17) and is only almost closed if a party holds all district seats (0.01 vs. 0.04). In other words, while party viability increases women candidates’ chances for placement in safe seat positions, the top position appears to be reserved for men candidates. All combined, we can accept our third hypothesis (H3).
Political experience in the form of incumbency benefits both men and women candidates when it comes to the allocation of safe seats, but has no impact on being placed as head of ballot. Spain follows a familiar pattern of higher turnover of women than men in public office, although Figure 3 indicates that the gender gap in experience has narrowed over time, especially with the introduction of the legislated quota. It should be mentioned, though, that as highlighted by the Wald test the effect of accumulated candidate experience does not differ across sex, contrary to our fourth hypothesis (H4).
Gender gap in candidate experience, 1982–2016.
Finally, variables operating at the aggregate level also have some impact on individual candidates’ chances of placement. The result of the Wald test points to the advantage of left-wing parties during the institutionalization phase of party quotas (1996–2004). During this period, a woman candidate’s average probability of being placed in a safe seat was 0.19—an increase of 4 percent compared to the previous period. Party quotas indeed leveled the playing field: without party quotas the average predicted probability of women candidates for a safe seat position is 0.15 compared to 0.21 for men, while a quota of 40 percent gives a likelihood of 0.19 for both.
Yet this does not apply to head of ballot positions, thereby suggesting again that parties may have safeguarded top positions against the increase in female competition brought about by positive action. While the predicted probability for female candidates hardly changes (0.7 without voluntary party quota vs. 0.8 if a quota of 40 percent is imposed), male candidates’ prospects of leading a party ballot increases from a predicted probability of 0.16 without quotas to 0.20 with a 40 percent quota.
However, the focus on contrasting the likelihood of a candidate being placed in a more viable position as opposed to a less viable position may actually mask gendered patterns of selection if we are interested in determining the persistence of the gender gap. We, therefore, turn to the analysis of the factors that help a candidate achieve a viable ballot position for the very first time. In other words, do party officials promote men and women candidates in the same way, or do we observe dissimilar treatment across sex?
The results in Table 5 reinforce our previous findings. The level of voluntary party quotas and the time period of their stronger institutionalization confirm the pattern for reaching a safe seat position for the first time. Party viability, though, does not have any impact on promoting women and men to viable list positions. 9 More importantly, the right-hand side of Table 5 suggests that parties do use double standards when assigning head of ballot positions, with more demanding criteria being imposed on women candidates. The results underscore that women need to show more political experience than their male counterparts to reach a top position. Incumbency improves a woman candidate’s average probability of being placed as the head of ballot by 7 percent, while four consecutive candidacies do so by 6 percent. 10 Incumbency or accumulated candidate experience, on the other hand, is not a requirement for men candidates to reach such a position, lending support to our fourth hypothesis (H4) when it comes to promoting women’s careers in politics.
First Time Safe Seat and Head of Ballot.
Standard errors in parentheses.
p < .1. **p < .05. ***p < .01.
In sum, our analyses paint a clear picture of gendered political recruitment. Institutions such as gender quotas and the degree to which they are effectively implemented are decisive in leveling the playing field and in improving women candidates’ chances of being elected. Nonetheless, this does not prevent parties from exerting strategic discrimination. Top positions have been protected from female competition and safeguarded for men candidates, while women candidates are put under more strain to prove themselves before being allowed to lead a party ballot.
Conclusion
Most research on women’s representation in parliament studies the final stage of the election process—who wins. This notwithstanding, the party selection stage is critically important. The Spanish case illustrates the relevance of looking at both the selection and the election of women. While Spain currently ranks sixteenth in the world in its parliamentary representation of women, with 39.1 percent of its lower house being female (Inter-Parliamentary Union 2018), women are not yet being elected at the same rates as men. The key lies in gendered selection patterns. Indeed, selection is particularly crucial in CLPR systems where such a large part of a candidate’s probability of winning a parliamentary seat depends upon being assigned a viable position on the party ballot to begin with.
Examining the selection of women to party ballots in Spain reveals that all parties discriminate against female candidates. Both the candidate pool of women and the share of safe seats allotted to women have increased significantly, particularly in left-wing parties where voluntary quotas have long been used and have become institutionalized in their practical enactment. However, when choosing candidates to head the ballot, we witness a strong bias in favor of male candidates. The candidate-level analysis, for its part, provides evidence that the strategic discrimination that parties apply to women is hard to overcome with candidate characteristics alone. The fact that incumbency increases women’s chances of being placed in a safe seat or as head of list on party ballots may indicate that the weight of candidates’ political experience is strongly gendered, with women being required to have much more political experience than men to obtain a viable candidacy.
Overall, our results suggest that candidate tickets reflect broader gender power dynamics entrenched in political parties. Women still appear to be perceived as “political underdogs” by party gatekeepers who instead fasttrack male candidates to viable positions, resulting in significant consequences for gender power distribution in politics. Not only do positions on electoral lists reflect candidates’ prominence within the party hierarchy (Franceschet and Piscopo 2014, 96; Threlfall 2005, 137), but also top-ranked candidates on party lists are likely to be awarded high-ranked positions in party executive bodies (O’Brien and Rickne 2014, 15). Male dominance of the positions at the top of the lists is instrumental for both securing the election of men candidates and for maintaining men’s positional power within parties (Childs and Murray 2014, 81), establishing a positive feedback loop for male candidates but a vicious circle for female candidates. Therefore, while Spanish electoral rules may be some of the most favorable toward equal gender representation, party gatekeepers continue to obstruct women’s political advancement. Further research is thus needed to unveil the informal norms that sustain such gendered strategic discrimination as well as to identify how the former can be counteracted with quota rules that produce effective parity representation.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the project Mulh(j)er e Poder: Women’s Political Representation in Portugal and Spain (PTDC/IVC-CPO/4088/2012), led by Nina Wiesehomeier. Financed by the Fundação para a Ciência e a Tecnología, Ministerio da Educação e Ciência, Portugal.
