The Weight on Her Shoulders: Marginalization of Women Legislators in Parliaments and Substantive Representation of Women

Abstract

Studies have widely documented that women's descriptive representation in parliaments enhances their substantive representation. We probe this relationship under varying levels of women's collective and individual marginality based on an original dataset documenting the parliamentary behaviour of Israeli legislators over eleven parliamentary terms (1977–2015). Using several measures of individual-level marginality we show that marginalized female legislators are more prone to engage in gender-related parliamentary activity than their less marginal counterparts, albeit only under a certain threshold of women's marginality as a group. The article elucidates the dynamic nature of the relationship between descriptive and substantive representation of disadvantaged groups by demonstrating that it is contingent on their collective standing in parliament and on the marginality of individual legislators as manifested in their strategic choices.

Introduction

In recent decades, post-industrial democracies have increasingly incorporated women into their legislatures, sparking extensive research into their role in advocating for women's interests. While scholars argue that this ‘feminization’ of politics enhances substantive representation, it remains uncertain whether women legislators truly make a substantial impact. This article seeks to address the question by exploring the conditions under which descriptive representation translates into substantive representation. By doing so, we aim to offer a nuanced understanding of the link between both.

Women legislators face marginalization due to cultural, institutional, and structural barriers that limit their influence on the legislative process (Barnes 2016). This marginalization, conceptualized as a lack of political power and influence in policymaking, involves numerous formal and informal obstacles. Despite increased representation of women in parliaments globally, their marginalization persists since it is not solely driven by women legislators' minority status. Women have unequal access to positions of power compared to men, with influential legislators holding disproportionate sway over legislative agendas and resource allocation. Informal barriers, such as exclusion from key leadership discussions and networks, and enduring negative stereotypes, further compound their challenges. This multidimensional marginalization underscores the complexities faced by women legislators.

We theorize that marginalized women legislators, who lack political power and agenda-setting resources, are more likely to substantively represent women's interests due to a mixture of push and pull factors. Specifically, we argue that marginalized women legislators will be more inclined to substantively represent women due to (1) the accessibility of opportunities at their disposal; (2) the prospect of collaboration with fellow women legislators, and (3) the advantages they can glean from such behavioural choices. As women legislators become less marginalized and gain more resources in parliament, their incentives to address diverse issues in their policy portfolio rise, reducing the benefits of substantively representing women's interests.

We further argue that it is crucial to examine the interplay between individual and group-based marginalization in shaping women legislators' policy portfolios. Our analysis reveals that group-based marginalization, such as women's collective power in the assembly or within their parties, moderates the impact of individual-based marginalization, such as parliamentary tenure, on women legislators' engagement in substantive representation of women. Specifically, the positive relationship between women legislators' marginalization and the substantive representation of women's interests only holds in contexts of low group marginalization.

To examine the relationship between women legislators' marginalization and substantive representation, we focus on Israel, a ‘most likely’ case for our analysis. Over the past two decades, Israel has seen a significant rise in women's descriptive representation in the Knesset, from 11.6 per cent in the 15th Knesset term (1999) to 25 per cent in the 25th term (2022). Despite this progress, women still face a glass ceiling, which is evident in their limited access to ministerial positions or committee memberships. Furthermore, Israeli society's adherence to traditional gender roles and the dominance of security concerns in politics perpetuates gender stereotypes and marginalizes women, who are often viewed as less competent in addressing such issues. This underscores the complexity of the increased numeric representation of women in parliament alongside entrenched gender norms and roles.

We utilize an original comprehensive dataset of private legislative initiatives sponsored by women and men legislators in the Israeli parliament (hereafter, the Knesset) between 1977 and 2015 (Knesset terms 9–19). The dataset includes 15,483 bills, of which we coded 1,099 as gender-related. We predict women legislators' likelihood to sponsor such bills as a function of their individual and group marginalization in parliament by estimating negative binomial regression models. We deduce a plethora of marginalization indicators to test our claim. Marginality at the individual level is tested using four indicators, each representing a different dimension of the concept: personal marginality (parliamentary tenure), party-related marginality (electoral list safety, parliamentary/partisan role), and structural marginality (coalition/opposition affiliation). Group-level marginality is tested via two indicators, the share of women legislators in parliament and the party. Our empirical analysis is two-pronged. First, in a series of moderation models, we demonstrate that women legislators are less/more likely to sponsor gender-related bills the less/more marginal they are in parliament, both individually and as a group. No such effect is registered for male legislators. Next, we show that this relationship obtains only in contexts of low group marginalization. When group marginalization is high, the effect of a woman legislator's level of individual marginalization on her likelihood to substantively represent women is neither significant nor substantial.

This article joins a bourgeoning literature of some studies where it is claimed that women in office make a difference, while others assert that they do not. We argue against either of these holistic approaches, contending instead that the relationship between women's descriptive and substantive representation is conditioned by the settings, incentives, and opportunities in each case. By doing so, we follow in the footsteps of previous work that has looked at the effect of marginality (Bailer et al. 2022; Barnes 2016; Höhmann 2020; Schwindt-Bayer 2010), but propose a more general theory regarding the dynamic relationship between women legislators' marginalization and their propensity to engage in the substantive representation of women. As we show in this article, marginality changes not only between legislators but also for each individual legislator (within legislators, as it were) in the course of their political careers, making them either more or less prone to represent women.

Furthermore, this article highlights the dynamics between collective and individual marginalization, arguing that the latter's impact is influenced by the former. By acquiring parliamentary or political power, women legislators can shed their marginality as individuals, and the collective marginality of women as a group in the respective parliament or party continues to constrain their behaviour. Both factors must, therefore, be examined in the case at hand. Finally, this article raises several new questions salient to women's substantive representation in politics and the conditions that facilitate it. Clearly, for an individual women legislator, marginality is an obstacle to power and influence, and she is much better off unencumbered by this troublesome barrier. However, to the extent that less marginalized women legislators tend to assign less value to representing women's interests, what does such empowerment entail when substantive representation of women is concerned? These questions are at the centre of this article.

Does Women's Descriptive Representation Entail Substantive Outcomes?

The nature of the relationship between the descriptive and substantive representation of women in parliament – that is, respectively, women's numerical presence and the degree to which the preferences and interests of women are represented – has been investigated in a large number of studies (Barnes, Beall, and Holman 2021; Beckwith and Cowell-Meyers 2007; Betz, Fortunato, and O'Brien 2021; Celis and Childs 2012; Clayton and Zetterberg 2018; Cowell-Meyers and Langbein 2009; Heath, Schwindt-Bayer, and Taylor-Robinson 2005; Lloren 2015; Reingold and Swers 2011; Rodríguez-Garcia 2015; Smooth 2011). This line of empirical inquiry is premised on Phillips's (1998) presence model, whereby legislators from disadvantaged groups make more authentic advocates of the latter's interests in parliament due to shared experiences and background (Atchison and Down 2009; Childs and Krook 2009; Kittilson 2008; Mackay 2008; Sawer 2012; Wängnerud 2009). For instance, Betz, Fortunato, and O'Brien (2021) found a correlation between women's representation in parliament and reduced import tax penalties on goods targeted at women. Similarly, Barnes, Beall, and Holman (2021) revealed that higher levels of pink-collar representation by women were linked to increased state spending on education and social services. Clayton and Zetterberg (2018) demonstrated that quota shocks not only diversified parliaments, they also shifted political priorities toward areas favoured by historically marginalized groups. Indeed, many studies divergent in terms of conceptual framework and methodology indicate differences in the parliamentary behaviour of men and women and show that women's descriptive representation often augments their substantive representation.

Nonetheless, such gender differences can be contingent on institutional and cultural factors, such that, according to some scholars, ‘the ability of female politicians to represent women's interests is context dependent’ (Koch and Fulton 2011, 2). Moreover, the linkage between descriptive and substantive representation grows weaker as parliamentary activities become increasingly constrained by institutional features (Childs and Krook 2009; Hawkesworth 2003; Höhmann 2020; Rocca and Sanchez 2008; Wüst 2014). The contingent nature of this link means that ‘an MP's background will not translate into a persistent and continuous representative effort on behalf of their social groups throughout their career’ (Bailer et al. 2022, 538).

To better understand this dependency, recent studies have focused on individual legislators' strategic considerations in substantively representing a disadvantaged group (Bailer et al. 2022; Carnes 2013; Davidson-Schmich 2016; Gilens 2012; Greene and O'Brien 2016; Grose 2005; Mansbridge 2015; Minta 2009; Öhberg and Wängnerud 2014). Thus, Bird, Saalfeld, and Wüst (2010) point to an entrepreneurial use of ethnicity by visible minority candidates during electoral campaigns, as well as by legislators in office. Saalfeld and Kyriakopoulou (2010) demonstrate the strategic use of ethnicity by British legislators through ‘toggling’, thereby elevating their own minority status, as Collet (2008) puts it, by building on their ‘experience with the ethnic community, knowledge of its norms, language, and cultural practices, and deeper connections with its key players’ (713). Höhmann (2020) shows that women legislators are more likely to concentrate on the representation of women's interests if their re-election is secured and if they do not have to fight for additional local votes from their district; however, his study only addresses electoral incentives and does not take into account the legislators' status in parliament. Bailer et al. (2022) take a further step, looking at legislators who share descriptive features with disadvantaged groups – women, migrants, low social class, and the young – and identifying a link between their time in office and their propensity to represent the interests of such groups. They show that legislators represent their respective groups at the beginning of their careers because this confers credibility when they have, as yet, no legislative track record and have had few opportunities to demonstrate expertise; later on, however, their efforts on behalf of the group are supplanted by other legislative activities.

When considering women legislators' strategic choices in representing women's interests, gender stereotypes play a significant role. Research suggests that women are often perceived as compassionate, trustworthy, and competent in areas related to empathy and care (Gordon and Miller 2001; Huddy and Terkildsen 1993), while men are seen as more authoritative and agentic, particularly in traditionally ‘masculine’ domains (Alexander and Andersen 1993; Holman, Merolla, and Zechmeister 2011; Koch 2000; Koch and Fulton 2011). Women legislators may encounter credibility challenges when addressing issues perceived as masculine (Koch and Fulton 2011) but may be viewed as more credible in nurturing areas. Consequently, it stands to reason that women in parliament might leverage such stereotypes for political gain and construct their legislative portfolios with an eye to public expectations regarding gender roles and party reputations (Itzkovitch-Malka and Friedberg 2018; Swers 2013).

Marginalization of Women in Parliaments

Notwithstanding considerable progress in recent decades, parliaments still adhere to traditional gender norms, which strongly affect the gender-role perceptions underpinning their operational practices. This culture is a remnant of an era when politics were ‘men's business’ when both formal and informal behavioural rules were created by men for the benefit of men. This gendered hierarchy precludes women from entering elite networks, which are typically dominated by men. Women's institutional disadvantage within parliaments diminishes the likelihood of their gaining influence in legislation or being appointed to senior or leadership positions (Barnes 2014; Barnes 2016; Heath, Schwindt-Bayer, and Taylor-Robinson 2005; Kerevel and Atkeson 2013; Schwindt-Bayer 2006; Victor and Ringe 2009; Wojcik and Mullenax 2017), and increases the likelihood of their exclusion from powerful committees (Heath, Schwindt-Bayer, and Taylor-Robinson 2005) whose members are better able to centralize information, influence legislation, negotiate support from colleagues, and affect important political outcomes. This marginalization prevents women from influencing legislation or policy (Heath, Schwindt-Bayer, and Taylor-Robinson 2005; Kerevel and Atkeson 2013; Victor and Ringe 2009).

We conceptualize women's marginalization in parliament as a lack of political power and influence in policy-making, created by numerous formal and informal obstacles. This conceptualization is in line with that of others in the field (see for example Barnes 2016; Schwindt-Bayer 2010; Senk 2023). In this context, it is important to distinguish between formal institutional structures and informal networks of power and collaboration. Previous research has shown that while formal rules usually do not discriminate against women, power hierarchies, informal norms, and expectations are still gendered, thereby favouring men and masculine behaviour (Erikson and Josefsson 2019; Lovenduski 2005). In line with this observation, marginalization is often described as women being marginalized within institutions rather than by the institutions themselves (Barnes 2016; O'Brien and Piscopo 2019; Schwindt-Bayer 2010).

Marginalization serves as an umbrella term, encompassing the multifaceted nature of women's out-of-power status in legislatures. In this article, we refer to several dimensions of marginalization within parliament using conventional concepts in legislative politics, which also reflect women's marginalized power and influence. We conceptualize women's marginalization as composed of both individual and collective factors. At the individual level, we use four indicators for marginality: personal marginality, measured by parliamentary tenure; party-related marginality, measured by electoral list safety and parliamentary/partisan role; and structural marginality, measured by coalition/opposition affiliation. Importantly, this conceptualization allows marginalization to vary between women legislators, and also during an individual legislator's political career. Some legislators are more marginalized than others, but even the same legislator can be either more or less marginal at different time periods and in institutional contexts.

The multidimensional character of women's marginalization is well reflected in our conceptualization. For individual marginality, we start with parliamentary tenure or seniority, which is a personal attribute per se, based solely on the legislator's political experience. In line with the discussion above, it is important to note that it is not necessarily seniority itself that is crucial but, rather, the networks of power, collaboration, and knowledge that come with it. Senior legislators often have more experience navigating legislative processes, access to prestigious committees, and influential leadership positions within the legislature. Next, we move to the second dimension of marginality, which is party-related with two indicators that bring in the party as an institution within which women are marginalized – electoral list safety and parliamentary/partisan role. These can be somewhat correlative with seniority, but still warrant an individual examination as there are many other factors which may affect such indicators. Finally, we have coalition/opposition affiliation as an indicator for the third dimension of individual marginalization, which is more structural. We address this attribute to the individual legislator, as it still directly affects her power and influence, but we acknowledge that this dimension involves broad considerations.¹ As for group-level marginality, we encompass the scope of marginalization using two indicators, addressing the share of women legislators in parliament and their share in the party each legislator belongs to.

How Marginalization Structures Women Legislators' Propensity to Engage in Gender-Related Policy Issues

According to Bailer et al. (2022), a legislator from a disadvantaged group is more prone to represent their interests at the beginning of his or her parliamentary career (though see Senk (2023) for different results²). These researchers demonstrate that legislators' representation efforts change along their career paths, such that their incentives to represent disadvantaged groups are stronger early on. We argue, on the other hand, that a legislator's professional trajectory is one of several factors that determine his or her marginality. Other grounds for marginality may endure throughout a legislator's entire parliamentary career, or not progress linearly as parliamentary seniority. Therefore, we propose a more general theory regarding the dynamic relationship between women legislators' marginalization and their propensity to engage in the substantive representation of women.

Given the pervasive marginalization of women legislators in parliament, our focus is on understanding why this marginalization primarily influences their behaviour in gender-related parliamentary activities rather than across a broad range of policy issues. As previously elucidated, the widespread marginalization of women legislators within parliament permeates their entire political experience, subjecting them to exclusion across a broad spectrum of policy issues. What then prompts our assertion that the manifestation of this marginalization will predominantly become apparent in their gender-related parliamentary behavioural patterns, and not across a wide range of policy issues? Viewing legislators as rational actors seeking to maximize ‘policy, office, or votes’ (Müller et al. 1999), we examine the factors influencing their decisions, particularly in light of their marginalized status. We focus on three key aspects: (1) the accessibility of opportunities for marginalized women legislators, (2) the potential for collaboration with fellow women legislators, and (3) the strategic advantages for women legislators in their behavioural choices. We elaborate on these points below.

1. (1) When it comes to parliamentary resources, marginalized legislators often find themselves sidelined from prestigious roles and high-profile policy areas due to the dominance of more powerful counterparts. Research has shown that women legislators, in particular, are directed towards lower-profile issues like gender policy by the male majority, limiting their opportunities (Heath, Schwindt-Bayer, and Taylor-Robinson 2005; Schwindt-Bayer 2006). In such cases, women legislators' space of opportunities is substantially restricted. Nevertheless, even a narrow window of opportunities can afford a considerable choice of policy issues to pursue. Why would women legislators choose to focus on gender-related policy issues when marginal?

2. (2) One significant pull factor is the opportunity for collaboration among women legislators. Research indicates that women tend to form more collaborative partnerships in parliament compared to men, with a higher proportion of their collaborators also being women (Barnes 2016). These collaborations help women overcome disadvantages and compete effectively with men (Bratton and Rouse 2011; Wojcik and Mullenax 2017). Additionally, collaborations often focus on shared interests, particularly gender-related policy issues. By working together, legislators gain a deeper understanding of women's issues and can develop legislation that reflects diverse perspectives. Given that women's concerns often transcend party lines, there is a strong tendency for women to collaborate across party divides on gender-related policy issues. Consequently, the high potential for collaboration with fellow women legislators in this area encourages women to prioritize addressing gender-related policies over other areas, especially when marginal.

3. (3) A second pull factor has to do with women legislators' strategic calculations and the advantages they anticipate gaining through the representation of women. Legislators' policy portfolios are structured by endogenous factors such as personal interest, prior experience, and ideology, and also by strategic calculations in a bid to come across as competent and credible legislators (Bailer et al. 2022). The latter consideration impels women legislators to take on activities in which they have personal experience, expertise, and, therefore, credibility (Butler 2014). Since women legislators are, by virtue of their gender, (stereotypically) considered to be experts on gender-related issues, they will try to ‘cash out’ on this attribute, and address gender-related policy issues in their legislation in order to appear as competent and credible legislators. For their part, women legislators – and especially those marginalized in parliament and deprived of other opportunities – can leverage their gender to influence these policy issues and, at the same time, secure political gains vis-à-vis their voters. All in all, by advocating for women's interests, marginalized women legislators are able to present themselves as highly credible representatives, thus – to a point – turning their disadvantage into an advantage.

In sum, women legislators' marginalization in parliament encompasses a wide array of their parliamentary experience. Nonetheless, we argue that the expression of this marginalization on their parliamentary behavioural patterns will be mostly evident with regard to gender policy, due to the limited options available to them coupled with the opportunity to collaborate with other women on these issues and the ROIs (Returns on Investment) such behaviour can produce.

As women legislators cast off the fetters of marginality – be it by attaining parliamentary tenure, winning a party or a parliamentary role, joining the government, etc. – they may feel compelled to ‘prove themselves’ by affirming their competence in a variety of policy issues (Swers 2007; Swers 2013). In establishing their credibility in areas other than representing a marginalized group, they signal to their colleagues and voters that they need no longer be typecast. At the same time, the range of options and resources at their disposal widens, the value they ascribe to group representation diminishes (Bailer et al. 2022), and they are less liable to promote gender-related bills.

Based on the above theory, we set forth three hypotheses. First, drawing on Phillips's (1998) presence model and previous empirical studies, we hypothesize that women legislators will engage more than their male colleagues in the substantive representation of women.

H1: Compared to their male counterparts, women legislators will initiate more gender-related legislation.

It must be kept in mind, however, that the object of the present article is to further explain the relationship between women's descriptive and substantive representation. Specifically, we focus on women's marginalization in parliament and its effect in terms of their likelihood to substantively represent women by initiating legislation to advance their own group's interests. Accordingly, we expect that:

H2: Marginalized women legislators will initiate more gender-related legislation compared to their less marginalized counterparts. No such effect of marginalization will be observable for men.

As already stated, we assume that marginality stems from both individual and group-related factors – a premise that is of especial salience in studying legislators who belong to and represent disadvantaged groups. Both of these contributing elements need to be investigated; although mostly interlaced, they can often be differentiated. For example, a woman's long tenure in parliament diminishes her marginalization, but since women as a group are heavily underrepresented in political institutions, her group marginality is still extremely high – as is often true of tokenism. Alternatively, women's descriptive representation can be high, while the status of individual women legislators continues to be marginal, as is evidenced by their lack of power and resources (Barnes 2016).

We claim that, as concerns women, group marginalization, here measured as the numeric presence of women in the chamber and their respective parties, moderates the effect of individual marginalization on the substantive representation of women. Thus, women can escape marginalization and expand their legislative portfolio, but only in an institutional environment that does not strongly marginalize women as a group. Conversely, irrespective of their marginalization level, women will find it harder to break out of their marginalized position if their number is low as legislators in parliament or a party list. In such contexts, push factors constrain women legislators' choices despite their increasing individual power. Put differently, with respect to women in politics, the negative relationship between their standing on par with men and their likelihood to initiate bills promoting women's interests obtains only in institutional contexts where women are not marginalized as a group, notably when their share in a parliament or a party is sufficiently large.

When group marginalization is high, a change in the level of individual marginalization – and specifically, a decrease thereof – cannot be expected to exert a measurable effect on women legislators' likelihood to substantively represent women by initiating legislation in their favour. In making strategic decisions regarding their policy portfolios, women legislators will factor in both their individual and their collective status. When the latter is low, they continue feeling marginalized, which makes them less prone to leave the secure policy arena of gender-related concerns – where they are considered competent and credible – and make inroads into other, less safe, policy areas. Additionally, much has been written on women's cooperation and collaboration in parliament. One potential mechanism for this interactive effect aligns with insights from Barnes (2016), who demonstrates that marginalized women in parliaments often collaborate with other women, typically on issues of shared interest. This collaboration serves as a strategic response to combat their marginalization (see also Swift and Van der Molen 2021). However, when women legislators are less marginalized as a group, evidenced by higher descriptive representation, they perceive collaboration as less necessary and thus engage in it less frequently.

Similarly, we propose that when women legislators collectively face marginalization in parliament, they are constrained within the realm of gender-related policy issues, regardless of their marginality status. However, as their collective status improves, lessening the overall marginality, individual legislators are less reliant on collaboration with other women legislators as a strategy. Consequently, they are more inclined to broaden their policy portfolio beyond gender-related issues. Thus, we hypothesize that:

H3: Among women legislators, the positive relationship between individual-level marginality and the sponsorship of gender-related bills will be less pronounced in contexts with high- compared to low group marginalization.

The Israeli Context

Israel is a parliamentary democracy with a proportional representation electoral system and, thus, coalition governments. Despite women's active participation in the public sphere since the pre-state era, their power as a group has been consistently constrained (Chazan 2011; Izraeli 2003). While gender inequality is not unique to Israel, some of the socio-cultural features of Israeli society exacerbate it by sustaining and strengthening existing gender-based stereotypes (Herzog 1999; Herzog 2004). Israeli society embraces traditional gender roles (Kamir 2007), as both Jewish and Arab sectors ascribe great importance to family values (Halperin-Kaddari 2004).

Furthermore, Israeli politics pivots around security concerns, and the resulting discourse endorses existing gender-based stereotypes and further marginalizes women, who are regarded as less competent to deal with these issues. This is not a new phenomenon, nor is it limited to Israel, but in Israel it is aggravated by the geopolitical environment. The securitization of politics has intensified stereotypical sexist perceptions and, due to a spillover effect, women are often seen as less competent to engage in politics as a whole, not just in security matters (Rahat and Itzkovitch-Malka 2012).

Since the 1990s, women's descriptive representation in the Israeli parliament has been steadily on the rise. The four decades that preceded this increase were not marked by measurable change, with the number of women legislators ranging from a low of seven (1988) to a peak of eleven (1992). In the last 20 years, women's descriptive representation in the Knesset has more than doubled – from 11.6 per cent in the 15th Knesset (1999) to 25 per cent in the 25th (2022). It is noteworthy in this connection that ultra-Orthodox Jewish parties prevent women from seeking political office or serving as elected representatives. In all other political parties, including Arab parties nowadays, women are free to serve as representatives. Despite the impressive improvement in the descriptive representation of women, a glass ceiling is still palpable when it comes to ministerial positions or committee membership (Shapira, Kenig, and Itzkovitch-Malka 2016).

The few studies that address the descriptive-substantive link in the Israeli political arena show that, compared to men, women initiate more legislations that are gender-related (Shapira, Kenig, and Itzkovitch-Malka 2016) and fewer on national security (Itzkovitch-Malka and Friedberg 2018). Indeed, women have been active champions of parliamentary acts on gender issues, leading to considerable progress in gender equality and women's rights in Israel.

Nevertheless, gender is not a salient policy theme in Israeli elections or public opinion discourse (Shamir and Gedalya-Lavy 2017). While, in their platforms, parties may profess a commitment to gender equality or promise policy changes in this regard, in practice few of them address such issues as part of their operational agenda. A legislator who seeks to promote women's interests through a private member's bill does not attract much notice unless the initiative clashes with his or her party's agenda or with the coalition guidelines.³ To all intents and purposes, these legislators act as policy entrepreneurs, unencumbered by partisan constraints – a specificity that renders Israel an optimal case to study legislative initiatives designed to boost women's substantive representation.

Empirical Strategy

Data and Measurement

This study draws on an original dataset of Private member bills submitted in the Israeli parliament by all legislators over the 38 years between 1977 and 2015, spanning 9–19 Knesset terms. As the ratio of private legislation in the Knesset is extremely high, our dataset was created using a random sampling method involving 15,483 bills.⁴ We stratify the sampling by individual legislators and count the number of gender-related policy bills per legislator in a given term. Thus, the unit of analysis is legislator-term. The dataset contains 1,451 legislator-term observations and 600 individual legislators in total. Since an overwhelming majority of legislative initiatives are co-sponsored, some of the bills sampled are coded more than once under different sponsors, rendering some observations more influential than others. We correct this bias using robust standard errors, thereby assigning lower weights to influential observations.

Israel has one of the highest rates of private member bills in parliamentary democracies (Tuttnauer 2020), as Members of the Knesset (hereafter: MKs) have no institutional limits on the number of private member bills they submit. Cabinet ministers and deputy ministers cannot initiate private member bills. The number of private member bills submitted by MKs has tremendously increased since the early 1990s and reached record numbers of around 4,000 bills on average per term in recent terms. Over the last five Knesset terms, MKs submitted, on average, more than 100 private member bills every month (Cavari, Rosenthal, and Shpaizman 2023). Nevertheless, the success rate of these bills was not met with a similar increase. An average of terms 17–19 shows that only 4–5 per cent of PMBs were successful and enacted as state laws (The Knesset 2019).

Private member bills are used on account of their importance in agenda setting (Bratton and Rouse 2011). Sponsoring such bills enables legislators to articulate the interests of different groups, which is an extremely important aspect of representation regardless of the initiative's ultimate success in every given case (Barnes 2016; Escobar-Lemmon and Taylor-Robinson 2014; Osborn and Mendez 2010; Piscopo 2011). While a legislator's goal is to shape public policy by passing laws, the latter step is not a requisite for representation, which necessitates that the policy-making process should reflect different perspectives, priorities, and concerns (Swers 2013). Furthermore, research has shown that sponsorship contributes to the success and perceived effectiveness of a legislature (Barnes 2016; Krutz 2005; Weissert 1991). Israel's political system is characterized by an exceptionally high amount of private member bills, most of which are intended as declarative pronouncements. It has been well documented that legislators use the initiation of private bills to draw attention to their parliamentary activity (Nikolenyi and Friedberg 2019) and to express certain interests and preferences, thus declaring their commitment to a policy issue. While such expository gestures are not tantamount to policy outcomes, salient sources such as private member bills, floor speeches, questions, etc. have been extensively used as data to explore substantive representation (Catalano 2009; Gerrity, Osborn, and Mendez 2007; Lloren 2015; Schwindt-Bayer 2006).

First, all bills were read and classified thematically using Schwindt-Bayer's taxonomy (2006)⁵ by three human coders.⁶ Overall, 7.5 per cent of the bills analyzed fell into the category of gender policy (N = 1,099), as they were ‘(1) directly constructed as being of importance to women, (2) presented as affecting only women, (3) discussed in terms of gender difference, (4) spoken of in terms of gendered effects, or (5) framed in terms of equality between women and men’ (Celis et al. 2014, 159). Thus, for example, these bills targeted the issues of maternity leave (for men and women), violence against women, sexual harassment, women in the workforce, LGBTQ rights, women's personal status, etc.

Our dataset spans an extensive legislative history, revealing the dynamic evolution of its focus and character throughout the years. Legislative initiatives reflect a shift from an earlier emphasis on formal equality within liberal feminism to the nuanced perspectives of cultural and radical feminism, prioritizing substantive equality. Initiatives in the late 1970s and 1980s primarily focused on employment, emphasizing equal opportunities and parity in retirement. By contrast, initiatives in the 1990s took a holistic approach to address women's grievances, including issues of sexual violence, single-parent families, and LGBTQ rights. This trend continued into the twenty-first century with initiatives adopting a critical gender mainstreaming approach. Efforts included appointing advisors on women's status in municipalities, accounting for gender impact in legislation, and addressing enduring gaps in workforce segregation and political power.

Dependent variable

Number of gender-related policy bills per legislator. This is a count variable capturing the number of gender-related bills submitted by legislator per Knesset term (see Table A1 in the online Appendix for variables' descriptive statistics). Empirically, this variable ranges from 0 to 19 bills (M = 2.68, SD = 4.00).

Main predictors

Our main predictor is a legislator's gender, which is a dummy variable that takes the value of 1 for women and 0 for men. The dataset comprises 1,280 men and 171 women legislators. The other predictors are indicators of a legislator's marginality, at both the individual and the group level.⁷

Individual marginality: As mentioned above, individual marginality can stem from a variety of factors: personal marginality, captured by parliamentary tenure; party-related marginality, captured by list safety and parliamentary role, which are mostly decided upon by the party leadership; and structural marginality, captured by government/opposition role which reflects on the individual legislator.

1. 1. Parliamentary tenure (personal marginality). Measured by the number of terms served (M = 2.8, SD = 2.07). The most marginal legislators are those who are new to the parliamentary system and have yet to acquire parliamentary experience and expertise. One's marginality decreases with tenure.

2. 2. List safety (party-related marginality). The safety of a legislator's position on the list. This variable is contingent on (1) the candidate's rank on the party list, and (2) the number of seats the party had won in the last elections. This measure (the formula is presented in the Appendix; Shugart et al. 2021) ranges from 0 to 1: for the highest and lowest level of marginality, respectively. The least safe position is that of a legislator who, in the last general election, had won the last (0), and the safest – the first (1), seat on the party's list. (M = 0.49, SD = 0.32).

3. 3. Parliamentary role (party-related marginality). A dummy variable which captures whether the legislator holds a parliamentary or party role. A value of 1 stands for president or vice-president of the parliament, committee chairperson, or party leadership (party leader or secretary of a parliamentary party group) (M = 0.24, SD 0.42). Legislators who hold any of these roles are unequivocally less marginal than others.

4. 4. Government affiliation (structural marginality). A dummy variable capturing whether the legislator's party belongs to the governing coalition (1) or the opposition (0). We take legislators in the governing coalition – even those with no ministerial position – to be less marginal than the ones in the opposition (M = 0.63, SD = 0.48).

Group marginality:

1. 1. Women's descriptive parliamentary representation. The share of female legislators in the Knesset upon its instalment. This variable varies between 5 per cent and 22 per cent (M = 0.11, SD = 0.5).

2. 2. Women's descriptive partisan representation. The share of women legislators within the legislator's party, which could differ greatly from the overall representation of women in the Knesset (M = 0.13, SD = 0.12).

Control variables

We control for three factors that are known to influence legislative behaviour in modern democracies.⁸ One such variable is prior experience and expertise (coded dichotomously for stereotypically masculine occupations such as armed forces, business, corporate, and general management, versus all other professions).⁹ Another is a legislator's party affiliation, measured as a categorical variable classifying legislators into right-wing, center-left, ultra-Orthodox, or Arab parties. The third is a legislator's ministerial position.¹⁰ We also control for the Knesset term, as it is important to account for temporal variation in gender policy trends.

Models and Estimation

Our dependent variable is the number of gender-related private member bills initiated by MK per term. As a result, we utilize a regression model for count data – negative binomial regression models – which is appropriate for over-dispersion count data (Cameron and Trivedi 2013). While this is not the only modelling strategy available, it has been used in similar works (see, for example: Bailer et al. 2022; Senk 2023). Poisson models proved less appropriate in this case due to zero inflation and over-dispersion of the dependent variable. We account for autocorrelation in our data caused by time dependencies. Following Bailer et al. (2022) we model time dependencies using a cubic spline for the time dimension to reduce autocorrelation. We also use robust standard errors to account for the clustering of observations caused by the data structure.

The empirical analysis follows two trajectories. We first run a simple model to validate the link between women's numerical presence and the substantive representation of women in parliament. We then run a series of moderation models, where we interact gender with various individual- and group-level marginality indicators to examine if and how a marginal status moderates the connection between gender and the substantive representation of women. In the second stage of the analysis, we examine the combined effect of group- and individual-level marginality on the behaviour of women legislators. To illustrate the substantive effect of our analysis, we present predicted probability/marginal effects plots using the observed-value approach (Hanmer and Ozan Kalkan 2013).

Results

We begin by empirically examining the basic premise of the presence model (Phillips 1998), according to which, in the policy-making process, individuals will be more inclined to represent the preferences and interests of the group whose descriptive features they share. Figure 1 clearly shows that this expectation is borne out in the case of women: as per H1, Women MKs are more than twice as likely to sponsor gender-related legislation compared to their male colleagues.

Figure 1. Difference between women and men in sponsoring gender-related bills.

Note: The predicted gender-related bills for women and men legislators between 1977 and 2015. The analysis draws on Model 1 in Table A2 in the Appendix. Control variables are set to their observed values. Capped vertical lines represent 95 per cent confidence intervals.

Figure 1 shows a strong and significant effect on gender. Women legislators are three times more likely than men legislators to sponsor gender-related policy bills. While this finding is not surprising, it corroborates H1 and lends further support to the behavioural gender differences established between men and women legislators.

Next, we present the results from a series of negative binomial regression models that interact gender with various marginality indicators and show how marginality influences the descriptive-substantive link. Figure 2 presents the differential effect of parliamentary tenure, a proxy for personal marginality, on the link between gender and the substantive representation of women. To understand the interaction dynamics, we plot the predicted number of gender-related bills for men and women given different levels of parliamentary tenure. The figure shows that the effect of tenure on men legislators' likelihood to sponsor gender-related bills is not statistically significant; however, for women legislators, a negative and statistically significant effect was found, especially when comparing the lowest levels of tenure with the highest. A plausible inference would be that, with fewer resources at their disposal, women legislators with little or no parliamentary experience will strategically choose to promote gender-related policy issues in their legislative portfolio to gain electoral and public support. With a longer tenure, they acquire more experience and resources, which allows them to step out of the gender niche and diversify their legislative initiatives.

Figure 2. Predicted gender-related bills by the duration of parliamentary tenure and gender.

Note: Predicted gender-related bills for women and men legislators across levels of parliamentary tenure (measured as the number of Knesset terms served). The analysis draws on Model 2 in Table A2 in the Appendix. Control variables are set to their observed values. Capped vertical lines represent 95 per cent confidence intervals.

Both individual party-related marginality indicators – parliamentary role and list safety – did not have a statistically significant effect on the sponsorship of gendered bills by women MKs (see models 3 and 4 in Table A2 in the online Appendix). While this is quite surprising, we can assume that the reason for this null effect has to do with the nature of both indicators, on which party leadership holds great influence. Holding a prestigious parliamentary or partisan role, or a safe position on the party list, makes legislators less autonomous in their behaviour, as such positions are governed by the party leadership. Thus, it could be that the effect of marginalization on legislators' behaviour is not fleshed out by these two indicators.

As for individual structural marginality, Fig. 3 confirms our expectation that women MKs who are affiliated with the opposition would be more prone to substantively represent women, compared to women affiliated with the coalition, while men MKs would not be susceptible to this difference. As already stated, we attribute this discrepancy to women's marginality at the individual level: opposition members enjoy far fewer resources and opportunities compared to coalition members and, even, backbenchers. It goes without saying that the tendency on the part of these women MKs to eschew gendered legislation has a significant impact on gender policy. A tenable inference regarding women MKs would, therefore, be that, in line with our theoretical expectations, one's marginality impels one to appeal to one's female electoral base to demonstrate one's competency as a legislator and promote one's image. For women coalition members, this incentive is less strong: they are incentivized to show that they can competently handle a wide range of policy issues, not only gender concerns.

Figure 3. Predicted gender-related bills by coalition affiliation and gender.

Note: Predicted gender-related bills in the 9–19 Knesset terms by government affiliation. The analysis draws on Model 5 in Table A2 in the Appendix. Control variables are set to their observed values. Capped vertical lines represent 95 per cent confidence intervals.

Next, we examine the effect of women's parliamentary group marginality on women MKs' proclivity to sponsor gender-related bills. Group marginality is measured as women's share in the parliament as a whole and in the legislator's own party. For a disadvantaged minority group such as women, descriptive representation usually, though not always, corresponds with its members' status in society and, in the case in point, also with general perceptions of gender roles and egalitarian values (Wängnerud 2009). Thus, this indicator is sufficiently inclusive to capture women's marginality not only in parliament but also in society at large. For women legislators, low levels of descriptive representation are clearly a disadvantage. Under these circumstances, offsetting their marginality through collaborations with other women (Barnes 2016) becomes less of an option. Isolated and powerless, women MKs are liable to turn to gender-related legislation. An increase in women's descriptive parliamentary representation, however, opens up opportunities for a division of labour, such that gender-related issues are pursued by choice rather than out of lack of choice.

Figure 4 confirms this expectation. With an increase in women's descriptive representation in parliament (left figure) or in a party (right figure), women legislators are less likely to sponsor gender-related bills.

Figure 4. Predicted gender-related bills across levels of women's descriptive representation in parliament/party.

Note: Predicted gender-related bills in the 9–19 Knesset terms across levels of women's descriptive representation (in parliament: left figure; in legislator's party: right figure). The analysis draws on Models 6 and 7 in Table A2 in the Appendix. X-axis values follow the dataset observed values. Control variables are set to their observed values. Capped vertical lines represent 95 per cent confidence intervals.

The effect of women's descriptive representation in parliament might be attributed to differences in feminist attitudes among early- and later-elected women legislators. We explore this by examining party ideology, finding no significant difference in the share of centre-left women elected early (65 per cent, 25/83) versus later (62 per cent, 38/133). Additionally, it should be noted that we control for party ideology, meaning that the models themselves also account, in some way, for this alternative explanation. Nevertheless, since party ideology is only a rough proxy for feminist attitudes, we cannot fully reject this alternative explanation. We address this point in the concluding remarks.

Taken together, the analysis presented above lends strong support to H2. Based on most of the indicators of marginality used in the analysis, both individual and collective, marginal women legislators initiate more gender-related bills compared to their counterparts.

The Combined Effect of Group- and Individual-Level Marginalization on the Behaviour of Women Legislators

From this point onwards the analysis focuses on the subset of women legislators only. Figure 5 is a descriptive display of the relationship between individual marginalization (measured by parliamentary tenure) and the number of gender-related bills sponsored by women legislators. It was created using a scatterplot and a linear best-fit line. Data is divided by high and low levels of group marginality. The two rows in Fig. 5 relate, respectively, to high and low levels of group or collective marginality of women (that is, [relatively] low and high levels of descriptive representation), and the charts in the top and the bottom row present distinct and markedly different trends. Under high levels of group marginality, the relationship between women's marginality and the number of gender-related bills is negligible, in both the parliamentary and partisan contexts. Put differently, under these conditions, individual marginality does not affect women's sponsoring of gender-related bills. However, under lower levels of collective marginality, such a relationship becomes clearly discernible: the less marginal women MKs are less likely to sponsor gender-related bills.

Figure 5. Number of gender-related bills sponsored by women legislators, by parliamentary tenure, in high and low collective marginality contexts.

Note: Trendline between parliamentary tenure and the number of gender-related bills submitted by legislators. 95 per cent Confidence Intervals. Figure is descriptive, and generated based on the raw data. It includes women legislators only. To construct high/low descriptive representation in parliament/party we used the mean value of the variables (for descriptive representation in parliament M = 0.10, low representation < 0.10, high representation ≥ 0.10; for descriptive representation in party M = 0.13, low representation < 0.13, high representation ≥ 0.13).

We corroborated this descriptive finding using robust multiplicative interaction models (Hainmueller, Mummolo, and Xu 2019) to examine the marginal effect of parliamentary tenure, as an individual marginalization indicator, on women legislators' likelihood to sponsor gender-related bills, with group marginalization as a moderator. Figure 6 thus presents the marginal effect of tenure on the initiation of gender-related bills given the changing levels of women's descriptive representation in parliament or in their party. This analysis, too, is based on the female subset of the sample only. Importantly, we do not use all individual-level marginality indicators for the interaction between individual and collective marginality but, rather, focus on tenure, as it proved to be the most consistent indicator of individual marginality in our previous analyses.

Figure 6. Estimated marginal effect of parliamentary tenure on the number of gender-related bills initiated by women legislators.

Note: Marginal effect of tenure on gender-related bills sponsored by women legislators (vertical axis) across levels of women's descriptive representation in parliament (left hand) and in the party (right hand). The left panel is based on model 1 in Table A3 in the Appendix, the right panel is based on model 2 in Table A3 in the Appendix. Bars are a histogram distribution of the moderator (women's descriptive representation in parliament/party). Marginal effect was generated using the observed-value approach, control variables are set to their observed values. Marked are 95 per cent confidence intervals.

According to Fig. 6, under high levels of women's group marginality – that is, when women's descriptive representation is low – either in parliament (left-hand graphs) or in a party (right-hand graphs), lower levels of individual marginality, proxied by parliamentary tenure, do not translate to fewer gender-related initiatives. However, in both the left- and right-hand charts, as we move right on the horizontal axis and women's marginality as a group decreases, the relationship between parliamentary tenure and gender-related bills becomes apparent: the longer the parliamentary tenure, the fewer gender-related bills are initiated – confirming our finding regarding H2. Thus, the relationship between women legislators' individual marginality and their preoccupation with gendered legislation only holds under low collective marginality, as per H3. When women are marginalized as a group, it is harder for them to diversify their legislative efforts regardless of personal standing; thus, in pursuit of political benefits, they tend to engage in gender-policy issues.

Furthermore, group marginalization influences women legislators' parliamentary behaviour both directly and indirectly. The former effect is captured by H2, whereupon we established that, as group marginalization decreases, women legislators are less likely to sponsor gender-related bills. In addition, however, collective marginalization conditions the effect of individual marginalization in this respect, limiting it to contexts of low marginalization alone.

Alternative Explanations

The trends obtained in the analysis could indicate that marginality increases a legislator's propensity to sponsor bills in general, not just those related to gender policy. In other words, if increased marginality per se encourages legislators to sponsor more bills, this should be true for a variety of policy issues. As recalled, we argue that it is not that marginalization only occurs around gender issues, but that its behavioural expression will be most evident with regards to gender related policy issues, due to the ROIs it provides women legislators. To support our argument and address these other possibilities, we conducted robustness checks by rerunning our models with different dependent variables, focusing on thematic policy issues like welfare, economy, and security. The results, detailed in Table A4 in the Appendix, do not show a clear trend. Marginalization does not seem to drive women legislators' bill initiation on any issue besides gender policy: an increase in women legislators' marginalization does not lead to increased bill initiation on welfare, education, or security, for example, and a decrease in women legislators' marginalization does not lead to decreased bill initiation. If it is marginalization per se that drives bill initiation, it is reasonable to assume it would affect women legislators' bill initiatives across policy issues and not just around gender-related ones. Given that we do not see these trends in any other policy issue but gender-related policy, this analysis helps mitigate concerns about alternative explanations, reinforcing the strength of our proposed argument, which suggests that the expression of marginalization on women legislators' parliamentary behavioural patterns is mostly evident with regard to gender policy.

Discussion and Conclusions

This article has addressed the complex relationship between the descriptive and substantive representation of women, and how the likelihood for a women legislator to promote the latter is conditioned by her marginal status within parliament. Our analysis has shown that individual-level marginality is positively correlated with the sponsoring of gender-related bills: marginalized women legislators initiate more such bills than their less marginalized counterparts. The effect was also found in connection to collective marginality, using women's descriptive representation in parliament or party as a salient indicator. That said, in the final stage of the analysis we demonstrated the complex nature of the relationship between marginalization at the individual and collective levels, such that the latter moderates the effect of the former on women legislators' propensity to substantively represent women. Put simply, under high levels of collective marginalization, a decrease in women legislators' individual marginalization does not diminish their likelihood to sponsor gender-related bills, as it does when their collective marginality is low. We attribute this effect to the overall gender climate in the Knesset. Collective marginalization impedes women legislators – even those whose personal status is secure – from leaving the comfort zone of gendered issues and venturing into other policy areas that do not accord with prevailing stereotypes.

Using women MKs as a case study, this article has shown that a legislator's incentive to represent the disadvantaged group to which they belong is strongly conditioned by their marginal position in parliament. However, this dynamic is far from simple, as one should take into account both the individual and the collective marginality of women legislators. Figure A1 in the Appendix demonstrates this complexity well. The figure presents the share of gender-related bills out of the total number of bills per Knesset term. No clear trend is apparent. Given the theoretical arguments we put forward in the article, we believe this is not surprising. Given the rise in the descriptive representation of women in the Knesset, more women are entering parliament. Thus, women as a collective become less marginal, a process we expect will lead to a decrease in the number of gender-related bills women initiate, especially those who are not individually marginalized. However, at the same time, the newly elected women are newcomers and, therefore, suffer from individual marginality. Therefore, they are expected to initiate more gender-related bills. These conflicting trends might cancel one another and hinder the possibility of seeing any clear trend at the macro-level data.

Our findings are subject to some scope limitations. While we do not test this empirically, we assume that the effect will likely be less strong among women legislators with rigid preferences regarding gender representation, from both ends of the continuum. For instance, for women legislators who hold a strong commitment to represent women voters, becoming less marginal in parliament will probably not make them let up on their efforts to advance gendered legislation, but rather redouble them. On the other side of the continuum would be those who do not set great store on women's issues. A change in the personal status of such a legislator would probably have no effect on her efforts to substantively represent women in parliament. Admittedly, we assume that the effect revealed in this investigation is based on the behaviour of women legislators in the middle range of this scale, who factor in their marginality when strategically choosing which policy issues to promote. This assumption lends itself to further research.

Another important scope condition has to do with the case studied in this research. We consider the Israeli case as ‘most likely’ for the examination of the connection between women legislators' marginalization and the substantive representation of women. While this study is not comparative, we believe it can be relevant to other national contexts. Israeli women legislators may be more marginalized than their counterparts; nevertheless, women legislators in Israel hold 25 per cent of the seats in the national parliament, which is similar to the global average, and places Israel in the middle of the global ranking (IPU, 2023). Moreover, the marginalization of women in politics is a far-reaching global phenomenon (Barnes 2016). Most parliaments are gendered institutions disadvantageous to women trying to enter politics and navigate their political careers. Thus, it is likely that women in other national contexts will also be affected by their marginality status while making behavioural choices. Scholars of women and legislative studies can adapt our empirical approach to test the generalizability of our findings in additional contexts.

Observed from a broad perspective, the individual-level trends described above could suggest that, with a decrease in their marginality, women legislators will be less liable to substantively represent women by sponsoring gender-related bills. However, this alleged tension can be partly resolved by thinking slightly differently about the implications of this study. A plausible interpretation of our findings – which indicate less promotion of women's issues by women legislators when the latter are less marginalized – might be explained by the increased success of less marginalized women in championing their proposals. In such a scenario, they may not find it imperative to repeatedly introduce similar legislation. This explanation is echoed in Barnes’ (2016) work, which posits a compelling argument suggesting that women engage in greater collaboration due to their marginalized status, and this cooperative approach contributes to their success. A noteworthy implication of her research is the expectation that, as women experience reduced marginalization, their collaborative tendencies may decrease. This decline could be attributed to the diminished necessity for utilizing collaboration as a tool to advance legislation, particularly when they are no longer facing significant marginalization. These dual insights potentially harmonize the apparent tension, suggesting that marginalization influences women legislators' behaviour by hindering success in policy initiatives, thereby prompting persistent attempts to achieve success. Conversely, when women are less marginalized, this dynamic undergoes a transformation. This explanation does not contradict our theory regarding the effect of marginalization of women's legislative behaviour but, rather, suggests the existence of an additional underlying influence or dimension. Thus, from a macro-level perspective, the substantive representation of women will not necessarily be harmed as women become less marginalized.

Finally, it should be noted that future research should further explore the mechanisms underlying our main findings. Since the trends we obtain in this research are based on correlational data, face-to-face interviews, or experiments, it can help unravel the reasons why marginalized women legislators focus more on gender-based legislators compared to less marginalized ones. It is hoped that the trends presented in this research, which raise compelling questions regarding the patterns of women's substantive representation, will be further pursued in future research.