Under what conditions do female politicians use their power more than male politicians to help women? This dissertation contributes to longstanding debates over descriptive and substantive
representation by drawing on cases of female and male presidential candidates and winners in Latin America. Turning first to concepts of representation, the chapter acknowledges criticisms of descriptive representation, including debates over its lack of accountability and its tendency to essentialize identities and interests. The chapter then elaborates on the stakes at hand in establishing a conditional link between descriptive and substantive representation. Empirical research has shown that the presence of women in political office positively correlates with and even causes policy outcomes favoring women, but women’s use of power to advance PWC is uneven. A theory of the probabilistic relationship between descriptive and substantive representation in the case of gender ideally would avoid essentialism and explain (1) why female politicians are more likely than their male counterparts to deploy their power to promote PWC (inter-gender variation); and (2) why some female politicians do so more than others (intra-gender variation).
Dominant accounts from political science argue that the number of female legislators, partisan ideologies, and formal powers determine divergences in women’s use of power to advance PWC, and yet each of these factors varies little in the cases of Bachelet and Rousseff. These theories also assume that rather than explain why female politicians would use their power more than their male counterparts to advance PWC. Socio-psychological theories would attribute variation in women’s use of power to advance PWC to the presence or absence of feminist consciousness. According to research from these fields, female politicians would be more likely than their male counterparts to benefit from feminism and be exposed to feminist ideas, but only women with nontraditional life experiences tend to develop feminist consciousness, preferences and attitudes. Bachelet and Rousseff’s biographies suggest that both made a series of highly nontraditional choices in both their private and professional lives, providing strong circumstantial evidence of their feminist leanings. While socio-psychological could offer an
account of both inter- and intra-gender variation, it falls short of adequately explaining the Bachelet- Rousseff puzzle.
The chapter then elaborates a novel, constituency-centered theory for the probabilistic link between descriptive and substantive representation. While the mainstream literature on the U.S. Congress emphasizes constituency influence on politicians’ decision-making, most studies looking at representation and gender so far have underappreciated this factor. I will argue that gendered and sex-related
characteristics of core and personal constituencies function as probabilistic mechanisms linking
politicians’ sex to the likelihood of using power to advance PWC. Compared to many existing theories, my constituency explanation better avoids the problem of essentialism and explains both inter-gender and intra-gender variation, providing new insights on the conditions under which descriptive improves substantive representation.
1. What Is at Stake? What Do We Know?
To many inside and outside of academia, the value of the mere presence of women in government is far from obvious. Further motivating the dissertation’s research question, this section describes the importance of understanding the link between descriptive and substantive representation and then reviews the state of empirical knowledge. I begin with Pitkin (1967) because having first defined descriptive representation, she remains one of its major critics. To understand the true meaning of representation, she identifies four types—formal, symbolic, descriptive and substantive—offering unique conceptual
advantages to understanding democracy. While formal and substantive forms require “acting for” constituents, symbolic and descriptive variants involve “standing for” them.
Formal representation refers to authorization and accountability through institutions such as elections. According to Pitkin, a downside to formal representation is that it does not specify what
representatives do. A key theme for her is that if we know what representatives do, then we can figure out what representation is. Symbolic representation also falls short in the sense that it relates to being rather than doing. It largely depends on whether constituents “believe in” their representatives and how they
emotionally respond to them (102). Since these representatives create symbols, this variant moves from a static to a dynamic concept of representation.
Many proponents of descriptive representation cite Pitkin (1967) as the originator of the concept, but do not acknowledge her criticisms of it (Alexander 2012; Sanbonmatsu 2003; Schwindt-Bayer and Mishler 2005). Pitkin finds fault in the static nature of descriptive representation, as descriptive “representing means being like you, not acting for you” (89). Anyone who looks like a constituency group member could play the role of a descriptive representative. She further notes the absence of an accountability mechanism: nothing can ensure that descriptive representatives will make decisions in ways desired by constituents. A final problem is that descriptive representation diverts attention from the activities of the representative to the composition of the government body. In sum, descriptive
representation, while still necessary to fully conceptualize representation, is less valuable to democracy because it is static, lacks accountability, and is unconcerned with representatives’ activities. In light of these shortcomings, descriptive representation could gain value if empirically linked to the form which Pitkin appears to favor most, substantive representation, which requires acting on constituents’ behalf (Dovi 2014).
Other scholars critique descriptive representation’s essentialism, or the premise that all women are fundamentally similar because of their common experiences, and hence identities and interests. Phillips (1995, 53) argues that the idea that any woman could represent all women invites essentialism. Childs and Krook (2008a, 101) agree that women are too diverse to assume that they are essentially similar. “If women have multiple identities and experience the world in different ways, the basis upon which women can act for other women seems by necessity either to succumb to essentialism or be underpinned by a denial of women’s differences.” These arguments diminish descriptive representation’s intrinsic value. Yet, defenders of descriptive representation—often those in favor of quotas for
historically marginalized groups—insist that presence matters (Franceschet, Krook, and Piscopo 2012). Scholars have argued for a greater presence of women in politics for other reasons aside from enhanced substantive representation—namely on the basis of pure justice and greater democratic legitimacy (S. L.
Dovi 2007; Phillips 1995). Yet, the prospects of more action on behalf of women remains one of the most intuitive and compelling motivations behind greater descriptive representation for women.
These debates have motivated a line of research on the conditions under which descriptive representation of women and other historically marginalized groups “works.” Given their diverse experiences and interests, when do descriptive representatives— to a greater extent than non-descriptive representatives—act on behalf of in-group constituents? Enhanced communication and trust are
important, albeit contingent, benefits of descriptive representatives (Mansbridge 1999). “Good”
descriptive representatives also are theorized to possess in-group ties, or relationships with disadvantaged sub-groups within the broader historically marginalized group (S. Dovi 2002; S. L. Dovi 2007). An example of this in the case of gender would be low-income women and organized groups that work to advance the interests of low-income women. Networks emerge as a conditional linkage between descriptive and substantive representation.
To sum up, while the value of substantive representation goes largely unquestioned, the
advantage of descriptive without substantive representation remains more controversial. Although there may be other benefits intrinsic to women’s political presence such as pure justice and democratic
legitimacy, substantive representation remains one of the most commonly cited ones, intuitively appealing to much of the broader public. The stakes are high for empirically demonstrating this positive relationship and for specifying the conditions under which it is most likely to hold.
Given the stakes at hand, it is unsurprising that an enormous body of empirical research explores this topic. Both observational and experimental studies have established a strong, probabilistic link between descriptive and substantive representation but so far have focused almost entirely on legislatures. Vast research in the U.S. demonstrates that female legislators are more likely than their male counterparts to deploy their power for the advancement of explicitly feminist issues or those related to women’s interests, such as social welfare (Bratton and Haynie 1999; Dodson 2006; Gerrity, Osborn, and Mendez 2007; MacDonald and O’Brien 2010; Osborn 2012; Reingold 2000; Saint-Germain 1989; Swers 2002; Swers 2013; S. Thomas 1994; Wolbrecht 2000). Some of the most prominent works includes Sue
Thomas’ (1994) and Michele Swers’ (2002; 2005; 2013; 2014) studies of women in the U.S. Congress.i Influential works examining the impact of women in U.S. state legislatures include Reingold’s (2000) study on the California and Arizona legislatures; Saint-Germain’s (1989) examination of the Arizona legislature; and Bratton and Haynie’s work on six U.S. state legislatures (1999). In short, the most influential research on the link between descriptive and substantive representation focuses on state or national congresswomen in the U.S.
Other works have examined European cases (Bratton and Ray 2002; Dahlerup 1988; Norris 1997; Wängnerud 2009), but far fewer studies outside of the developed world have explored the relationship between the presence of female leaders and women-friendly policy outcomes. The most frequently-cited comes from developmental economics.ii Chattopadhyay and Duflo’s (2004) study, which leveraged randomized gender quotas in local councils in India, significantly advanced scholars’ confidence in a causal relationship between descriptive and substantive representation. A limitation of this kind of experimental study is of course external validity. Strengthening scholars’ confidence in the validity of these findings, other studies have examined the link between descriptive and substantive representation for legislatures in Latin America (Franceschet and Piscopo 2008; Schwindt-Bayer 2010).
Virtually all of these studies produce positive results, but acknowledge that some female
politicians use their power much more than others to advance PWC. Perhaps more than any other scholar, Dodson (2006, 8) has stressed variation in women’s impact: “Although feminist empirical scholars’ choice of words may have sometimes (unfortunately) conveyed an image of unity and consistency among women … feminist empirical researchers have long realized that…the relationship between gender and attitudes/behaviors is probabilistic rather than deterministic.” She points out that this research also shows that the impact of politicians’ sex changes over time and according to specific issues (e.g. social welfare vs. explicitly feminist policy, see for example, M. C. Escobar-Lemmon, Taylor-Robinson, and Schwindt- Bayer 2014)).
Scholars therefore have offered compelling observational and experimental evidence that women tend to, but do not always, use their power more than men to promote PWC. There are two main
conclusions to this review of the empirical knowledge. First, an important shortcoming in this literature is that it has yet to fully explore the link between descriptive and substantive representation in the executive branch, which in many countries is actually more powerful than the legislature. Second, any new theory on the probabilistic linkage between descriptive and substantive representation should not only explain variation in women’s use of power to advance PWC (intra-gender differences), but also whether and why female politicians are more likely than their male counterparts to do so overall (inter-gender differences). I now will examine the contributions and shortcomings of existing theories on variation in women’s impact from political science and socio-psychological research.
2. Political Science Theories for Variation Among Female Politicians
Dominant theories from political science tend to argue that factors such as women in the
legislature, party ideology and formal powers either facilitate or curtail women’s use of power to advance PWC. Examining these variables reveals that each is too similar in the cases of Bachelet and Rousseff to adequately explain their dramatic divergence. Exclusively focused on intra-gender variation, these theories assume rather than explain why female politicians would be more likely than their male counterparts to advance PWC in the first place. As mentioned above, a new theory of the probabilistic relationship between descriptive and substantive representation should be able to explain both intra- and inter-gender variation.
Women in the legislature The first factor that may enable or hinder women’s pursuit of PWC derives from critical mass theory, one of the most cited and controversial accounts of women’s varying impact. Proponents argue that a greater proportion of female legislators fosters a more pro-women legislative climate and enhances possibilities for supportive alliances (Childs and Krook 2008b; Dahlerup 1988). Female representatives may be more likely to push for PWC bills when they observe greater numbers of female legislators because they anticipate greater chances of success (Thomas 1994). Critical mass theory predicts that once the percentage of women in legislatures reaches a certain threshold, for example 30 or 40%, then policies improving women’s status are more likely to be debated and passed.
Some empirical studies show that more women in U.S. state legislatures and European parliaments lead to different policy priorities (Gertzog 1995; Saint-Germain 1989; Thomas 1994).
Typically applied to legislatures, critical mass theory could be adapted to explain variations in female executives’ use of power. The argument would acknowledge that male and female presidents decide which legislation to send to Congress after strategically considering the legislation’s chance of success. Female presidents who observe a significant number of potential allies would anticipate a favorable reception for PWC bills and thus would send more of them to Congress. Female presidents would then be more likely to promote PWC when the number of women lawmakers achieves the critical mass threshold.
A major limitation of this theory is that it assumes rather than explains why female presidents— like female legislators—possess greater motivations than their male counterparts to pursue PWC. It therefore provides a theoretical account only of intra-gender but not inter-gender variation. Moreover, politics and gender scholars have repeatedly noted its conceptual flaws. Exactly what proportion
constitutes a critical mass? Several studies have found results that contradict expectations (Carroll 2001; Crowley 2004; Reingold 2000). Difficulties in pinpointing a threshold combined with these empirical anomalies have prompted some scholars to abandon the theory (Childs and Krook 2008b).
Critical mass theory also would have problems explaining the Bachelet-Rousseff divergence. Both presidentas faced low numbers of women in the legislature during their first terms. In Chile from 2006-10, only 15% of the Chamber of Deputies and 5.3% of the Senate were comprised of women, averaging 10.2%. In Brazil from 2011-14, 8.6% of the Chamber and 16% of the Senate were women, averaging 12.3% (Inter-Parliamentary Union 2016). Bachelet’s first administration coincided with more female legislators in the Chamber of Deputies than in the Senate while the opposite was true for
Rousseff’s first term. Yet it is unclear why this difference would matter because both chambers are needed to pass legislation. More relevant is the fact that these percentages fall significantly below the 30- 40% mark required for expectations of a policy impact. The proportion of women in the legislature cannot fully account for the Bachelet-Rousseff divergence.
Party ideology Partisan ideology may act as a powerful determinant of the degree to which female politicians use their power to promote PWC. Women from the left are more likely to deploy their power to advance PWC than women on the right in part because women’s movements in the U.S., Europe and Latin America historically have associated with the left (Beckwith 2000; Ríos Tobar, Godoy Catalán, and Guerrero Caviedes 2004; Jaquette 2009).iii Left-leaning female politicians will likely have more ties to feminist networks than female politicians who hold conservative ideologies. Relatedly, politicians with PWC preferences are more common in left-leaning parties in Latin America (Beckwith 2000; Morgan and Hinojosa 2015; Roza 2010; Schwindt-Bayer 2010).iv
Partisan ideology appears as a major predictor of variation in women’s use of power to promote PWC, but both Bachelet and Rousseff’s parties appear ideologically similar. The limited data available indicate that center-left parties, providing these presidentas’ partisan base, featured similar degrees of women-friendliness (Morgan and Hinojosa 2015; Murillo, Oliveros, and Vaishnav 2010). The manifestos of Bachelet’s Socialist Party (PS) and Rousseff’s Workers’ Party (PT) have made a similar number of gender equality mentions, and depending on the indicator, anywhere from 20-30% of the leaders within these parties are female. Furthermore, evaluated on a 1-10 scale where higher numbers indicate greater progressiveness, the PS earned a score of 3.2 while the PT scored of 3.8 in terms of their stances on abortion issues; and they scored of 3.5 and 5.7 in terms of their stances on social issues (Morgan and Hinojosa 2015). Relevant characteristics of the PS and PT are more similar than different, and some of the differences that do exist suggest that Rousseff would advance more PWC than Bachelet. Party ideology therefore cannot easily account for Bachelet’s and Rousseff’s divergent use of power.
A comparison of Bachelet’s and Rousseff’s coalitions further suggests more similarities than differences in the pro-women positions of these parties. In 2006, Bachelet was backed by the PS and the Party for Democracy (PPD). She also sustained crucial support from the socially conservative Christian Democrat Party (DC). Similarly, in 2011, Rousseff was primarily backed by the PT, but allied with the more conservative (or at least, less ideologically-oriented) Party for the Democratic Brazilian Movement (PMDB). Without DC collaboration in Chile and PMDB collaboration in Brazil, probably neither
presidenta would have won their elections—much less pursued their policy agendas once in office. Other statistics on women’s presence in parties reveal additional similarities in terms of their women-
(un)friendliness. In recent years, just 18% and 17% of DC and PMDB leaders have been women, respectively, and only 11% of both parties’ congressional candidates and 10% of their elected legislators have been women (Morgan and Hinojosa 2015). Bachelet and Rousseff therefore appear similarly constrained by their parties’ and coalitions’ ideologies.
Institutions and formal power Many political scientists argue that formal institutions shape politicians’ decision-making. Institutions distribute power, and politicians with more power may pursue their a priori policy goals to a greater degree than those with less power. Again holding constant women’s PWC motivations, institutional theories suggest that female policymakers with more formal power will push for more PWC than those with less power.
Some empirical evidence from the U.S. and Latin America helps confirm the institutional expectation. Swers (2005) finds that female congresswomen increase their activity on social welfare issues (deemed a PWC issue in the U.S. context) when their party gains majority status. Others have shown that the marginalization of women in Latin American legislatures curtails their powers, and by extension, their impact on PWC (Barnes and O’Brien 2014; Heath, Schwindt-Bayer, and Taylor- Robinson 2005; Schwindt-Bayer 2010). In sum, stronger institutional powers seem to foster a greater pursuit of pro-women policies by female policymakers.
Although more formal power could augment female politicians’ pursuit of PWC, this explanation has trouble accounting for the Bachelet-Rousseff puzzle. Presidents in both countries not only draft legislation but also sets the legislative agenda. They also seem to use their legislative powers to comparable degrees—sending a similar number of bills to Congress (about 94 annually in Chile from 2000-10 and 96 annually in Brazil from 2003-14). They also enjoy similar success rates, ranging from 65% to 78%.v In short, while plausible, institutional explanations also fail to account for the Bachelet- Rousseff puzzle. Like the critical mass and partisan ideology accounts, this explanation does not tell us
why we should expect female politicians to use their power more than their male counterparts to advance PWC.
3. A Feminist Consciousness Explanation for Variation among Female Politicians
Unlike these dominant theories from political science, theories of how individuals acquire feminist consciousness, preferences and attitudes may be able to explain both intra-gender and inter-