Abstract
Are women in office more likely to providvae access to women’s lobby groups than men in office? If so, how can women’s strategic lobbying increase the responsiveness of male legislators? This paper presents a field experiment examining how women and men in state legislatures respond differently to women’s organizational lobbying. My findings suggest that substantial gender gaps do exist; women are twice as likely to respond to a women’s issue group’s simple meeting request. That said, meeting requests signaling constituent mobilization have heterogeneous effects across legislator gender, doubling the likelihood that a male legislator will respond and effectively closing gender gaps in responsiveness. My results identify how women’s lobbying can employ distinct lobbying strategies on descriptive and nondescriptive representatives to successfully gain their attention. In distinguishing differing pathways toward maximizing opportunities for women’s organizational inclusion in policymaking, this paper importantly informs women’s groups lobbying in state legislatures, wherein low levels of descriptive representation often persist.
Studies of women’s representation largely confine their focus to mechanisms of descriptive representation, often proposing that women are best served by elected officials with whom they share a marginalized gender identity (Phillips 1998; Sapiro 1981; Williams 1998). This focus is well-merited; an impressive body of literature demonstrates that legislators are likely to take increased interest and action on behalf of constituent groups that share their descriptive characteristics (Broockman 2013; Burden 2007; Butler and Broockman 2011; Canon 1999; Carnes 2012; Grose 2005; Reingold 1992, 2000; Whitby 2000). Women in politics bring gendered life experiences to political institutions, ultimately reflecting different interests than their male counterparts (Gilligan 1992; Mansbridge 1998; Sapiro 1981; Thomas 1991). Research suggests that political activity by female legislators is substantively different than male legislators otherwise similar in partisanship, seniority, and district characteristics (Dodson 2006; Reingold 2008). In state legislatures specifically, Holman and Mahoney (2018) find that especially with the presence of a woman’s caucus, increases in women’s aggregate descriptive representation leads to increases in women’s collaboration on women’s interest legislation (even across party lines).
If descriptive representation leads to substantive representation, the current political outlook in American legislative politics looks historically hopeful for women. The 2018 election cycle brought the proportions of women serving in state legislatures to new heights. Women now make up 28.7 percent of all state legislators nationwide, marking a sizable increase from 25.1 percent in 2017. 1 In states like Nevada (50% women) and Colorado (47% women), women’s numerical representation in state legislatures mirrors that in the general population, achieving a representational ideal of proportionality optimal for aggregative and deliberative democracy (Mansbridge 1998).
Nevertheless, the impressive electoral gains for female descriptive representation in 2018 did not extend consistently across the United States. In seventeen states, women make up less than 25 percent of legislators; this statistic fails to reach even 20 percent in six of these states. Women living in states like West Virginia (14.2% women), South Carolina (15.9% women), and Mississippi (13.8% women) face a dramatically less positive outlook if descriptive representation constitutes the only clear pathway toward substantive representation. For the numerous states in which the number of women elected into office fails to keep up with national gains, the exploration of alternative means toward the representation of women’s political voice becomes vital.
Women’s organized advocacy might be one way in which women’s representation can be achieved in the face of low levels of descriptive representation. On the national level, organized interests influence politicians and policy through a number of strategies with varying levels of success (Baumgartner et al. 2009; Grasse and Heidbreder 2011; Kollman 1998, e.g.). 2 Strategic lobbying activities available to state-level women’s advocacy organizations to gain attention from state legislators can provide an important link to women’s substantive representation. However, if women’s organized advocacy can be instrumental in women’s representation for states with low levels of descriptive representatives, a prerequisite is that men must be responsive to women’s issue groups. The field experiment presented in this paper explores this prerequisite, focusing on a group’s ability to secure legislative access.
Lobbying groups can use different strategies for outreach depending on whom within the legislature they target for access; with less certain allies, a group can use stronger lobbying tactics to underline their potential value to a legislator. Based on the links between descriptive and substantive representation, I argue that female legislators will be more likely than their male counterparts to provide a women’s issue group access at face value. However, I also argue that this baseline gender gap should render the effectiveness of different lobbying tactics heterogeneous across gender. If women in office are likely to be responsive to a women’s group regardless, heightened lobbying tactics signaling electoral or informational benefits are unlikely to affect observable change. In contrast, I argue that male legislators should be more sensitive to the same heightened lobbying tactics, shaping an environment wherein the effectiveness of a women’s issue lobbying strategy in securing legislative access depends on a legislator’s gender identity.
Little is currently understood about how descriptive representation is linked to the inclusion of marginalized groups’ advocacy organizations in policymaking. In fact, lobbying is often studied with an eye toward its relationship with representational inequity. Thus counterintuitively, this paper identifies how lobbying in state legislatures can help facilitate, rather than inhibit, increased democratic inclusion. Women’s organizational lobbying emerges as having strong potential power to effectively close observed gender gaps in legislative responsiveness, increasing opportunities for women’s substantive representation overall.
Organizational Access and Influence
To have influence on the legislative process, an advocacy organization must have access to legislators. A lobbyist or organization’s access to policymakers is considered essential by political analysts and advocates alike (Austen-Smith 1993; Herndon 1982; Langbein 1986; Ornstein and Elder 1978; Sabato 1985; Snyder 1990). Sabato (1985, 127) notes, “political analysts have long agreed that access is the principal goal of most interest groups, and lobbyists have always recognized access is the key to persuasion.” Organizational access thus marks a crucial gateway to further policy influence; without it, influence in policymaking is improbable. Therefore, a study of the conditions under which a legislator grants access to an advocacy organization is effectively a study of the conditions under which influence becomes possible.
While access to political decision-makers, the ability to get a “foot in the door,” is fundamental to political influence, it is also unequally distributed (Dahl 1957, 1961; Garson 1974, 1978; Lowi 1979). Inequalities in access are paramount to the study of marginalized group influence, as “old boys” networks shape uneven opportunities for insider and outsider groups. That said, even among mainstream and insider organizations, access to elected officials is not guaranteed to all groups who ask. A legislator’s time is limited, fixed, and valuable. As such, legislators cannot provide access to all groups hoping to collaborate on policy or share policy preferences (Hall 1996).
It is important to note that access does not necessarily lead to significant influence on legislative decisions on policy formation. While access provides the critical opportunity to have group concerns heard, it does not guarantee that a legislator will take action on a group’s insight, requests, or advice. But group access reflects inclusion in the policymaking process, which is of paramount importance to this paper. Historically, marginalized groups have been ineffective at influencing policy that might otherwise challenge social and institutional inequality in large part because of barriers to access. Therefore, given that access is important, privileged, and reflective of a legislator’s priorities, it is highly relevant for observation and analysis in and of itself.
Benefits from Providing Access
What influences an individual legislator to set aside time to hear the group’s concerns? My expectations in answering this question rely on a theoretic framework similar to Hall and Deardorff’s (2006) model of lobbying as a legislative subsidy. To assess the value of providing a group access, legislators estimate the benefits of time spent with that group given the costliness of their time. I characterize the potential benefits to legislators in providing an advocacy organization legislative access under three categories: electoral, informational, and intrinsic. Electoral benefits relate to how much electoral payoff a legislator can expect by working with a particular issue group. For instance, a legislator is likely to estimate attractive benefits from providing access to or partnering with a group with strong capacities to mobilize his or her constituents or to contribute to his or her campaign. Informational benefits are more broadly defined, as valuable information can take many forms. 3 Generally speaking, however, informational benefits emerge, in Hall and Deardorff’s (2006, 69) terms, as a “legislative subsidy,” or an offer of any valuable information to relieve the legislator of some of their work burden in policymaking.
Finally, benefits can also arise intrinsically, that is, from the positive or negative feelings a legislator gets from working with or on behalf of a particular group (Broockman 2013). Intrinsic benefits relate to a legislator’s personal preferences and life experiences. For a marginalized group facing structural barriers to access, intrinsic benefits become a critical factor to consider, as a legislator’s own experience of marginalization as a member of an identity group can shape their political priorities. As Burden (2007) argues, legislators use their personal preferences in making decisions on how to vote, which bills to sponsor, and how to allocate their time. While an elected official’s personal preferences, ideologies, and life experiences are difficult to reliably measure, theories of descriptive representation suggest that legislators emerging from distinctive identity groups (especially those facing historic marginalization) are expected to prioritize the interests of their identity group in solidarity (Dawson 1995; Gay 2004). In fact, intrinsic motivations to represent one’s marginalized identity in part form the basis for arguments underlining the need for descriptive representation in democracy. Phillips (1998) argues that descriptive representation allows women to give voice to preferences, issues, and interests previously overlooked in political deliberation. Importantly, great tension in the literature exists regarding any individual descriptive representative’s ability to represent the diverse interests of an entire marginalized subpopulation (e.g., Dovi 2002). The link between female legislators and the prioritization of “women’s issues” broadly defined must be probabilistic, as the connections between descriptive identity and legislative behavior is far from absolute. However, evidence does suggest that when the links between descriptive and substantive representation do emerge, they are motivated intrinsically (e.g., Broockman 2013). It is therefore reasonable to consider the intrinsic, more psychological benefits an elected official might derive from providing access to a group with whom they share common life experiences or identities.
Benefits and Allies
Whether the benefits of working with an advocacy organization derive electorally, informationally, or intrinsically, the greater the benefits a legislator attributes to time spent with an advocacy organization, the more likely the legislator will be to accept a meeting request from that organization. For a legislator, perceived benefits must arise out of an overlap between his or her policy priorities and the group’s. The resulting implication is that lobbying groups will have the most success accessing legislators with whom this overlap is clear—in other words, with legislators they understand to be allies. How a legislator perceives the benefits of working with an organization shapes the strength of potential alliances between legislator and group. While some alliances between legislator and group might be relatively strong, alliances with other legislators might be less certain. Ultimately, an advocacy organization can use different strategic lobbying tactics to alter a legislator’s perceptions of the benefits they offer, bolstering ties to less certain legislative allies and increasing their likelihood for access. 4
For an advocacy organization representing a historically marginalized group, alliances based on intrinsic benefits are once again of marked interest. Links between descriptive and substantive representation would suggest that men and women in office are motivated differently to respond to a women’s advocacy organization based on a gender group affinity (intrinsic benefits). This paper thus argues that female legislators will be more likely than their male counterparts to provide access to a women’s group, all else being equal. 5 I therefore predict that at a baseline level of responsiveness, with no additional lobbying tactics to increase legislative responsiveness used, there will be a gender gap in the likelihood a legislator will provide access to a women’s organization.
Strengthening Alliances with Strategic Lobbying
Legislator perceptions of intrinsic benefits, especially as they arise from elements of descriptive representation, cannot be easily manipulated by an advocacy organization: a lobbyist cannot change a legislator’s experience of in-group or out-group gender identity. 6 In contrast, a legislator’s perception of a group’s electoral or informational benefits are changeable. As such, where intrinsic benefits are unavailable or where alliances are relatively weaker, an advocacy organization can use additional strategic lobbying tactics to strengthen potential alliances by emphasizing the electoral or informational benefits they can offer to a legislator. My field experiment thus manipulates an organization’s use of lobbying strategies that signal electoral or informational benefits to evaluate their impact on the responsiveness of legislators—both male and female—to a women’s group.
Past studies of lobbying activity have identified two strategic methods to successfully achieve legislative inclusion, each of which provide either informational or electoral benefits (e.g., Burstein and Linton 2002; Grossman and Helpman 2001; Kollman 1998). First, advocacy organizations can use what are commonly known as “inside strategies” for lobbying (Gais and Walker 1991; Grossman and Helpman 2001; Kollman 1998). Inside strategies refer to the provision of specialized information, or “legislative subsidies.” In offering legislative subsidies, an advocacy organization can make the provision of access to their organization more likely. Empirically, the value of lobbying in terms of information and expertise provision is widely explored (Austen-Smith 1993; Austen-Smith and Wright 1996; Hansen 1991; Potters and Winden 1992; Rasmusen 1993; Wright 1996). In addition, organizational surveys (Berry 1977; Gais and Walker 1991; Heinz et al. 1993; Nownes and Freeman 1998; Schlozman and Tierney 1986) confirm the validity of models asserting the informational benefits of lobbying, describing the activities undertaken by most strategic interest groups as aimed at gathering and providing expertise to lawmakers. I thus expect that inside lobbying tactics, or meeting requests specifically signaling that an organization can offer valuable expertise, will increase legislative responsiveness to a women’s advocacy organization.
Advocacy organizations can also use “outside strategies” in lobbying. “Outside” lobbying strategies denote attempts to mobilize constituent or public support to affect greater legislative attention and political action (Goldstein 1999; Kollman 1998). Organizations wield valuable power to shape constituent opinion, which can in turn shape legislator behavior (Grossman and Helpman 2001). Scholars observe “outside lobbying” to have noteworthy influence on the choices of elected officials (Kollman 1998), and advocacy organizations large and small are often credited with pivotal roles in mobilizing public opinion campaigns to measurably shape policy outcomes across diverse issues (Skocpol et al. 1993; Soule and Olzak 2004; Weldon 2002, 2012). An advocacy organization can use outside lobbying tactics to make engaging with their organization appear electorally beneficial. I thus expect that if a women’s organization uses outside lobbying tactics on a legislator when requesting access, their likelihood of seeing a response will increase. In turn, I also predict that the greatest increase in responsiveness will emerge when a women’s organization uses both inside and outside lobbying tactics simultaneously.
Finally, given my expectations that a gender gap will emerge in the baseline responsiveness to a women’s organization, I argue that both inside and outside lobbying tactics are likely to have a greater impact on men than on women. The logic here is quite straightforward: when an organization seeks access to a legislator to discuss an issue that legislator cares little about, the provision of extra incentives to pay attention to their group is worthwhile. But if the organization is seeking access to a strong ally, the investment in extra lobbying should have little effect on the already high likelihood of response. I therefore test the prediction that inside and outside lobbying tactics will have heterogeneous effects on male and female legislators, showing a relatively strong impact on men but a limited effect on women.
My experiment focuses on gender gaps as they relate to legislative responsiveness to a women’s issue group specifically. While I argue that gender gaps in responsiveness to a women’s issue group are a function of intrinsic benefits, my study cannot confidently determine if instead female legislators are more likely to respond to all issue groups. There are theoretic grounds for the prediction that based on their female gender identity, women in office might be more responsive to all organizations seeking access. For instance, Lazarus and Steigerwalt’s (2018) theory of gendered vulnerability would suggest women are more likely than men to prioritize activities directly related to reelection and should therefore be more responsive to constituent organizations. Ultimately, the decision not to test for this alternative mechanism was carefully made with a mind toward ethics, as pursuing this line of inquiry would have demanded a significantly larger experimental sample. While I am clearly interested in gendered differences in legislator behavior, my motivating interest in conducting this experiment related to a gendered issue group’s ability to gain legislative access. I speak to this point again in my concluding discussion, and at greater length in the online appendix.
The study of organizational access, especially for organizations seeking increased representation for marginalized groups, has much to gain from assessing how descriptive and nondescriptive representatives differ in their willingness to include such groups in the policymaking process. While women’s organizations might face gender gaps in seeking legislative attention, they might also possess the power through strategic lobbying to bridge gender gaps. Such power could be pivotal in state-level politics, where gains for women and women candidates often lag behind those in the national political arena. The remainder of this paper applies these motivations to a preregistered experimental design. 7
Experimental Design
I designed and implemented an auditing experiment that sends a sample of 600 state legislators across eight states an email containing a meeting request from a fictitious women’s organization dedicated to advocating on behalf of female victims of sex trafficking. For my experimental treatments, I manipulate the text of the emails, randomly assigning legislators into four treatment groups receiving varied requests for a meeting: one group signaling constituent mobilization, one group signaling expertise collection, one group containing both signals, and a control group receiving a simple meeting request signaling neither inside nor outside lobbying activity. Legislative access was then operationalized by measuring the response rates to the meeting request, allowing me to make comparisons across treatment groups as well as across gender.
A research design utilizing randomized field experimentation is optimal to my investigation in two critical ways. First, the use of randomization in the experiment allows me to account for confounding variables such as partisanship and ideology. Second, observational inquiry into the inclusion of special interest groups in policymaking faces significant endogeneity problems. Looking retrospectively at lobbying partnerships between women’s issue groups and individual legislators, it is impossible to determine if the group’s lobbying tactics triggered legislator responsiveness or if the group targeted legislators whom they knew would respond. This issue of reverse causality is only heightened when considering my interest in isolating gender, a strong potential confounding variable, from the influence of lobbying tactics on access provision. Experimental analysis allows me to disentangle lobbying tactics from shared gender identity as they influence legislator responsiveness independently.
The Sample
My population of interest is the universe of state legislators in the United States. For ethical reasons, I select a sample of U.S. states from which to draw state legislators that is as limited in number as possible to maintain statistical power. I include elected officials from Michigan, Pennsylvania, South Carolina, Wisconsin, Missouri, Massachusetts, Rhode Island, and California. 8 In determining which states to include in the study, I faced certain practical obstacles in terms of timing and sample size. Numerically, I needed to strike a delicate balance in reaching the threshold of individual male and female legislators to provide sufficient statistical power while simultaneously ensuring that the experiment extend into as few states as possible. At the same time, the experimental conditions required that each legislature be in session through the period in which the experiment would be fielded. Within these limitations, however, I carefully select states to maintain balance within the sample in terms of professionalism in the legislature, the proportion of women’s serving in the legislative body, and overall partisan composition. A full discussion of state-level descriptive statistics for these three variables is presented in the online appendix.
Table 1 presents state-level descriptive statistics for these three variables. For legislative professionalism, I refer to the second-dimension measurements collected by Bowen and Greene (2014) with Rosenthal’s (2000) conceptualization of legislative professionalism in mind. I use this measurement strategy in light of the potential relationships between legislative staff support and benefits to policymakers in providing access to interest groups. Differences across states in the degree to which legislatures are support-intensive and work-intensive (Bowen and Greene 2014) might shape differences in the way individual legislators value the expertise of interest groups. I include states with high levels of professionalism expecting that such states should provide for the most conservative estimates of positive effects. That said, maintaining balance within the sample is also critical toward my objective of evaluating differences in gender across various types of legislatures, including those that are more likely to prioritize interest group incorporation.
State-Level Descriptive Statistics.
I thus also pay close attention to variation in gender compositions in my process for state selection. Scholars of the links between women’s descriptive representation and substantive representation debate the importance of “critical mass” in shaping whether female representatives will emerge as activists for women’s issues rather than “token” women within an institution (Beckwith and Cowell-Meyers 2007; Dahlerup 1988; Kanter 1977). I therefore take care to include states with both high and low proportions of female legislators to account for potential institutional differences in how women officeholders prioritize partnerships with women’s issue groups. Because this paper is motivated by an interest in states with particularly low levels of women’s descriptive representation, I include states like South Carolina and Pennsylvania with especially few women in office, while also including Rhode Island as a state with a relatively high proportion of female legislators. The remaining states fall around, if not slightly below, the 2018 nationwide average (25.3%). All data on state legislature gender composition were gathered using reports published by the Center for American Women and Politics (CAWP 2017).
Finally, states sampled for experimental analysis reflect the best balance in overall partisan composition possible. As shown in Table 1, the states included in the experiment represent both Republican and Democratic controlled legislatures, varying in terms of strength of majority party. In the aggregate sample of legislators across all eight states, 64.1 percent are Democrats. Past research demonstrates nuanced relationships between gender and political party. Partisan gaps among female congressional officeholders, for instance, have grown over the past 30 years, shaped by gendered differences across party in recruitment practices and campaign contributions (Crowder-Meyer and Cooperman 2018; Thomsen and Swers 2017). Among elected female representatives, Swers (2002) finds that Democratic and Republican women diverge in their tendencies toward women’s issue activism at the federal level, especially in varying contexts of partisan control. Ideologically, Osborn et al. (2019) show that female state legislators are increasingly more polarized than their male colleagues, with important implications for women’s representational policymaking. If Democratic and Republican women legislate differently—not just from men, but from each other—it is important to attend to potential partisan-gender differences in responses to my experimental stimuli.
However, given the partisan gaps across women officeholders overall, the number of Republican women in my sample is limited, rendering a confident examination of these partisan dynamics largely out of reach. I account for partisanship in my experimental design with my selection of states, my choice of the relatively nonpartisan issue of sex trafficking, and finally with my use of block randomization by party and state in selecting the sample of male participants to mirror the sample of female legislators. Nonetheless, I do estimate triple-interaction analyses between gender, party, and treatment group to provide some exploratory insight. The results of these regressions are presented in the online appendix along with a more in-depth discussion. 9
For each of the states in the study, I include all female representatives in my sample. An equal number of male legislators are randomly selected to match the female legislators in number, state, and party. 10 I then randomly assign legislators to treatment groups across the entire sample, subsequently testing my randomization scheme for any differences across gender or party. The results of this χ2 test were successful, producing a p value of .8934. Importantly, this experiment was approved in expedited review by my university’s Institutional Review Board (IRB). Nonetheless, auditing experiments on public officials by academics must be carefully designed with attention to ethics. A full discussion of my ethical considerations and use of deception can be seen in the online appendix.
Experimental Stimuli and Treatment Conditions
The experimental stimulus consists of a meeting request to discuss issues of sex trafficking, with specific emphasis placed on assistance to victims of sex trafficking rather than to the policing of sex traffickers. Support for victims of sex trafficking represents an optimal choice for a “women’s issue” in that it is both nonpartisan and relatively nonsalient. 11 If the issue had been drawing heavy attention in the media or considered “owned” by a particular party (Petrocik 1996), discerning whether differences in responsiveness stemmed from experimental stimuli or from concerns of partisanship or salience would be impossible. 12
The meeting requests appeared to come from a newly formed organization of local women dedicated to assisting fellow female victims of sex trafficking. 13 Each emailed meeting request contained the subject “Women Against Violence and Exploitation,” to send a strong signal of the group’s collective identity. Women Against Violence and Exploitation (WAVE) was also the name of the organization engaging in experimental outreach and was a fictitious organization I created for the purposes of this study.14,15 Meeting requests were delivered to each legislator on their direct email address, a method for outreach frequently used in each of the states included in the study. 16
Within the text of each emailed meeting request, I independently manipulated a mobilization treatment factor and an expertise treatment factor. The randomly assigned treatments consisted of two levels, yielding four total treatment arms. The control condition for mobilization factor sent a simple meeting request to introduce the group and speak about the importance of policy supporting the victims of sex trafficking. In contrast, the mobilization condition for this factor included a reference to a working petition on the importance of sex trafficking policy. The email’s language emphasized the group’s success in mobilizing the legislator’s constituents specifically, and stated that the petition had already been signed by a specific number of constituents residing in the most populous counties and/or cities within the legislator’s district. 17 The control condition for expertise factor once again contained a simple request to meet, while the treatment condition underlined the organization’s collection of expertise specific to the most populous counties legislator’s district. The email text for all four treatment arms can be viewed in the online appendix.
The experimental output then observes if and how the legislative office responded to the meeting request. 18 I fielded the experiment from April 10 to 12, 2018. Each legislator was allotted 10 business days to respond to the meeting request. For legislators responding to the meeting request, I subsequently sent an email informing them that the organization was suspending action on the project indefinitely. While these emails were sent from the fictitious organizer and only to those legislative offices that responded to the initial request, a final debriefing email was then delivered on June 22, 2018, from my personal email account. 19
Experimental Results and Analysis
Across the eight state legislatures included in the study, 254 out of the 600 legislators emailed responded to the women’s organization’s meeting request, yielding an overall response rate across all legislators of 42.3 percent. Each response was carefully coded to produce different measures of legislative responsiveness. A breakdown of response rates by states as well as a description of my coding protocol and various responsiveness measures can be seen in the online appendix. Following Butler and Broockman (2011) and Broockman (2013), I analyze an objective binary dependent variable for if the organization received a response from the legislative office contacted. 20
Table 2 shows the rates of response broken down by treatment groups as well as by legislator gender, evaluating gender gaps across the treatment groups, the control groups, and the overall sample. In addition, the final row in this table shows the differences across gender for each treatment group and for the total sample. Among the emails that did not signal any specific lobbying tactics, 52.3 percent of female legislators responded whereas only 23.5 percent of male legislators responded, a large and statistically significant difference of 28.7 percent (p = .0005). Across the entire sample, the 12 percent difference in response rates across women (48.3%) and men (36.3%) in office was also statistically significant, demonstrating a clear gender gap. On its face, the results support my expectation that men and women in office respond at different rates to a women’s organizational meeting request assuming no additional lobbying tactics. Nonetheless, while differences are strong across the control groups and the sample at large, gender gaps are not consistent across treatment groups. Interestingly, there appears to be almost no difference in the response rates between female legislators and male legislators in the mobilization treatment group. The difference across gender within the combined treatment group is also relatively small; none of the treatment groups saw significant gender gaps in response rates.
Response Rates by Gender and Treatment Group.
p < .1. **p < .05. ***p < .01.
Taking a closer look at the differences in response rates across treatment and control groups, Table 3 shows treatment effects overall, evaluating gender gaps in treatment effects specifically. Overall, the mobilization treatment effect had a positive and significant effect on legislator responsiveness to a women’s group meeting request, increasing the likelihood a legislator would respond by 17.3 percent. But this overall effect might derive from the treatment’s effect on male legislators specifically. Significant differences emerge across gender in observed treatment effects, providing support for my expectations that men would see greater change in responsiveness from additional lobbying tactics. While the mobilization treatment had an impressive impact on male legislators, increasing their response rate by a substantial and significant 29.5 percent, female legislators saw only a slight increase in the likelihood of response. Similarly, the combined treatment effects also show an overall positive effect on legislator response rate. However, whereas men saw a strong and statistically significant increase of 19.1 percent (p = .02) in response to the combined treatment, women in office were actually less likely to respond (though this difference also fails to reach statistical significance).
Differences in Response Rates.
For a more thorough examination of the findings presented in Tables 2 and 3, I conduct an ordinary least squares regression analysis predicting whether the legislative office responded to the emailed meeting request sent by the fictitious women’s advocacy organization. Importantly, interpretations of these analyses and those regression analyses that follow must account for the fact that I could not randomize legislator characteristics, and some confounding variables might therefore be at play. With this in mind, additional models evaluating state fixed effects and controlling for specific legislator characteristics are included in the online appendix, the results of which are all consistent with those presented below.
I estimate a baseline model (1) first which predicts effects of the experimental treatments without the inclusion of (female) gender interactions by treatment group, and then estimate a full interaction model (2), adding the interaction terms to test the hypotheses that the relationship between the experimental groups and response rates differ based on a legislator’s gender.
The regression results presented in the coefficient plot in Figure 1 21 strengthen the conclusions drawn from Tables 2 and 3. First, interacting the treatment groups by gender adds important nuance. The baseline model appears to underestimate the influence of gender substantially. Whereas the baseline model predicts that female legislators are 11.3 percent more likely to response to the women’s meeting request, the interaction model predicts this positive increase at 26.5 percent. Employing an F-test to compare the nested models in the two regressions produces a statistically significant χ2 statistic of 6.7891 (p = . 07). I can therefore confirm that female legislators not only respond differently to the women’s advocacy group’s request for access at the baseline, but to the organization’s lobbying tactics as well.
Coefficient plot comparing baseline and interaction models for predicting legislator response rates to a meeting request from a women’s issue group.
The heterogeneous effects show particularly interesting results for male legislators. In the interaction model, the coefficient for male legislators receiving the meeting request signaling constituent mobilization estimated an increase in response rate by 27.1 percent. Looking at the negative and statistically significant coefficient for female legislators, the net effect of signaling constituent mobilization on women in office is estimated at virtually zero. Furthermore, the coefficient estimated for male recipients of the mobilization treatment mirrors the estimated difference between female and male legislators; lobbying tactics signaling constituent mobilization appears to close the gender gap almost entirely. Figure 2 underlines this point well by comparing the predicted rates of response for men and women in office in the control groups and the mobilization treatment groups. While 50.1 percent of women in the control group are predicted to respond to the women’s meeting request, only 23.6 percent of men in the control group are predicted to respond. This marks a statistically significant difference of 26.5 percent. In contrast, men in the mobilization treatment group are predicted to respond at a rate of 50.7 percent, which exceeds the predicted rate of women in the mobilization treatment group by a statistically insignificant 0.9 percent.
Comparing mobilization treatment effects across gender.
The regression results also show that the effect of the combined treatment on men is positive and statistically significant while being entirely negated by the statistically significant negative coefficient estimated for women. In fact, the net effect of the combined treatment on female legislators is unexpectedly negative albeit small (−.037). Figure 3 depicts these results graphically as well. It should be noted, however, that given the noneffects of the expertise treatment, it is likely that the constituent mobilization piece of the combined treatment drives these results.
Comparing combined treatment effects across gender.
Ultimately, the experimental results produce strong findings. Lobbying tactics that signal an organization’s ability to provide electoral benefits to a legislator can act as additional persuasion for a legislator to provide access. Informational benefits seem to have less of an influence on access provision, at least in this context, as emphasis on expertise provision failed to trigger any strong increases in the rates of response. But significant gender gaps emerge in observed responsiveness to a women’s group: when additional lobbying tactics that emphasize electoral or informational benefits are not employed, women’s groups are likely to see more success targeting female legislators for access than male legislators. The results of this experiment suggest that by signaling electoral benefits, a women’s advocacy group can effectively tighten if not close gender gaps in responsiveness. Whereas women in office might be more likely allies to women’s organizations than men, women’s organizations can use strategic lobbying tactics to shift the balance in their favor. On the other hand, my findings also suggest a ceiling effect in efforts to lobby female legislators, wherein additional lobbying tactics fail to provoke any real increase in the rate of response.
Discussion
I conducted a field experiment to explore the opportunities offered by women’s advocacy for increased women’s organizational inclusion in the legislative process. I argued that women in office would be more likely allies to a women’s issue group than men in office, and as such more likely to respond to a women’s issue group’s meeting request. I also predicted that while additional lobbying tactics signaling electoral and informational benefits to a legislator would increase responsiveness, such lobbying tactics would face ceiling effects when targeting female legislators. Men, in contrast, would show substantial room for improvement in their base-level responsiveness; they would thus be influenced strongly and positively to the same organizational lobbying tactics that proved effectively inconsequential for women.
My analysis showed substantial support for these predictions, revealing interesting and conditional gender gaps in the ways men and women in office respond to women’s advocacy and varying lobbying tactics by a women’s group in their willingness to provide access. Women in office appear to matter for organizations advocating on a women’s issue, increasing a group’s potential for access and inclusion. My research lends strong support for links between descriptive and substantive representation while at the same time presenting this linkage through the previously understudied lens of lobbying and organized advocacy. In turn, I also show that by leveraging a legislator’s electoral self-interest, a women’s group can indeed garner attention from those in office who at first glance might be overlooked as unexpected allies. More generally, it appears from this experiment that a group’s intrinsic value and electoral value can matter a great deal for lobbying efforts seeking access to legislators.
In contrast, a group’s informational value did not appear to matter at all. The lack of evidence supporting my prediction that expertise provision should stimulate increased access provision is puzzling, and merits further investigation. Given that extant literature characterizes the inclusion of lobby groups in policymaking in terms of “legislative subsidies” (Hall and Deardorff 2006), why did the offer of specialized expertise not increase legislative interest? The answer might relate back to Mayhew’s (1974) seminal work, such that electoral incentives are the principal driver of a legislator’s behavior in this scenario as well. My results definitely indicate that for male state legislators, the “electoral connection” (Mayhew 1974) is strong; a petition with only .5 percent of support by registered voters more than doubled their likelihood to provide access. However, the null results might also be reflective of experimental design. The value of expertise might relate to its source, and it is possible that expertise provision from an unknown organization without more informative references was insufficient to provoke increased attention. Therefore, the weakness of the expertise treatment relative to the more explicit mobilization treatment is also worth consideration in these unexpected findings. A few additional questions remain unanswered. First, these findings arise out of an examination of responsiveness to a women’s group advocating on the needs of victims of sex trafficking. Sex trafficking can be defined as a women’s issue in that it predominantly effects women specifically, but it is unique in comparison to other issues traditionally defined as “women’s issues” such as abortion or reproduction issues (e.g., Reingold 2000; Swers 2002) or anti-discrimination issues (e.g., Wolbrecht 2000) in that it cannot be easily tied to a particular ideology or political party. It remains unclear if the trends observed above with regard to gender gaps and constituent mobilization lobbying hold if the women’s organization of interest is advocating on an issue more partisan, or more salient. Additional research is thus needed before these results can be broadly generalizable to lobbying efforts of all women’s issue groups in U.S. states.
Second, my theory rests on the assumption that the mechanisms linking descriptive and substantive representation for women derive from intrinsic benefits. But this represents only one side of the story. In fact, scholars propose two central theories to explain the linkages between descriptive and substantive representation: the first explanation is intrinsically or personally motivated, but scholars have also often suggested that linkages between descriptive and substantive representation arise from how legislators perceive the strength of electoral support offered by in-group constituencies. If a legislator perceives a great deal of electoral potential for a given constituency group, they are likely to allocate more time to address policy issues specific to that particular constituency, and vice versa; they are likely allies to that constituency group and their policy preferences. There thus remains an underlying question as to what drives the gendered differences I observe in this experiment. Broockman’s (2013) field experiment speaks elegantly to this debate in the literature, examining intrinsic verses extrinsic (i.e., political or electoral) motivations behind the links between descriptive and substantive representation among black politicians. His findings do support the conclusion that shared group identity shape the links between descriptive and substantive representation with regard to race. In terms of gender scholarship, compelling research also suggests that women in office often feel compelled to, in Pitkin’s (1967), “stand in” and “act for” women as a group (Carroll 2002; Reingold 2000; Swers 2002). As such, I believe my assertions that the gender gaps observed here reflect intrinsic benefits of descriptive representation are well-founded. But future examination of these trends should pay more comprehensive attention to disentangling these two driving potential mechanisms. 22
Finally, in limiting my field experiment to the examination of lobbying tactics of a women’s issue group only, I leave another question unanswered regarding mechanisms and gender gaps in legislative responsiveness. That is, are women in office more responsive than their male counterparts to basic requests for access by interest groups across all issue areas? Given female legislators’ greater disposition toward the prioritization of constituency services (Lazarus and Steigerwalt 2018; Richardson and Freeman 1995), there is reason to expect women in office to more readily respond to group requests for meetings (or to simply pay closer attention to their legislative emails). The question as to whether the gender gaps emergent in this study relate to differences in responsiveness to women’s advocacy groups or advocacy groups at large remains unclear, and merits further investigation. But importantly, this experiment was designed to speak to advocacy organizations seeking progress on women’s representation specifically, especially in states where progress in levels of descriptive representation continues to lag. For such organizations, the source of gender gaps in legislative responsiveness to their efforts for inclusion is largely irrelevant. What matters is that these gaps have been observed to exist, and that they have also been shown to be bridgeable. Furthermore, within the study of interest groups and lobbying, very little is currently understood about how legislators respond to and engage with groups that advocate on issues that disproportionately affect women. This paper represents an effort to fill that gap in the literature.
This study and its findings have substantial implications for women’s representation in the U.S. state legislatures. A great deal of important policymaking impacting the everyday lives of women in the United States happens in state legislatures. Women in office matter for women, but in states like Tennessee, Alabama, and Mississippi where (as of 2020) women make up 15.2, 15.7, and 16.1 percent of the state legislature respectively (CAWP 2020), alternative mechanisms for increased representation for women are crucial. By shedding light not only on gender gaps in representation, but also on how gender gaps can be closed for overall increases substantive representation, this research can act as a beacon for women’s organizations who might otherwise overlook underexplored pathways toward progress.
Supplemental Material
Online_Appendix – Supplemental material for Getting a High Heel in the Door: An Experiment on State Legislator Responsiveness to Women’s Issue Lobbying
Supplemental material, Online_Appendix for Getting a High Heel in the Door: An Experiment on State Legislator Responsiveness to Women’s Issue Lobbying by Elizabeth Wiener in Political Research Quarterly
Footnotes
Acknowledgements
I thank Beth Reingold, Zachary Peskowitz, Alexander Bolton, Amy McKay, David Miller, Nichole Bauer, Jon Rogowski, Alice Kang, and the anonymous reviewers for their generous, thoughtful, and insightful feedback on previous drafts. Many of my colleagues at Emory University provided meaningful insight and guidance in the research design and execution of this paper, including Kirsten Widener, Danielle Villa, Nancy Arrington, Maggie Macdonald, Devon Thurman, Anna Gunderson, Laura Huber, and the broader FCJ community.
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) received no financial support for the research, authorship, and/or publication of this article.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
