In the summer of 2021, the city of Carmel, Indiana issued its first proclamation recognizing June as LGBTQ Pride Month. The document—signed by Mayor Jim Brainard—articulated the city’s commitment to equality by acknowledging the struggles of LGBTQ Americans and declaring itself a place of tolerance and mutual respect. Despite Carmel’s Republican leanings, city residents largely expressed support for the measure, saying it was overdue and provided LGBTQ people renewed hope. In June 2022, the Horry County Council in South Carolina rescinded an LGBTQ pride proclamation, just one month after it passed. They did so in response to mobilization by Christian conservatives. The diverging trajectories of LGBTQ pride proclamations in Carmel and Horry County reveal how local elected officials have influence over the recognition of LGBTQ people through nonpolicy mechanisms and that recognition is a source of conflict. As a result, we focus on the behavior of mayors and their responsiveness to public-facing constituency requests about issuing an LGBTQ pride proclamation on behalf of lesbian and gay Latine constituents.
Local politics have been a critical venue for the LGBTQ movement to advance its agenda (Haider-Markel Reference Haider-Markel2010), which includes winning policy but also gaining legitimacy and recognition (Proctor Reference Proctor2022a). Since the Stonewall rebellion in June 1969, LGBTQ people have held annual demonstrations and marches in June in cities around the United States, building their movement’s visibility (Armstrong and Crage Reference Armstrong and Crage2006). City governments were also the first to pass nondiscrimination ordinances in the early 1970s (Smith and Haider-Markel Reference Smith and Haider-Markel2002). Alongside policy developments, mayors began to issue proclamations to recognize LGBTQ pride, which have been met with resistance from the public (Faderman and Timmons Reference Faderman and Timmons2009; Levy Reference Levy2015; Simmons Reference Simmons2012). These conflicts in local politics continue to today. Many cities still oppose recognizing LGBTQ people, pride events still generate counterprotests, and conservatives are currently using local-level politics to censor LGBTQ visibility. Thus, how mayors represent LGBTQ people is as salient today as it was 50 years ago.Footnote 1
Studies of government officials’ responsiveness to marginalized constituencies have found mixed evidence of discrimination (Broockman Reference Broockman2013; Einstein and Glick Reference Einstein and Glick2017; Lowande and Proctor Reference Lowande and Proctor2020; White, Nathan, and Faller Reference White, Nathan and Faller2015). This literature typically examines whether shared characteristics between a constituent and representative increase responsiveness (Broockman Reference Broockman2013; Kalla, Rosenbluth, and Teele Reference Kalla, Rosenbluth and Teele2018) or whether officials are more responsive to constituents from advantaged groups, regardless of the officials’ identity characteristics (Lowande and Proctor Reference Lowande and Proctor2020; White, Nathan, and Faller Reference White, Nathan and Faller2015). A weakness in these studies is that they struggle to explain findings such as greater responsiveness to lesbians than gay men in the absence of greater responsiveness to straight couples compared to LGB couples (Lowande and Proctor Reference Lowande and Proctor2020) and men’s greater responsiveness to women than to other men (Kalla, Rosenbluth, and Teele Reference Kalla, Rosenbluth and Teele2018). Relatedly, studies about local Latine representation are inconclusive about the relationship between representatives and constituents. Some studies find that Latines are more likely to be elected in majority of Latine cities (Cuéllar Reference Cúellar, Orr and Morel2018), while other work finds that Latine mayors de-emphasize their identity in campaigns (Orr, Morel, and Fraga Reference Orr, Morel and Fraga2018). Thus, it is unclear how race and ethnicity shape the constituent-representative relationship among Latines. Similarly, research finds that LGB representatives de-emphasize their identities in campaigns (Haider-Markel Reference Haider-Markel2010), raising further questions about how sexual identity shapes the constituent-representative relationship. Therefore, we ask: are local elected officials responsive to constituency requests from Latine gay men and lesbians and does responsiveness vary across these subgroups?
This paper answers this question by drawing on theories of intersectionality to examine mayoral responsiveness to Latine gay men and lesbians. By theorizing about representation at the intersection of multiple marginalizations (Bauer and Cargile Reference Bauer and Cargile2023; Cohen Reference Cohen1999; Montoya et al. Reference Montoya, Bejarano, Brown and Gershon2022; Strolovitch Reference Strolovitch2007), our approach can uncover inequalities that are masked in single-axis approaches that dominate studies about responsiveness to marginalized constituencies (Glenn Reference Glenn2002; Masuoka and Junne Reference Masuoka and Junne2013; Medenica and Fowler Reference Medenica and Fowler2020). We draw from the key assumption of intersectional scholarship that intersecting categories of difference produce distinct experiences of marginalization (Hancock Reference Hancock2007). In the context of this study, this assumption means that Latine gay men and lesbians are intersectionally marginalized constituencies whose social positions are a product of sexuality, gender, and race. In addition, we evaluate representation by examining the multiple, intersecting positions of constituents relative to a representative. For example, we examine the responsiveness of Latine representatives to lesbian and gay Latine subgroups who share a racial identity but are marginalized based on gender and sexuality. We make similar comparisons across race, gender, and sexuality to assess variation in responsiveness through an intersectional lens.
Similar to other studies, we use an audit experiment to examine the responsiveness of mayors to LGB Latines. In audit studies, researchers pose as constituents who are contacting public officials to obtain information about government services (Costa Reference Costa2017). An advantage of these studies is that scholars can identify real-world discrimination in responsiveness, which is otherwise challenging to examine through alternative approaches such as surveys or interviews with public officials and constituents (Button, Wald, and Rienzo Reference Button, Wald and Rienzo1997). In addition, audit experiments take place in a natural setting, meaning that the information request is something the official would encounter in their position. Thus, scholars can draw conclusions about the behavior of officials. Finally, we acknowledge that there are ethical considerations when researchers contact public officials, which we discuss in the paper. However, we posit that the social benefits of this research outweigh the potential risks.
Our study makes several contributions to studies of responsiveness, LGBTQ and Latine politics, and local politics. This study contributes to the audit study literature by measuring responsiveness to a public-facing request. Most audit studies are requests for private information, which are low-cost ways for officials to respond to marginalized groups because they can be kept from the public. As a result, existing studies likely understate the prevalence of discrimination. We find that mayors are more responsive to Latine lesbian couples than gay couples. These findings align with other research about sexuality that finds gay men are viewed more negatively than lesbians (Bettinsoli, Suppes, and Napier Reference Bettinsoli, Suppes and Napier2020; Herek and Capitanio Reference Herek and Capitanio1999). We also find that LGBTQ mayors were more likely to respond to our inquiry than straight, cisgender mayors, that Latine mayors were less likely to respond to our inquiry than non-Latine mayors, and that women mayors were more likely to respond to our inquiry than men. These findings complicate single-axis frameworks which predict that shared identity characteristics with a representative should increase responsiveness. In our study, it varies by sexuality, gender, and race. Finally, we find that political contexts shape responsiveness. Mayors from cities where LGBTQ people are protected from discrimination were more likely to respond to our request. Thus, public policies are constraints on the behavior of mayors.
Our findings have several implications. First, our results demonstrate that inequalities are not additive. If they were, we would have found that Latine lesbians were responded to less than gay men. We also did not find evidence that mayors are more responsive to constituents when they share characteristics. Patterns of responsiveness vary depending on which characteristics are shared between constituent and representative. As a result, studies of representation, particularly descriptive representation, should use intersectional approaches to theorize about constituent-representative dynamics. Our results also suggest that political science needs to pay more attention to gender dynamics within LGBTQ scholarship. There is a growing body of evidence that public officials are less responsive to gay men than lesbians, which should be brought into conversation with a growing body of research about discrimination against transgender people. Finally, our results suggest that nondiscrimination policies have “spillover effects” on constituency service, reflecting their importance to improving LGBTQ people’s representation.
Discrimination in Constituent Service
This paper uses an audit experiment to examine mayors’ responsiveness to Latine same-sex couples who request that their city issue a pride proclamation. There is a growing literature in political science that uses audit experiments to study whether public officials discriminate against marginalized groups (Landgrave and Weller Reference Landgrave and Weller2022), including racial and ethnic minorities (Einstein and Glick Reference Einstein and Glick2017), women (Kalla, Rosenbluth, and Teele Reference Kalla, Rosenbluth and Teele2018), and gay men and lesbians (Lowande and Proctor Reference Lowande and Proctor2020). In previous studies, researchers have contacted public officials to obtain private information about government and constituency services (Costa Reference Costa2017; Landgrave and Weller Reference Landgrave and Weller2020), such as voting (White, Nathan, and Faller Reference White, Nathan and Faller2015), access to public housing Einstein and Glick (Reference Einstein and Glick2017), and obtaining a marriage license (Lowande and Proctor Reference Lowande and Proctor2020). These studies reach varying conclusions about the extent to which government actors discriminate.
Worth emphasizing, however, is that these experiments measure discrimination from private information requests. The problem, however, is that discrimination often manifests through public-facing dynamics of politics. As a result, public officials may behave differently when the act is public. The public–private distinction is especially relevant in LGBTQ politics, whereby politicians have been unwilling to publicly support LGBTQ constituents despite privately claiming to support them (Proctor Reference Proctor2022a). Therefore, it is unclear whether the absence of discrimination in previous research is an artifact of the private nature of the request (Lowande and Proctor Reference Lowande and Proctor2020). More generally, responsiveness to private requests is low cost for public officials, who can prevent the public from observing their behavior. As the aforementioned examples demonstrated, public officials have rescinded LGBTQ Pride proclamations after their behavior was made visible. An implication of these dynamics is that existing audit studies likely understate the extent to which public officials discriminate. Our study addresses this problem by requesting constituent services that are public-facing rather than private. In doing so, we measure the willingness of public officials to provide visibility to LGBTQ people through a pride proclamation. Thus, our study demonstrates how audit experiments can measure discrimination in contexts that require public officials to represent a marginalized group publicly.
Local Representation
Representatives’ responsiveness to constituents is a critical component of democratic politics. Scholars have examined factors that explain the representation of marginalized groups in local politics (Haider-Markel, Joslyn, and Kniss Reference Haider-Markel, Joslyn and Kniss2000; Holman Reference Holman2017; Trounstine Reference Trounstine2010). Studies find that electoral institutions and gender stereotypes shape where and how women run for office (Anzia Reference Anzia2019; Crowder-Meyer, Gadarian, and Trounstine Reference Crowder-Meyer, Gadarian and Trounstine2015; Krebs and Wagner Reference Krebs and Wagner2023) and that the size of minority populations is associated with increased representation (Trounstine Reference Trounstine2010). Orr et al. (Reference Orr, Morel and Fraga2018) find that approximately 67% of Latine elected officials serve at the municipal level. Most Latine mayors represent cities that are majority Latine (Cuéllar Reference Cúellar, Orr and Morel2018). Studies show that Latines are less likely to win elections when turnout is low (Hajnal and Trounstine Reference Hajnal and Trounstine2005). As the Latine population has increased in the United States, Latines have moved from large urban centers to smaller cities and towns, leading to increased local representation (Fraga et al. Reference Fraga, Garcia, Hero, Jones-Correa, Martinez-Ebers and Segura2010; Fry Reference Fry2008; Orr, Morel, and Fraga Reference Orr, Morel and Fraga2018). These studies suggest Latine mayors would be responsive to Latine constituents, who facilitate their election. Other research finds that Latine mayors de-emphasize their Latine identity in campaigns (Orr, Morel, and Fraga Reference Orr, Morel and Fraga2018), which suggests that shared identity between Latine mayors and constituents may be less salient. Furthermore, existing research is inconclusive about the relationship between Latine constituents and representatives at the local level.
There are similar representation patterns among LGBTQ representatives but scholarship is limited. First, a majority of LGBTQ representatives serve at the local level. The Victory Fund, an interest group that collects data on LGBTQ representation, reported that 68% of openly LGBTQ elected officials serve at the local level. Second, research finds that LGB representatives de-emphasize their sexual identities in campaigns (Haider-Markel Reference Haider-Markel2010). Other research about local politics has studied processes of how affirmation and morality politics shape LGBTQ politics (Button, Wald, and Rienzo Reference Button, Wald and Rienzo1997; Cravens Reference Cravens2015; Mucciaroni Reference Mucciaroni2011; Wald, Button, and Rienzo Reference Wald, Button and Rienzo1996). A major difference between LGBTQ and Latine representation at the local level is that the LGBTQ population is numerically small (Sherrill Reference Sherrill1996). This difference has implications for responsiveness that are not addressed in the literature. On the one hand, LGBTQ people can be ignored because they are not influential. On the other hand, LGBTQ representatives may be especially likely to respond to constituency service requests from LGBTQ constituents. Our study addresses this gap.
Policies are also an important component of representation. Previous studies reveal that local officials are responsive to the public opinion of their constituents (Tausanocitch and Warshaw Reference Tausanocitch and Warshaw2014). Studies also find that policies vary based on the composition of local governments (Trounstine Reference Trounstine2010), the size of minority populations (Cravens Reference Cravens2015; Cuéllar Reference Cúellar, Orr and Morel2018; Haeberle Reference Haeberle1996), and ideology (Warshaw Reference Warshaw2019). Relatedly, cultural and morality politics influence the adoption of pro-LGBTQ policies (Cravens Reference Cravens2015; Haider-Markel, Joslyn, and Kniss Reference Haider-Markel, Joslyn and Kniss2000; Wald, Button, and Rienzo Reference Wald, Button and Rienzo1996). Other studies find that discrimination decreases when LGBTQ ordinances pass, but the conclusion is drawn from self-reports from officials who face incentives to under-report discrimination (Button, Wald, and Rienzo Reference Button, Wald and Rienzo1997). In addition, we do not know whether nondiscrimination ordinances influence the responsiveness of mayors to LGBTQ people. As a result, our study examines how policies influence responsiveness at the local level.
The Representation of Latine LGB People
In this section, we examine research about descriptive and substantive representation based on race and ethnicity, sexuality, and gender. In doing so, we identify gaps that our study addresses. Scholars have argued that the descriptive representation of underrepresented groups should lead to substantive representation because of the shared experiences between the representative and group members (Mansbridge Reference Mansbridge1999). Research finds, for example, that Latine representatives are more likely to represent issues that are salient within the Latine community (Wallace Reference Wallace2014). Studies have also shown that Latines’ descriptive representation increases political engagement and empowers group members (Barreto Reference Barreto2010; Rocha et al. Reference Rocha, Tolbert, Bowen and Clark2010; Sanchez and Morin Reference Sanchez and Morin2011). Other work, however, shows a weaker relationship between Latine representatives and constituents (Jones Reference Jones2016). Furthermore, a debate exists about the overall quality of Latine descriptive representation and whether representatives adequately represent intersectionally marginalized subgroups (Beltrán Reference Beltrán2010; Clifford Reference Clifford2012; Fine and Avery Reference Fine and Avery2014; Griffin and Newman Reference Griffin and Newman2007; Mendez Reference Mendez2018; Pleites-Hernandez and Kelly Reference Pleites-Hernandez and Kelly2022; Rocha and Matsubayashi Reference Rocha and Matsubayashi2013; Rocha et al. Reference Rocha, Tolbert, Bowen and Clark2010). These debates and competing findings suggest further research is needed. Our study contributes to these debates by examining responsiveness to Latine lesbian and gay couples, who are marginalized subgroups constituted at the intersections of sexuality, gender, and race.
LGBTQ people and issues are understudied in literature about Latine representation. This gap is notable for several reasons. First, research has demonstrated that the likelihood of a city having an LGBTQ nondiscrimination ordinance decreases as the size of the Latine population increases (Cravens Reference Cravens2015), suggesting the potential influence of straight Latine people on pro-LGB policy adoption. This is consistent with intersectional approaches which hypothesize that Latine representatives would follow the public opinion of straight Latines and be less responsive to Latine same-sex couples (Cohen Reference Cohen1999; Strolovitch Reference Strolovitch2007). Second, scholars have written about the influence of religion on the political attitudes and behavior of Latines, finding evidence that Latines have been less supportive of same-sex marriage than other groups (Sherkat Reference Sherkat2017). Studies also show that Protestant Latines are more conservative than Catholic Latines (Chaturvedi Reference Chaturvedi2014; Ellison, Acevedo, and Ramos-Wada Reference Ellison, Acevedo and Ramos-Wada2011). These findings imply that Latine representatives may be unwilling to publicly represent LGB people through pride proclamations. Third, on the other hand, studies reveal that Latine LGB people are more politically active than straight Latines (Moreau, Nuño-Pérez, and Sanchez Reference Moreau, Nuño-Pérez and Sanchez2019). If representatives pay attention to mobilized subgroups, then they may be more responsive (Bishin and Smith Reference Bishin and Smith2013). Furthermore, the shared in-group racial identity between representative and constituent could lead to responsiveness relative to representatives who do not share a racial identity with constituents. This study provides a direct test of those competing expectations.
The literature on LGBTQ representation has also examined descriptive representation. Studies have shown that LGBTQ representatives are more likely to introduce and sponsor LGBTQ policies and that descriptive representation is correlated with passing pro-LGBTQ bills (Haider-Markel Reference Haider-Markel2010; Hansen and Treul Reference Hansen and Treul2015; Reynolds Reference Reynolds2013). These studies suggest that LGBTQ representatives would be responsive to LGBTQ constituents relative to straight representatives, especially since we are inquiring about a pride proclamation. Since the constituency service request is about LGBTQ pride, the representative-constituent relation between Latine same-sex couples and LGBTQ representatives is different from the relation when race is the shared identity characteristic. Most importantly, the salience of the issue differs because the shared sexual orientation between representative and constituent is aligned with the issue in the constituency service request. As a result, we can reasonably expect higher response rates from LGBTQ representatives than from Latine representatives. At the same time, approximately two-thirds of local elected officials who are LGBTQ are White and just 14% are Latine, which suggests that race may still be salient. Hindman (Reference Hindman2019) and Murib (Reference Murib2023) show that White gay men and lesbians are represented at the expense of LGBTQ people of color. These dynamics of exclusion may carry over to LGBTQ representatives and their responsiveness to Latine same-sex couples. Thus, it is imperative to test rather than assume responsiveness.
Theoretical Expectations
Our theoretical expectations are derived from research that uses intersectional frameworks and methodologies. As previous research has demonstrated, intersectionality is not a uniform approach and there are multiple ways in which researchers can conduct intersectional analyses. In this study, we use an intercategorical approach to examine responsiveness to Latine same-sex couples. Initially articulated by McCall (Reference McCall2005), the intercategorical approach assumes “that there are relationships of inequality among already constituted social groups,” which become the focus of analysis and explication (1784–5). Thus, intercategorical approaches place emphasis on relationality, using existing categories as reference points in analyses (Glenn Reference Glenn2002; Montoya et al. Reference Montoya, Bejarano, Brown and Gershon2022). This scholarship leverages analysis of multiple groups and systematic, relational methodologies to explain inter- and intragroup differences and inequalities (Masuoka and Junne Reference Masuoka and Junne2013; Medenica and Fowler Reference Medenica and Fowler2020; Proctor Reference Proctor2022b). Within our study, we fix the race and sexual orientation characteristics of our constituents while varying their gender. This allows us to examine intragroup differences in responsiveness to lesbians and gay men. In addition, we analyze intergroup differences among representatives by measuring responsiveness to our inquiry based on the identity characteristics of mayors. Because of this complexity, our study has scope conditions that facilitate making relational comparisons at the intersection of sexuality, gender, and race.Footnote 2
Since we are not comparing responsiveness between White and Latine LGB people, it is important to make clear how our approach is intersectional. Our study employs an empirical framework that centers the relational positionality of constituents and representatives across multiple characteristics (Montoya et al. Reference Montoya, Bejarano, Brown and Gershon2022). As a result, we can evaluate how Latine, LGBTQ, and women representatives respond to Latine LGB people compared to non-Latine, straight people, and men. This approach provides a rich set of comparisons that allow us to disentangle how representatives from marginalized groups are responsive to subgroup members. In addition, it allows us to take seriously the ways in which Latine, LGBTQ, and women representatives face distinct constraints that could produce differences in responsiveness. Existing audits that use an intersectional approach have already identified that resource and institutional constraints affect responsiveness (Bauer and Cargile Reference Bauer and Cargile2023), although the focus was on state representatives. One possible constraint in this study is the Latine population, since most Latine mayors represent cities that are majority Latine (Cuéllar Reference Cúellar, Orr and Morel2018). In addition, the number of Latine representatives is greater in states with larger Latine populations (Casellas Reference Casellas2009) and increases in Latine population are negatively associated with cities adopting LGBTQ nondiscrimination policies (Cravens Reference Cravens2015). Thus, Latine mayors may face pressures to represent their straight Latine constituents at the expense of LGB Latines.
LGBTQ representatives face different constraints than Latine representatives, which could lead to differential responsiveness. In other words, representatives are embedded within distinct contexts that may affect their behavior. First, LGBTQ people are a numerically small group and lack the geographic concentration to be a majority electoral constituency. This could increase the likelihood that an LGBTQ representative would respond. Second, the content of our request is about an LGBTQ political issue making a response more likely from an LGBTQ representative. Furthermore, on a general level, the positionality between representatives who share a sexual identity but not a racial identity with a constituent is distinct from the positionality of representatives who share a racial identity but not a sexual identity. These differences are further accentuated when gender is brought into the mix, demonstrating a need for an intercategorical approach. Moreover, we still have more to learn about the representative-constituent relationship with regard to marginalized subgroups. Our study contributes to this gap by examining responsiveness to Latine same-sex couples.
Our theory builds on previous audit experiments by using an intersectional framework to identify competing hypotheses about responsiveness to Latine gay and lesbian couples. Previous audit experiments have generally theorized about discrimination along a single axis of marginalization or treated intersecting axes of marginalization as additive (Crenshaw Reference Crenshaw1989; Hancock Reference Hancock2007), although there are some exceptions (Bauer and Cargile Reference Bauer and Cargile2023). Audit experiments that examine discrimination against LGB people have used putatively White names and not theorized about race (Bailey, Wallace, and Wright Reference Bailey, Wallace and Wright2013; Lowande and Proctor Reference Lowande and Proctor2020; Mishel Reference Mishel2016; Tilcsik Reference Tilcsik2011). As a result, previous findings likely do not reflect experiences with discrimination among LGB people who are from marginalized racial groups. In addition, these studies reach varying conclusions about whether gay men and lesbians face discrimination compared to straight people and whether lesbians and gay men are discriminated against in the same way.
We argue that audit studies should theorize about marginalization through an intersectional lens. We do so by developing hypotheses that recognize how intersecting categories of marginalization produce mutually constitutive, distinct experiences. In this framework, experiences of marginalization are not the sum of the categories through which people are marginalized. An additive approach would hypothesize that Latine lesbians would be responded to less than Latine gay men because they experience marginalization due to their race, gender, and sexuality while Latine gay men experience marginalization through race and sexuality but not gender. Findings from previous research demonstrate how these assumptions are flawed. For example, some audit studies find that men respond to women constituents at higher rates and others find that public officials do not respond at different rates to men and women (Kalla, Rosenbluth, and Teele Reference Kalla, Rosenbluth and Teele2018; Thomsen and Sanders Reference Thomsen and Sanders2020). An additive framework would hypothesize that women should be responded to less than men in these studies. The evidence suggests otherwise.
In contrast, our intersectional framework recognizes that intersecting categories of marginalization lead to distinct experiences with discrimination. Latine gay men and lesbians occupy distinct positions within intersecting hierarchies of sexuality, race, and gender. As a result of these distinct positions, we allow for the possibility that Latine gay men are responded to at lower rates than lesbians. Research on sexual prejudice lends credence to this possibility. Studies have shown that men and women provide more positive evaluations of lesbians than gay men (Bettinsoli, Suppes, and Napier Reference Bettinsoli, Suppes and Napier2020; Herek Reference Herek2000; Herek and Capitanio Reference Herek and Capitanio1999). These findings suggest that the intersection of sexuality and gender shapes the contours of LGB discrimination. More generally, they lend support to our argument that intersectional frameworks should inform audit studies. In service of this argument, we test competing hypotheses about whether responsiveness to constituencies varies based on their distinct position within intersecting systems of sexuality, gender, and race:
H1a: Mayors, on average, will be more responsive to Latine lesbian couples than to Latine gay couples.
H1b: Mayors, on average, will be more responsive to Latine gay couples than to Latine lesbian couples.
Second, we hypothesize that the identity characteristics of mayors influence their responsiveness to Latine same-sex couples. Extant research suggests that representatives have an intrinsic motivation to represent communities based on mutually experienced descriptive attributes like gender, race, sexuality, or membership in a historically marginalized population (Dovi Reference Dovi2002; Phillips Reference Phillips1998; Pitkin Reference Pitkin1967). These representatives, by virtue of their lived experience, tend to be acutely aware of their group’s policy prerogatives and pursue them when in office. Moreover, given the frequency in which marginalized groups are sidelined within majoritarian institutions, descriptive representatives are often the best advocates for their own group’s rights (Mansbridge Reference Mansbridge1999). Research has demonstrated that women, Latine, and African American legislators more ardently promote their group’s interests than men and White lawmakers (Griffin and Keane Reference Griffin and Keane2006; Christian R. Grose Reference Grose2005; Juenke and Preuhs Reference Juenke and Preuhs2012; Swers Reference Swers2005). This has come by way of roll-call voting (Wallace Reference Wallace2014), bill sponsorship (Rocca and Sanchez Reference Rocca and Sanchez2008; Swers Reference Swers2005), meaningful symbolic actions (Tate Reference Tate2004), constituency service (Broockman Reference Broockman2013; Christian R Grose Reference Grose2011), and the allocation of federal projects (Christian R Grose Reference Grose2011). Likewise, research finds that LGBTQ representatives are more likely to sponsor pro-LGBTQ legislation at the state and federal level (Haider-Markel Reference Haider-Markel2010; Hansen and Treul Reference Hansen and Treul2015). In addition, representatives from marginalized groups are also more likely to support policies that help other marginalized groups (Bratton and Haynie Reference Bratton and Haynie1999). This research motivates our second set of hypotheses about the influence of Mayors’ identity characteristics on responsiveness to Latine LGB couples:
H2a: LGBTQ Mayors, on average, will be more responsive to Latine LGB constituents than non-LGBTQ Mayors.
H2b: Latine Mayors, on average, will be more responsive to Latine LGB constituents than non-Latine Mayors.
H2c: Women Mayors, on average, will be more responsive to Latine LGB constituents, than Mayors who are men.
Third, in addition to the identity characteristics of constituents and mayors, we hypothesize that institutional and partisan electoral contexts influence mayors’ responsiveness to public-facing service requests. Since the 1970s, cities and states have passed ordinances that protect lesbian, gay, and bisexual people from discrimination. The cumulative number of ordinances has increased over time (Button, Wald, and Rienzo Reference Button, Wald and Rienzo1997; Haider-Markel and Joslyn Reference Haider-Markel and Joslyn2013). Some scholars question how well these ordinances are enforced (Riccucci and Gossett Reference Riccucci and Gossett1996), while others find that LGB people are more likely to live and work where they exist (Lewis and Pitts Reference Lewis and Pitts2011). We use nondiscrimination policies that protect gay men and lesbians from discrimination to measure institutional constraints that might influence a mayor’s willingness to issue a pride proclamation.
Although LGBTQ people are protected from employment discrimination at the federal level (Bostock v. Clayton County 2020), they lack protection from discrimination in housing and public accommodations. The absence of broader federal protections has produced a patchwork of protections across the United States (MAP 2020). Thus, local and state ordinances are critical policies that protect the rights and liberties of LGB people. We argue that nondiscrimination ordinances can have consequences beyond their intended effect of providing legal protections to LGB people. In particular, they may shape the behavior of mayors and how they publicly represent their constituents through service provision. Mayors who represent cities where LGB protections exist at the local or state level are embedded within institutional contexts in which discrimination is (legally) prohibited. As a result, they should be more likely to respond to service requests from LGB constituents than mayors who represent cities where local or state protections do not exist. Thus, we advance the following hypothesis:
H3: Mayors of cities with existing LGBTQ protection ordinances will, on average, be more responsive than mayors of cities without existing LGBTQ protection ordinances.
Finally, we emphasize that local partisan contexts are also important political institutions that affect the behavior of mayors. While individual voters do not constitute an “institution,” their aggregate behavior in the electorate enables and constrains the behavior of elected officials, who require their support to win public office (Key Reference Key1968). Therefore, the partisan contexts in which mayors are embedded should influence constituency responsiveness. Partisan contexts are especially relevant when evaluating responsiveness to LGB people because of the polarization of LGB issues in American politics (Bishin, Freebourn, and Teten Reference Bishin, Freebourn and Teten2021; Karol Reference Karol2009; Saraceno, Hansen, and Treul Reference Saraceno, Hansen and Treul2021). Mayors embedded within Democratic partisan contexts should be expected to respond to LGB constituent requests at higher rates than mayors embedded within Republican electoral contexts. As such, we advance our final hypothesis:
H4: Mayors of cities in increasingly Democratic counties will, on average, be more responsive than mayors of cities in Republican-leaning counties.
Experimental Design
We conducted an experimental audit of mayors from approximately 1400 cities across all 50 states. Following previous studies that have examined the responsiveness of city leaders (e.g. Moy Reference Moy2020), our sample comes from a database collected by the United States Conference of Mayors (USCM). The USCM is a nonpartisan organization that primarily represents the chief executives of US cities with a population of 30,000 or more, but occasionally partners and collects information on select municipalities with less than 30,000 residents. Figure 1 displays the cities included in our dataset. Mayors were randomly assigned to receive an email inquiring about their city’s plans to recognize June as LGBTQ Pride Month. In each case, the sender was a recently engaged member of the city’s LGBTQ community, writing on behalf of themselves and their same-sex partner. Given this study’s emphasis on multiple, intersecting forms of marginalization we opted to use aliases that were (1) among the most common Latine first and last names and (2) unambiguously conveyed the gender of the sender. The aliases we used in the study were Maria Rodriguez, Diego Rodriguez, Alejandra Martinez, and Carlos Martinez.Footnote 3
Grey dots correspond to the 1396 cities in which mayors received our experimentally manipulated email message.
Our decision to focus on same-sex marriage and LGB constituents was motivated by Lowande and Proctor (Reference Lowande and Proctor2020), which found greater responsiveness to lesbian couples than gay couples. The goal of this research was to test whether similar differences emerged in a different representative context involving elected officials rather than bureaucrats. We decided to link same-sex marriage to LGBTQ pride due to backlash against LGBTQ pride in recent years. We also acknowledge that some of this backlash has been motivated by antitransgender politics, which we do not address in this manuscript. Furthermore, we acknowledge that there are many important LGBTQ issues that we could have used instead of same-sex marriage. As a result, we note that this study examines responsiveness to LGB people who participate in normative institutions, which could impact the degree to which we observe discrimination. Finally, we chose to make a public-facing request for a pride proclamation because we were concerned about the ethics of lobbying officials over policy (Bergan Reference Bergan2009) and because most audits request private information. The template used to send messages to mayoral offices is displayed below.
Subject Line: Proclamation Information Request
Mayor [Mayor Last Name],
My name is [man/woman sounding first name]. I am writing because my [boyfriend/girlfriend] and I recently got engaged and are planning a wedding to take place in June of this year. As members of [city name]’s LGBTQ community, we were wondering if the mayor’s office was planning on issuing a proclamation recognizing June as LGBTQ Pride Month. Many members of our community continue to face hardship because of their identity and we believe this would be a reaffirming gesture from the city we all love. If not already on the agenda, is this something that your office might consider? If the city already plans to issue a pride proclamation, how can we obtain a copy?
Thank you,
[man/woman sounding first and last name]
Ethical Considerations
It is widely understood that experimental research has undergone a revolution within political science (Grose Reference Grose2014; Saraceno Reference Saraceno2020). As such, extensive work has documented the ethical considerations that must be considered when conducting experimental research, especially when involving public or elected officials (e.g. Butler Reference Butler2014; Einstein and Glick Reference Einstein and Glick2017; Grose Reference Grose2014; McClendon Reference McClendon2012). We made every effort to conform to these guidelines in the experiment described above. For instance, we contacted mayors in their professional capacity and requested information that was (in virtually all instances) only available from their offices. We further worked to minimize the time cost for mayoral staff by limiting our interaction to a single email, regardless of whether a response was received. Specific words like “planning” and “agenda” were used to relate our information request to an easily-identifiable calendar event. Lastly, our email sought only to gauge whether the mayor’s office would be open to a pride proclamation request, rather than an official request for such.
One potential concern about our research design is that it could be interpreted as lobbying for nondiscrimination ordinances due to the public-facing nature of the request. Previous research has shown, for example, that email communications that lobby legislators influences their behavior (e.g. Bergan Reference Bergan2009). We acknowledge this concern and researchers conducting audit experiments should take step to minimize lobbying. However, we also argue that issuing proclamations is a common, nonpolicy tool used by public officials. Many local governments have websites where requests can be made or where information about making requests can be obtained. For example, in the city of Rockford, Illinois, the local government has a dedicate page for making proclamation requests.Footnote 4 We also note that an internet search turned up similar websites for cities around the country. Often, these webpages also make clear that proclamations are not endorsements. As a result, we believe that the mayors’ offices receiving our requests for a pride proclamation can distinguish them from lobbying for policy.
We also acknowledge that even relatively minor interventions can impose burdens when considered in the aggregate. With this in mind, we follow Lowande and Proctor (Reference Lowande and Proctor2020) in estimating the externalized costs to the public associated with our study.Footnote 5 Among the 646 email responses we received, the average number of typed words—including salutations, signatures and contact information, and substantive responses—was 90. If it is assumed that mayors and their staff type at the speed of the typical American, the average time to compose a response was roughly 2 min. We can use the same approach to calculate that the average time spent reading our 128-word email was roughly 30 s. It follows that a conservative estimate (deliberately overestimated) for the average time cost borne by those who engaged with our experiment is 3 min. Thus, the cumulative amount of time spent by those who theoretically read our email but did not respond (750 × 30 s) and those who read and responded to our email (646 × 3 min) is 2,313 min or roughly 38.5 h. By multiplying this time cost by the median hourly wage of mayors,Footnote 6 we approximate the externalized cost of our experiment to be roughly $1,993.15.
Response Coding and Dependent Variables
We evaluate mayoral responsiveness to LGB Latine constituents by coding how they respond to the request described above. Two trained research assistants were tasked with independently coding four distinct categories of mayoral responsiveness: response, consideration, commitment, and congratulations. They were not given information about the research question, the nature of the experiment, identifying information for mayoral offices, or the names of the fictitious senders. The research assistants were first instructed to consider whether a nonautomated response was received from the mayor’s office. We consider this the lowest responsiveness threshold—one that could be met by simply responding to our email request without necessarily addressing all of the questions posed. Indicating only that a pride proclamation had been made at some point in the past, for instance, was coded simply as a response. It is worth noting that automatic responses or “out of office the office” messages are commonly set up by elected officials to indicate that the email has been received and that a follow-up should be expected. If only an automated response was received after 14 days, the observation was coded 0 for all responsiveness categories.
In the cases where a nonautomated message was received, the research assistants were asked to code whether the mayor’s office would consider a request for a pride proclamation. A message indicating willingness to do so is substantively important as it implies that a public official is open to providing visible recognition to LGBTQ people—although it falls short of an explicit commitment. Consistent with this definition, messages like, “I would be more than happy to send something to the Council for their evaluation” or “[t]he city has not done this type of proclamation before, but that doesn’t mean we can’t – go ahead and fill out the attached form,” were coded “1”. Our third response category goes further by identifying the instances in which an explicit commitment to issue a proclamation recognizing pride month is made by the mayor or their staff. This included cases where our sole email was a catalyst that led to a pride proclamation and cases where a pride proclamation was already in the works. Theoretically, a stated commitment is distinctly significant as it demonstrates a mayor’s willingness to make a public-facing, symbolic action in support of a minoritized group—despite the consequences that may arise. Following previous work, our research assistants also measured whether the response indicated a congratulatory message related to the couple’s recent engagement. It is common to receive a congratulatory response when discussing an impending marriage. Thus, the lack of such a message can provide an, albeit blunt, insight into mayoral attitudes toward same-sex couples (2020). Lastly, when the research assistants disagreed, we adjudicated the discrepancy between coders (roughly 8% of coded responses).
Results
We start by noting that, mayors responded at a rate of 44.1%, indicated they would consider issuing a pride proclamation at a rate of 39.7%, committed to issuing a pride proclamation at a rate of 29.1%, and offered a congratulatory message at a rate of 25%. As illustrated in Figure 2, we received responses from mayoral offices across the country. Next, we conduct bivariate analysis to test our hypotheses. Once we establish baseline levels of responsiveness, we provide additional evidence through a multivariate regression analysis that includes our treatment indicator, local ordinance measures, partisan electoral contexts, mayors’ identity characteristics, and controls for city population, county-level Latine population, county-level LGBTQ population, and county-level evangelical population.
States are shaded based on the percentage of the sample that responded to our study. We present these data at the state-level to prevent disclosure about who responded to our inquiry and who did not.
Differential Responsiveness to Lesbian and Gay Couples
We find mixed evidence of systematic differences in responses to Latine lesbian and gay couples across our dependent variables. Figure 3 illustrates the difference-in-means response rates for Latine lesbian and gay couples across our four dependent variables: overall response rate, considering issuing a proclamation, commitment to issue a pride proclamation, and congratulatory message. The difference-in-mean for overall response rate to lesbian and gay couples is marginally significant (diff = 0.05, p = 0.06, two-tailed). Mayor’s offices responded to the lesbian couple at a rate of 46.6% compared to a rate of 41.6% for the gay couple. Mayor’s offices were 42.7% likely to consider issuing a pride proclamation for the lesbian couple and 36.9% likely to do so for the gay couple (diff = 0.058, p < 0.05, two-tailed). The difference-in-means for committing to issue a pride proclamation and congratulatory messages do not reach statistical significance. Thus, we find that mayors’ offices are more responsive to and more likely to consider issuing a pride proclamation for lesbian couples than gay couples but they make explicit commitments to completing constituency service requests at similar rates.
Plots illustrating the difference in mean response rates (and 95% confidence intervals) between Latine lesbian couples and Latine gay couples. Full results can be found in Table A1.1 in the Online Appendix.
Mayor Identity Characteristics
Figure 4 plots the difference-in-mean responsiveness to LGB constituents by mayors’ identity characteristics. In these models, we examine whether mayoral identity shapes responsiveness to our inquiry regardless of whether the inquiry is from a lesbian or gay constituent. In other words, does the decision to respond to Latine same-sex couples vary by the identity characteristics of mayors? We find support for our hypotheses that LGBTQ and women mayors will be more responsive to Latine LGB constituents than straight, cisgender mayors and men. Mayors who identify as LGBTQ are approximately 28 percentage points more likely to reply to LGB Latine constituents than mayors who are straight and cisgender. Likewise, LGBTQ mayors are 30 percentage points more likely to consider a pride proclamation and 43 percentage points more likely to commit to issuing a proclamation. LGBTQ mayors are also more likely to say congratulations to their constituents than straight, cisgender mayors (p < 0.01 for all comparisons). These results demonstrate that descriptive representation of LGBTQ people in public office increases responsiveness to LGB constituents.
Plots illustrating predicted response rates (and 95% confidence intervals) by Mayor identity characteristics. Full results are in the Online Appendix Tables A1.2–A1.4.
We do not find greater responsiveness between Latine mayors and LGB Latine constituents compared to non-Latine mayors. In fact, Latine mayors were approximately 9 percentage points less likely to respond to LGB Latine constituents than non-Latine mayors (p < 0.05). Likewise, they are approximately 10 percentage points less likely to commit to issuing a pride proclamation relative to non-Latine mayors. We do not observe a difference between Latine and non-Latine mayors for consideration of issuing a pride proclamation and congratulatory message. This evidence demonstrates that the decision to respond to Latine same-sex couples varies by mayor identity characteristics. When it comes to groups constituted at the intersection of race and sexuality, LGB Latine constituents are responded to less by Latine representatives despite sharing an in-group racial identity. Thus, a shared in-group identity between a constituent and a representative does not inevitably lead to greater responsiveness. A possible explanation that is consistent with intersectional scholarship is that a majority of Latine mayors are straight and therefore less likely to respond to their LGB Latine constituents.
Finally, women mayors are more responsive to Latine LGB constituents than mayors who are men. Women mayors are approximately 9 percentage points more likely to respond, consider, and commit to issuing a pride proclamation relative to men. Women representatives are also more likely to say congratulations than men representatives (p < 0.01 for all comparisons). In addition, although women mayors are more responsive overall compared to men, panel 3 in Figure 4 shows that women mayors are still less likely to commit to issuing a pride proclamation compared to LGBTQ mayors. This further demonstrates that LGBTQ descriptive representation is critical for increasing the visible recognition of LGB people through pride proclamations at the local level.
Local Political Institutions
We find strong support for our third hypothesis that local institutions affect responsiveness to Latine same-sex couples. Our analysis demonstrates that mayors in cities where there are local or state ordinances protecting LGB people from discrimination are more likely to respond, more likely to consider and commit to issuing a proclamation, and more likely to say congratulations than mayors who represent cities without legal protections (p < 0.01, two-tailed, for all models). The difference-in-mean mayoral responsiveness by local LGB ordinance status is reported in Figure 5. In cities with nondiscrimination ordinances, the overall response rate was 50.1% compared to 34.2% in cities where there are no local protections, a difference of 16 percentage points. We find that 46% of mayors in cities with protections would consider issuing a proclamation compared to 29.4% of mayors in cities without protections. Likewise, there is a 15.9 percentage point difference when it comes to making explicit commitments to issue a proclamation. Mayors who represent cities with LGB protections committed to issuing a proclamation in 35.1% of responses compared to just 19.3% of responses from Mayors who represent cities without LGB protections. Mayors representing cities with LGB protections were about 30.1% likely to issue a congratulations in their response, while mayors representing cities without LGB protections said congratulations at a rate of 17.2%. Latine same-sex couples received greater responsiveness from mayors who represent cities with LGB protection ordinances compared to mayors who represent cities without LGB protection ordinances. Thus, local institutional contexts in which mayors are embedded affect their responsiveness to LGB Latine people.
Plots illustrating the difference in mean response rates (and 95% confidence intervals) between cities with local LGBTQQ+ protection ordinances and those with no such protections. Full results are in the Online Appendix Table A1.5.
Partisan Electoral Contexts
Our analyses of the effect of local partisan contexts on mayoral responsiveness reveals support for our fourth hypothesis. Using county-level Democratic vote share from the 2020 presidential election, we find that overall responsiveness, consideration of and explicit commitment to issue a proclamation, and including a congratulatory message increase as partisan electoral context shifts from most Republican to most Democratic (p < 0.01, two-tailed). Figure 6 presents the predicted level of responsiveness as share of Democratic vote at the county-level increases. When Democratic vote share is 25%, for example, the predicted response rate is 31.7%. In contrast, when Democratic vote share is 75%, the predicted response rate is 52.6%. We can further contextualize these results by examining the threshold in which response rates are predicted to exceed 50%. The predicted overall response rate does not surpass 50% until a county is 68% Democratic. This suggests that a large partisan constituency at the county-level is necessary for the probability of response to exceed that of a coin flip. Thus, as with our LGB ordinances analysis, we find that local partisan contexts influence the behavior of mayors.
Plots illustrating predicted response rates (and 95% confidence intervals) across various levels of Democratic presidential vote share. Full results are in the Online Appendix Table A1.6.
Multivariate Regression
We also conducted multivariate analysis using ordinary least squares regression to test the robustness of our bivariate results and to compare our hypotheses in a single model. Results from logistic regression are reported in Table A1.7 in the online appendix and are substantively and statistically similar. We regress responsiveness measures on our treatment indicator, mayors’ identity characteristics, local ordinances, and electoral context in addition to controls for LGBTQ population in a county, logged city population, county-level Latine population, and county-level evangelical population. The regression coefficients are reported in Table 1 and demonstrate that our results are robust to additional model specifications. We again find support for hypothesis 1a. Mayors were more likely to respond to requests from lesbian couples relative to gay couples as well as more likely to consider and commit to issuing a pride proclamation. The effects are similar in magnitude compared to the bivariate analysis. Mayors were 5.4% more likely to respond and 6.2% more likely to consider issuing a proclamation when the constituency request came from a lesbian couple (p < 0.05). Mayors are 4.4% more likely to commit to issuing a pride proclamation for lesbian couples, noting that this coefficient only reaches statistical significance if we relax our threshold to p < 0.1. We find null effects for the effect of couple gender on congratulatory messages. Thus, our evidence consistently shows that lesbian couples receive greater responsiveness than gay couples.
Ordinary least squares regression predicting overall response rate, willingness to consider issuing a proclamation, commitment to issuing a proclamation, and the inclusion of a congratulatory message
${{\rm{\;}}^{{\rm{***}}}}p$
< 0.001,
${{\rm{\;}}^{{\rm{**}}}}p$
< 0.01,
${{\rm{\;}}^{\rm{*}}}p$
< 0.05,
${{\rm{\;}}^ \circ }p$
< 0.1.
We find support for hypothesis 2a but not 2b. Mayors who identify as LGBTQ are 24% more likely to respond, 27.1% more likely to consider, and 38.7% more likely to commit to issuing a pride proclamation. They are also 25.5% more likely to say congratulations to the couple. Thus, we have robust evidence that the descriptive representation of LGBTQ people in mayor’s offices increases responsiveness to LGB constituents. While descriptive representation by LGBTQ mayors increases responsiveness, we do not observe a similar effect for Latino mayors. We also find support for hypothesis 2c. Women mayors are more responsive to Latine same-sex couples than men who are mayors. Women mayors are 7.4% more likely to respond and 11.3% more likely to include a congratulatory message. Likewise, when the threshold for statistical significance is relaxed to p < 0.1, women mayors are 5.8% more likely to consider and 5.3% more likely to commit to issuing a pride proclamation. These results demonstrate that identity characteristics of mayors are factors that shape the decision to respond to constituency requests from Latine same-sex couples.
With evidence that lesbian couples are responded to more than gay couples and evidence that mayor identity characteristics are associated with responsiveness to our inquiry, we examine whether responsiveness to lesbians is moderated by mayoral identity characteristics. Thus, we respecify three multivariate regression models that interact with our treatment and mayor identity characteristics. The results for these regressions are reported in Tables A.10–A.12 in the online Appendix. We find substantively similar main effects in these analyses. Lesbians are responded to more than gay men and LGBTQ and women mayors are more responsive to requests from Latine same-sex couples than straight, cisgender representatives and men. Latine mayors are not more responsive to Latine same-sex couples than non-Latine representatives. The interaction terms, however, do not reach conventional levels of statistical significance.
The multivariate regression also demonstrates support for our institutional ordinance hypothesis (H3). Ordinances protecting LGB people from discrimination are associated with increased mayoral responsiveness to constituency service requests from Latine same-sex couples. Mayors from cities with ordinances protecting LGB people from discrimination were 9.5% more likely to respond, 9.6% more likely to consider issuing a proclamation, 9.4% more likely to commit to issuing a proclamation, and 7.9% more likely to include a congratulatory message in their response (p < 0.01 for all models, two-tailed). Thus, across all of our responsiveness outcomes, mayoral responsiveness is influenced by the existence of policies that protect LGB people from discrimination.
The results for the influence of county-level Democratic vote share on responsiveness are less robust. First, county-level vote share is only statistically significant and positively associated with receiving a response and willingness to consider issuing a pride proclamation if we relax our threshold for evaluating statistical significance to p < 0.1. Second, county-level Democratic vote share does not reach conventional levels of statistical significance in the models for committing to issuing a pride proclamation or saying congratulations. Thus, we conclude that the partisan electoral contexts in which mayors are embedded are less likely to influence mayoral behavior than constituent identity characteristics, mayor identity characteristics, and nondiscrimination ordinances.
Beyond our key variables of interest, multivariate regression reveals that controls for Evangelical population and Latine population are negatively associated with responsiveness to Latine same-sex couples. Consistent with existing literature, we find that county-level Evangelical population is negatively associated with responsiveness (e.g. Oldmixon and Calfano Reference Oldmixon and Calfano2007). The coefficient for Evangelical population is statistically significant, negative, and large in magnitude for all four measures of responsiveness. Thus, the size of the Evangelical population in a county is a constraint on mayoral responsiveness to LGB people. The size of the Latine population in a county is also negatively associated with responsiveness and large in magnitude. This result is not consistent with theories that rely on a single-axis framework to conceptualize the relationship between constituents and representatives (i.e. that mayors would be responsive to Latine same-sex couples as the Latine population increases). LGB population did not reach conventional levels of statistical significance. Bivariate results for responsiveness on LGB and Latine populations are reported in the appendix.
Discussion
Differential Responsiveness to Latine LGB Couples
This research contributes to a growing body of evidence that government officials are more responsive to lesbians than gay men. Across our analyses, the evidence demonstrates that mayors are more responsive to requests from Latine lesbian couples compared to gay couples. These results align with other national audit experiments that find that bureaucrats are more responsive to lesbians than gay men (Lowande and Proctor Reference Lowande and Proctor2020), while they diverge from older audit experiments where callback rates to gay men and lesbians varied with no discernible pattern (Bailey, Wallace, and Wright Reference Bailey, Wallace and Wright2013). Other studies also find that gay men are more negatively stereotyped and penalized as political candidates compared to lesbians (Doan and Haider-Markel Reference Doan and Haider-Markel2010; Golebiowska Reference Golebiowska2003), a finding that extends beyond the United States (Magni and Reynolds Reference Magni and Reynolds2021). Research about Latina lesbians and bisexuals has also found that feeling accepted within the LGBTQ community, which could be increased by officials’ recognition of LGBTQ people, increased sociopolitical involvement (Battle and Harris Reference Battle and Harris2013). Thus, our findings suggest that Latine lesbians and bisexuals may become more politically active due to greater responsiveness. We caution, however, against overinterpreting that Latine lesbians are better represented due to the relative invisibility of Latine LGB people in politics. Furthermore, other evidence shows that gay men are more likely to serve in public office than lesbians, which may be an outcome of gay men’s over-representation in the LGBTQ movement (Murib Reference Murib2023). Thus, evaluations of lesbian and gay representation are complex and nuanced.
Our evidence that public officials are less responsiveness to gay men compared to lesbians is consistent with cross-national evidence that gay men are more negatively evaluated than lesbians (Bettinsoli, Suppes, and Napier Reference Bettinsoli, Suppes and Napier2020). Studies have shown that straight people have more negative personal reactions to gay men than lesbians (Herek Reference Herek2002) and negatively stereotype gay men for violating gender roles (Madon Reference Madon1997). Gay men also have distinct experiences with bias and discrimination at the intersection of race and sexuality (Ward Reference Ward2008). Within the Latine community, gay, and bisexual men’s experiences with racism, homophobia, and biphobia are associated with negative social outcomes (López et al. Reference López, Yuan, Booth, Wei and Friedman2023). This research demonstrates how the positionality of Latine gay men and lesbians in hierarchies of sexuality, gender, and race constitute distinct experiences of marginalization.
Our findings of greater responsiveness to Latine lesbian couples are in productive conversation with audit studies that examine responsiveness by gender. Kalla, Rosenbluth, and Teele (Reference Kalla, Rosenbluth and Teele2018) find null effects for differential responsiveness to inquiries from men and women. In our study, however, lesbian couples are responded to more than gay couples, demonstrating how the intersection between sexuality and gender shapes responsiveness. We also found that women mayors are more responsive to inquiries from Latine LGB constituents than men (Thomsen and Sanders Reference Thomsen and Sanders2020, see also). These findings align with research showing that women are stereotyped as “better” at constituency service (Crowder-Meyer, Gadarian, and Trounstine Reference Crowder-Meyer, Gadarian and Trounstine2015), research that women are more pro-LGBTQ (Herek and Capitanio Reference Herek and Capitanio1999), and evidence that women are more successful in stereotype congruent elected positions (Anzia Reference Anzia2019). Other research similarly shows that representatives from marginalized groups are more likely to represent the concerns of marginalized groups (Bratton and Haynie Reference Bratton and Haynie1999). As a result, activist groups that focus on electing pro-LGBTQ candidates may consider targeting women, even if they are not LGBTQ.
Descriptive Representation
Our bivariate analyses showed that Latine mayors were less responsive to Latine LGB constituents than non-Latine mayors, while multivariate model revealed that size of the Latine population was negatively associated with responsiveness. This evidence demonstrates that descriptive representation does not inevitably lead to greater responsiveness for intersectionally marginalized constituencies. The conventional wisdom is that elected officials are more responsive to groups as their size in the electorate increases and when they are descriptively represented. In a single-axis framework, this implies that as the size of a racial group increases a public official would become more responsive to people from that group—regardless of sexuality. Our results do not align with this expectation. Thus, our findings contribute to a growing debate about whether descriptive representation and size of the Latine population improve Latine representation (Clifford Reference Clifford2012; Fine and Avery Reference Fine and Avery2014; Griffin and Newman Reference Griffin and Newman2007; Rocha et al. Reference Rocha, Tolbert, Bowen and Clark2010). Other scholarship has similarly shown that intersectionally marginalized groups, such as non-citizens and Latinas, are inadequately represented in Latine politics (Beltrán Reference Beltrán2010; Mendez Reference Mendez2018; Rocha et al. Reference Rocha, Tolbert, Bowen and Clark2010). Thus, a growing body of research lays bare the limitations of theorizing about responsiveness along a single-axis of marginalization.
Although Latine mayors were not more responsive to Latine same-sex constituents, LGBTQ mayors were more likely to respond than straight, cisgender mayors. The descriptive representation of LGBTQ people at the local level increases responsiveness to constituency service requests for a pride proclamation. One possible explanation for greater responsiveness among LGBTQ mayors is that our request was about the recognition of LGBTQ people. In addition, LGBTQ pride is likely perceived as an LGBTQ issue rather than a Latine issue by Latine representatives, especially if they are straight (Wallace Reference Wallace2014). While descriptive representation is associated with responsiveness, we do not find evidence the size of the LGBTQ population influences responsiveness. We speculate that this is because they are a numerically small constituency (Sherrill Reference Sherrill1996). As a result, mayors may not feel constrained by population size. Other studies similarly find that LGBTQ population does not influence representation outcomes for LGBTQ people (Saraceno, Hansen, and Treul Reference Saraceno, Hansen and Treul2021), while some find evidence that group size does improve representation (Haider-Markel Reference Haider-Markel2010; Hansen and Treul Reference Hansen and Treul2015). This emerging debate offers a path for future research on LGBTQ representation.
Nondiscrimination Ordinances
This article demonstrates that local institutions shape responsiveness to constituency service requests from intersectionally marginalized groups. Mayors who represent localities with nondiscrimination ordinances that protect LGB people from discrimination are more likely to respond to inquiries from Latine same-sex couples. They were also more likely to consider and commit to issuing a pride proclamation. We argue that this evidence shows one way in which nondiscrimination ordinances translate into better representation for marginalized groups in the form of responses to constituency service requests. Although issuing a pride proclamation is not directly related to employment and housing policies, these ordinances create institutional contexts that influence the behavior of mayors to be more responsive to Latine same-sex couples.
Readers might be concerned about whether nondiscrimination ordinances are endogenous to responsiveness since these policies are not random and it is possible that mayors from cities with these ordinances would respond anyway. While we agree with these concerns, we note that our results are robust across model specifications that included controls for the Democratic vote share of a county, a measure that should theoretically be correlated with the non-random distribution of LGB ordinances and increased likelihood of a response to Latine same-sex couples. Even after accounting for county-level partisanship, nondiscrimination ordinances are associated with increased responsiveness across all dependent variables. In fact, when both measures are included, nondiscrimination ordinances were more likely to retain statistical significance. Our models also accounted for variables that are associated with LGBTQ politics and minority representation at the local level, including evangelical mobilization (Button, Wald, and Rienzo Reference Button, Wald and Rienzo1997; Wald, Button, and Rienzo Reference Wald, Button and Rienzo1996), Latine population (Cravens Reference Cravens2015), LGB population (Haider-Markel Reference Haider-Markel2010), population size (Trounstine Reference Trounstine2010), and partisanship (Tausanocitch and Warshaw Reference Tausanocitch and Warshaw2014). Thus, we can reasonably conclude that nondiscrimination policies have an independent influence on responsiveness.
An additional concern with our measure is that it combines state-level and city-level ordinances. As a result, all cities within states that have statewide nondiscrimination ordinances are coded as having ordinances, regardless of local policy. Our data are coded this way because we use Movement Advancement Project data that identifies nondiscrimination ordinances at either the state- or local level. When a state has nondiscrimination policies, MAP classifies the entire population as protected. When a state lacks an ordinance, they identify which cities have nondiscrimination ordinances. A related concern is whether mayors are thinking about nondiscrimination policies, since ignoring a request for a pride proclamation is not a violation of them. It is also reasonable that this is likely when a policy exists at the state rather than local level. To account for these concerns, we rerun our analyses with dummy variables for state and local ordinances and find that the effect of policy is similar for both measures (see online appendix). Thus, our data show that state and local policies influence mayors’ behavior. These findings suggest that lobbying at both the state and local level could be effective ways for activists to reduce discrimination.
Comparing Across Audit Studies
Since we are interested in explaining how the intersection of race, sexuality, and gender affects responsiveness, we contextualize our findings against studies of responsiveness to putatively White LGB people as well as other audit studies more broadly. Compared to Lowande and Proctor (Reference Lowande and Proctor2020), response rates to putatively Latine same-sex couples in our study are approximately 30 percentage points lower. Although one possible explanation is that bureaucrats are more likely to respond to constituency service requests than elected officials, a 30 percentage point difference in response rate is so large in magnitude that we speculate that it is unlikely just a difference in job function. Consistent with this speculation, in another study of elected officials, the overall response rate to putatively White LGB people was 58% (Saraceno n.d.), a difference of 14 percentage points compared to our results. Together, this body of evidence suggests that race of the respondent influences responsiveness to LGB people. Furthermore, we are only able to identify these differences because we used an intersectional framework to examine the responsiveness of government officials. At the same time, our ability to say more is limited because our study does not explicitly test responsiveness to White and Latine LGB couples and because there are so few studies that examine responsiveness to LGB people. Future research should further examine responsiveness by directly comparing responsiveness to White and Latine LGB people and how race, gender, and sexuality intersect to shape representation.
When comparing our findings to other audits, there is mixed evidence that Latine same-sex couples are responded to less than Latine people in other studies. In Einstein and Glick (Reference Einstein and Glick2017), the lowest response rate to putatively Latine constituents was 53%, which is 9 percentage points greater than the overall response rate in our study. On the other hand, Kalla, Rosenbluth, and Teele (Reference Kalla, Rosenbluth and Teele2018) and Landgrave and Weller (Reference Landgrave and Weller2020) had lower rates of responsiveness. We expect that these differences across studies are, in part, due to differences in the nature of the inquiries in audit studies. Scholars have used a wide range of treatments, which makes it difficult to assess the degree to which underlying variation in responsiveness is a product of the request being made rather than its sender. Likewise, as Landgrave and Weller (Reference Landgrave and Weller2022) argue, names can signal multiple traits that are not explicitly mentioned. Therefore, differential responsiveness across these studies might also be from representatives making assumptions about unobserved characteristics of the people who make constituency service requests. These are general limitations of audit studies. Nonetheless, our response rate was also substantially lower than other studies of local-level officials in the United States, which suggests that difference in constituency request likely does not entirely explain differences across studies. The response rate in our study would rank near the 21st percentile of the distribution in Costa (Reference Costa2017) meta-analysis (7th out of 33 studies in the US). Finally, although studies since Costa’s analysis might affect this “ranking,” we also do not expect the distribution to drastically change. Moreover, we have discussed many articles published since then in this manuscript. While we cannot draw strong conclusions, the overall evidence allows us to speculate that the intersection of sexuality, gender, and race shaped how public officials responded to our request for an LGBTQ pride proclamation, including the low rate of responsiveness compared to studies making private information requests. To further assess these claims, additional research should use intersectional frameworks to examine officials’ responsiveness to constituents.
Conclusion
Our study conducted an experimental test of mayoral responsiveness to Latine same-sex couples making a public-facing constituency request for an LGBTQ pride proclamation in their city. We found that mayors are more responsive to Latine lesbian couples than gay couples. In addition, LGBTQ mayors and women mayors are more responsive to Latine same-sex couples, while Latine mayors were not. We also found that responsiveness increased in cities where LGBTQ people are protected from discrimination. Partisan electoral contexts and LGBTQ population were less influential on responsiveness. Our findings have implications for research about the responsiveness of public officials to marginalized groups.
First, our study highlights how intersectional frameworks can help political scientists explain how marginalized groups interact with the state. By recognizing the ways in which race, gender, and sexuality are mutually constitutive, our study demonstrates that inequalities are not additive. If that were the case, we would have found that Latine lesbians were responded to less than Latine gay men. Likewise, responsiveness did not increase as the Latine population increased or when a mayor is Latine. The seemingly contradictory nature of these findings are only contradictory because scholars tend to theorize about inequality along a single axis. On the other hand, our results are aligned intersectional frameworks that show how intersecting forms of marginalization affect representation (Cohen Reference Cohen1999; Strolovitch Reference Strolovitch2007). Our findings also diverge from a recent audit study that used an intersectional approach (Bauer and Cargile Reference Bauer and Cargile2023), although the constituent name conveyed limited information about them. The conflicting findings across these studies suggest the need for additional audit studies that use intersectional frameworks.
Our results suggest that the study of LGBTQ politics should explicitly theorize about gender differences. A growing body of research shows that lesbians and gay men experience marginalization in distinct ways (Bernhard and Sala, Reference Bernhard and Salan.d.). Furthermore, we note that these gender dynamics in LGB politics appear to contradict dynamics of antitransgender politics. Whereas lesbians have some advantages relative to gay men, transgender women are attacked more than transgender men. For example, opponents of allowing transgender people to participate in sports aligned with their gender identity have targeted transgender women, while there is limited discourse about transgender men. These are part of a broader pattern of constructing transgender women as a threat to cisgender women. Furthermore, they suggest that gender dynamics must be centered in research about LGBTQ politics.
One limitation of our study is that we did not examine responsiveness among Black representatives. Future research should conduct a parallel study to test whether Black representatives are responsive to Black LGBTQ constituents. Previous research suggests Black representatives would be less responsive (Cohen Reference Cohen1999), as do the results in our study. However, an updated study may yield new insights about the representation of Black LGBTQ people. An alternative research design could examine interminority politics, testing whether Black representatives are responsive to LGBTQ Latines or whether Latine representatives are responsive to LGBTQ Black people. In other words, there are many ways in which future scholarship can build on our intersectional analysis of responsiveness to LGBTQ Latines.
Second, our research demonstrates that local institutions shape the behavior of elected officials. While some have questioned whether nondiscrimination ordinances are effective, we find that mayors from cities where LGB people are protected from discrimination were more responsive. Although we noted some concerns about endogeneity, we argue that this evidence suggests that policies create institutional contexts that extend beyond formal protections written into law. Additional research should be conducted to determine whether and how nondiscrimination laws have “spillover” effects to other forms of representation. We recommend expanding these studies to examine responsiveness to transgender constituents, who face severe discrimination and are under attack from right-wing conservatives. There are fewer nondiscrimination ordinances protecting transgender people from discrimination and it is unclear if they would constrain the behavior of public officials.
This study broke new ground by requesting information about constituency services that are public-facing and recognize marginalized groups. In doing so, our study is different from previous audit studies that request private information. Future audit studies should further explore the ways in which public-facing constituency requests are different from private requests. In our view, they are different because public-facing acts bring visibility and recognition to marginalized groups. This claim could be further tested by designing a study that makes different types of public-facing requests. Some of those requests could call for public recognition of marginalized people and some requests could call for public recognition of an advantaged group or a nonpolitical cause. This design would reveal whether public officials were less responsive to the request recognizing the marginalized group. Finally, we conclude by situating this future research in the current political climate. Conservatives are using local politics to censor the visibility of LGBTQ people, Black people, and other marginalized groups. Political science must be able to explain how public officials represent constituents whose very existence and visibility are political. Thus, research must examine the public-private dynamics of discrimination.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/rep.2024.21
Competing interests
The author(s) declare none.
Open access
