Introduction
The invocation of rights and the mobilization of the law is often seen as fundamental for successful civil rights movements (Schneider Reference Schneider1986; Crenshaw Reference Crenshaw, Gotanda, Peller and Thomas1995; McCann Reference McCann1994). Often in response to pressure from activists and shifts in public opinion, lawmakers have the ability to recognize new rights and extend existing protections. Critical legal scholars in contrast shed doubt on this position, arguing that law tends to legitimatecontemporary inequalities in power and rights (Tushnet Reference Tushnet1984). The literature on employment discrimination often sees the impact of law as largely symbolic, forcing firms to create policies that signify legal compliance without actually changing their internal practices (Edelman Reference Edelman1992).
Although judges and legislators are typically at the heart of these debates, scholars have recognized the capacity of legal intermediaries to influence the law in favor of a particular group (Talesh and Pelisse Reference Talesh and Pelisse2019). Intermediaries are actors situated between the rule-makers and rule-takers – such as human resources managers or in-house counsel – who often have the role of interpreting or implementing the law (Edelman Reference Edelman1992; Pélisse Reference Pélisse2019; Talesh and Pelisse Reference Talesh and Pelisse2019). Their interpretations of the law might become regular practice and, in some cases, could even be adopted by the courts (Edelman et al. Reference Edelman, Uggen and Erlanger1999). Particularly in areas with legal ambiguity, then, intermediaries have the potential to either “facilitate or inhibit social change,” playing a significant (but often overlooked) role in the law’s transformative potential (Talesh and Pelisse Reference Talesh and Pelisse2019: 112). In this paper, we document a case where a legal intermediary – the U.S. Equal Employment Opportunity Commission (EEOC) – fostered an expansion of civil rights law, which was only later confirmed by the Supreme Court.
Scholars have begun to explore how institutional values and priorities compete to shape intermediaries’ interpretations of, and compliance with existing law and the implications this might have for social change (Edelman Reference Edelman2016; Pélisse Reference Pélisse2019; Talesh and Pelisse Reference Talesh and Pelisse2019). With respect to employment discrimination law, this research has typically examined how actors such as human resources employees, in-house counsel, or insurance companies shape the meaning of the law, offering interpretations favorable to their employers or clients that then shape largely symbolic compliance (Edelman et al. Reference Edelman, Uggen and Erlanger1999; Talesh Reference Talesh2015; Talesh and Pelisse Reference Talesh and Pelisse2019). This research has paid more attention to how non-legal professionals interpret existing law and less attention to a legal intermediary’s potential to recognize, foster and circumscribe a new civil right. In this article, we examine how a legal intermediary expanded the boundaries of the law, asserting the availability and scope of a new legal right.
We take up this question through an analysis of the EEOC extension of sex discrimination law to sexual orientation and gender identity (SOGI) discrimination. As for earlier civil rights, scholars, activists, politicians and others turned to mobilization of the law as the primary means to secure protections for LGBTQ individuals (Andersen Reference Andersen2005; Boutcher Reference Boutcher2011; Boutcher and McCammon Reference Boutcher, McCammon, Snow, Soule, Kriesi and McCammon2018). These efforts met with some success. As early as 1982, some states began to pass laws prohibiting discrimination based on sexual orientation and, after some delay, gender identity (Badgett Reference Badgett2001). Protection at the federal level, however, was much slower, with attempts to pass a federal non-discrimination law repeatedly failing. It was not until 2020 when the Supreme Court in Bostock v. Clayton County (2020) ruled that SOGI discrimination were federally prohibited under Title VII of the Civil Rights Act.
Eight years earlier, however, the EEOC was already accepting charges of SOGI discrimination. In 2012, the EEOC concluded that discrimination against gender minority employees is rooted in gender stereotypes and other sex-based considerations and, accordingly, was prohibited sex discrimination; a few years later, the agency reached the same conclusion about sexual minorities (Baldwin v. Foxx 2015; Macy v. Holder 2012). Thus, although employees, employers and judges are often the focal points in understanding discrimination law outcomes (e.g. Berrey et al. Reference Berrey, Nelson and Beth Nielsen2017; Best et al. Reference Best, Edelman, Krieger and Eliason2011; Hirsh and Kornrich Reference Hirsh and Kornrich2008; Hirsh and Lyons Reference Hirsh and Lyons2010; Roscigno Reference Roscigno2019; Tomaskovic-Devey and McCann Reference Tomaskovic-Devey and McCann2021), federal law surrounding SOGI employment discrimination has been notably shaped by an administrative agency’s interpretation of the law.
In this article, we examine the capacity and limits of the EEOC to act as an agent of social change through the expansion and restriction of SOGI law during its processing of discrimination charges. We consider the role of the EEOC and its actors as legal intermediaries, shaping the definition and implementation of EEO law in an area of legal ambiguity (Edelman Reference Edelman1992; Edelman and Stryker Reference Edelman and Stryker2005; Talesh and Pelisse Reference Talesh and Pelisse2019). Without clear guidance from the Court, our findings support that the EEOC drew on related Title VII caselaw on sex discrimination, political currents, pragmatism in processing charges and EEOC actors’ understandings of the agency’s purpose in order to create a legal interpretation of SOGI discrimination. In this respect, the agency displayed a surprising power to engage in lawmaking, rather than rule following, in its handling of SOGI charges. At the same time, the EEOC emphasized cases consistent with earlier legal practice, limiting the scope of the new law it had itself promulgated.
To understand the degree to which the EEOC pushed for expansion of sex discrimination law, and what values and priorities might have been guiding its decisions, we focus on the agency’s categorization of SOGI charges as more or less likely to demonstrate that discrimination has occurred. On intake, the EEOC evaluates each charge and assigns a processing category in accordance with its Priority Charge Handling Procedures. An “A” signifies that the charge contains evidence that indicates reasonable cause of discrimination and/or that it involves one of the agency’s strategic priorities, a “B” indicates that more information is needed to assess the merits of the charge and further investigation is warranted; and a “C” indicates the charge lacks merit, is untimely, and/or the EEOC does not have jurisdiction. EEOC priorities include “addressing emerging and developing issues,” which – in its Strategic Enforcement Plan for 2013–2016 – encompassed “coverage of lesbian, gay, bisexual and transgender individuals under Title VII’s sex discrimination provisions, as they may apply” (EEOC 2012). An analysis of patterns in the EEOC’s processing categories for SOGI charges, therefore, provides insight into what types of discriminatory conduct the agency viewed as likely to be successful – including areas the agency might see as appropriate extensions of sex discrimination law to gender identity and sexual orientation claims. Accordingly, we ask: What alleged issues contained in SOGI charges are more likely to have received a categorization of “A,” indicating that the agency viewed the charge as likely to demonstrate discrimination under Title VII? Our analysis of processing categorization of charges provides insight into how the EEOC, as a legal intermediary, made decisions that ultimately pushed the boundaries of existing law.
We draw on two datasets created from confidential 2012–2016 EEOC charge data for SOGI discrimination. We first follow the standard approach in employment discrimination research of using the quantitative Integrated Management System (IMS) dataset created by EEOC administrators to understand the relationship between alleged issues and processing categorization. We then use a unique dataset created from a content analysis of the charge narratives themselves to understand the detailed issues alleged in SOGI discrimination charges. Our coding of the charge narratives allows for a more in-depth understanding of how the EEOC evaluated specific discriminatory experiences that are not captured in the quantitative administrative dataset. In doing so, our research provides insight into how legal intermediaries make decisions that act to promote social change.
Theoretical and empirical literature
The ability of the law to effectively advance rights is contested. Critical legal scholars express skepticism that a legal system designed to promote the interests of those in power can be harnessed to empower disadvantaged groups. They argue that working within the legal system limits the ability to imagine new solutions to inequalities and simply reinforces the established order (Tushnet Reference Tushnet1984). Compounding this effect, countless studies have found that the litigation process provides advantages to “repeat players,” such as corporations and others with greater organizational and economic resources (Albiston Reference Albiston1999; Galanter Reference Galanter1974; Nielsen Reference Nielsen2000). These advantages, over time, serve to shape legal precedent in their favor, producing yet another obstacle to legal victory for those with fewer resources. On the other hand, Crenshaw (Reference Crenshaw, Gotanda, Peller and Thomas1995) points to civil rights achievements won through the courts as evidence that calling upon the language of equality and rights can be a powerful tool for change precisely because it reinforces the dominant ideology, thereby legitimating rights claims (see also Schneider Reference Schneider1986; McCann 1998).
Social movements can also produce shifts in access to rights, though never truly independent from the assistance of other actors. Law and movement scholars often point to the National Association for the Advancement of Colored People’s (NAACP) legal campaign around desegregation that culminated in Brown v. Board of Education (Greenberg Reference Greenberg1994; Meyer and Boutcher Reference Meyer and Boutcher2007; Tushnet Reference Tushnet1994) as the primary textbook example of successful legal mobilization, but social movements dating back to the 18th and 19th centuries relied on litigation as part of their strategies (Boutcher and McCammon Reference Boutcher, McCammon, Snow, Soule, Kriesi and McCammon2018; O’Connor Reference O’Connor1980; Wiecek Reference Wiecek1977). Following Brown v. Board of Education, however, a wide range of social movements increasingly relied on legal mobilization as a key strategy to advance their causes (Chen and Cummings Reference Chen and Cummings2013; Meyer and Boutcher Reference Meyer and Boutcher2007), including the LGBT movement in the case of marriage equality. Ultimately, courts can be at the forefront of change by recognizing claims to rights, or they can follow shifts in public opinion produced through social movements (Banaszak and Ondercin Reference Banaszak and Ondercin2016).
Courts are not the only institution, however, through which disputes are funneled. As Albiston et al. (Reference Albiston, Edelman and Milligan2014) note, the dispute process can be imagined as a tree, with varying branches offering dispute paths that might or might not be fruitful. For employment discrimination disputes, one of these branches involves the processing of formal charges through the EEOC, a requirement of the 1964 Civil Rights Act. The EEOC might funnel these charges toward several resolutions, including mediation, litigation brought by the EEOC, or provision of a “right to sue” letter to the charging party. The latter could result in a charging party choosing to “lump it” and not pursue their claim any further, or their following a branch toward litigation within the courts (Felstiner et al. Reference Felstiner, Abel and Sarat1980–81; Michelson Reference Michelson2007; Sandefur Reference Sandefur, Pleasence, Buck and Balmer2007). The EEOC’s determination regarding a charge’s merit serves to direct charges down varying dispute paths, legitimating some claims and extinguishing others.
This function of the EEOC is particularly important in situations involving legal ambiguity, where the law is unsettled or emergent. In these cases, the EEOC has the potential to act as an agent of social change by constraining or expanding the law based on its assessment of the case’s strength. For SOGI discrimination charges prior to 2020, the EEOC had the option to categorize all charges as lacking merit due to the absence of federal protection for sexual orientation or gender identity, but instead processed these charges under the umbrella of sex discrimination. This decision not only served to broadly expand legal rights but also resulted in the agency playing a fundamental role in evaluating the types of SOGI discrimination that would be protected under the law.
In making determinations about the strength and viability of SOGI claims, the EEOC was influenced by external guidelines in the form of legal precedent surrounding sex and gender discrimination (e.g. Price Waterhouse v. Hopkins 1989). But institutions are not solely governed by a set of external rules, as exemplified by organizational theory on inhabited institutions (Binder Reference Binder2007). Internal actors interpret and shape the meaning of the institution in ways that might reinforce official policies or rules (internal and external) or might be in the spirit of their understandings of the institution’s purpose (Binder Reference Binder2007; Hallett and Ventresca Reference Hallett and Ventresca2006). Through interpretation and application of institutional logics, therefore, actors within organizations such as the EEOC are empowered to actively shape rules and processes, including potentially expanding legal rights.
Research on the role of legal intermediaries provides insight into how actors such as EEOC commissioners, case managers and mediators could contribute toward social change through their interpretation and application of legal principles. Given that laws often lack explicit detail regarding the criteria of legal violations (Edelman Reference Edelman1992), those who are the subject of such laws or who are charged with implementing them must develop their own standards of compliance.
Research on legal intermediaries has recognized that intermediaries can be legal or non-legal professionals but share the common characteristic of being situated between the rule-makers and the rule-takers (Pélisse Reference Pélisse2019). Lawyers, managers, various bureaucrats and even judges have been analyzed as legal intermediaries, interpreting and implementing the laws at differing levels (Pélisse Reference Pélisse2019). Legal intermediaries might draw upon professional experience, institutional logics and their understanding of law and compliance in order to develop these standards (Talesh Reference Talesh2015; Talesh and Pelisse Reference Talesh and Pelisse2019). Their interpretations have the potential to become common practice and, in turn, to influence the law if they are viewed by the courts as rational (Talesh and Pelisse Reference Talesh and Pelisse2019). In this respect, legal intermediaries can play a larger role than lawmakers in shaping legal outcomes (Edelman and Stryker Reference Edelman and Stryker2005; Talesh and Pelisse Reference Talesh and Pelisse2019). For example, Edelman et al. (Reference Edelman, Uggen and Erlanger1999) found that many of the ways that organizations developed employee due process grievance procedures were soon understood to have legal origins and, in turn, had influence on judicial interpretations of compliance. Notably, however, much of the research on employment discrimination has focused on the role of intermediaries in fostering symbolic compliance on the part of organizations through grievance procedures, EEO training, or the creation of diversity and EEO officers (see, e.g., Dobbin Reference Dobbin2009; Edelman Reference Edelman2016; Edelman et al. Reference Edelman, Petterson, Chambliss and Erlanger1991; Reference Edelman, Uggen and Erlanger1999; Talesh and Pelisse Reference Talesh and Pelisse2019).
In the following section, we more closely examine the history of the EEOC, including the latitude provided to the EEOC in interpretation and application of the law and the deference afforded to EEOC decisions by the U.S. Supreme Court.
The EEOC as legal intermediary
The EEOC was founded in 1964 as the primary federal agency tasked with monitoring enforcement of Title VII of the newly enacted Civil Rights Act. Specifically, Title VII prohibits employment discrimination based on race, color, religion, sex or national origin, with the main goal of desegregating America’s workplaces. Despite being tasked with enforcing the full scope of the law, early leaders made it clear that they saw their role as mainly focusing on race (Stainback and Tomaskovic-Devey Reference Stainback and Tomaskovic-Devey2012), in large part due to heightened mobilization around the civil rights movement (Pedriana and Stryker Reference Pedriana and Stryker2004). As the women’s movement gained more momentum in the early 1970s and with increased national attention on gender discrimination, the EEOC began to expand its focus to include sex discrimination (Pedriana Reference Pedriana2004).
Although officially a nonpartisan agency, the EEOC’s history has always been heavily influenced by politics. For instance, the five commissioners that sit atop of the EEOC are appointed by Presidents – no more than three can be from the same party – and have historically followed the political leanings of presidential administrations (Stainback and Tomaskovic-Devey Reference Stainback and Tomaskovic-Devey2012). In this regard, the changing political context can shape the agenda and policy priorities of the agency. In addition, the EEOC has been heavily influenced by external social movements that have mobilized to pressure the EEOC into action throughout its history, initially through the civil rights movement and then the women’s rights movement, and more recently the LGBT movement.
In its early days, the EEOC was marked by uncertainty about how best to accomplish its mandate. Equal employment law was novel and unsettled due to the lack of regulatory guidance and the absence of court rulings shaping the newly enacted law. Indeed, the early period of the agency has been characterized as “toothless” due to its lack of resources, the absence of a clearly defined mission and its unorganized and chaotic structure (Pedriana and Stryker Reference Pedriana and Stryker2004). However, despite its “weak” origins, this period of unsettled law meant that the EEOC had significant leeway to define its mandate. Although the agency was largely constrained to “collecting data and issuing symbolic statements” (Stainback and Tomaskovic-Devey Reference Stainback and Tomaskovic-Devey2012: 87), the early EEOC was also actively shaping itself into a powerful agency through statutory construction (Pedriana and Stryker Reference Pedriana and Stryker2004). Early “activists” within the agency used this period of uncertainty to creatively interpret the EEOC’s mandate as broadly as possible (Blumrosen Reference Blumrosen1970). Further, the agency’s early statements about what defined discrimination were “essentially seen as law in the eyes of both firms and eventually the courts” (Stainback and Tomaskovic-Devey Reference Stainback and Tomaskovic-Devey2012: 90).
With the passage of the Equal Employment Opportunity Act of 1972, the EEOC gained considerable regulatory power. Most importantly, the agency was granted the power to engage in its own legal mobilization and bring lawsuits against employers (Stainback and Tomaskovic-Devey Reference Stainback and Tomaskovic-Devey2012). One landmark example is Griggs v. Duke Power Co. (1970)where the Supreme Court ruled that employment practices that had disparate impact between groups of workers can be considered discriminatory even if it was not intentional. Following Griggs, the 1970s saw a string of Supreme Court cases that defined the boundaries of equal opportunity law, and more importantly, strengthened the regulatory authority of the EEOC. These cases played a pivotal role in shaping legal precedent and influenced how courts would interpret antidiscrimination law.
Historical accounts suggest that the peak of the EEOCs regulatory power began to recede in the 1980s following the election of Reagan and his stance against civil rights. Reagan appointed Clarence Thomas to head the agency and budget cuts and staffing levels soon followed (Stainback and Tomaskovic-Devey Reference Stainback and Tomaskovic-Devey2012). Despite increasing numbers of charges being filed, the number of EEOC-initiated lawsuits declined by more than 50% (Burbridge Reference Burbridge1986). Moreover, the number of class-action cases – arguably one of the stronger mechanisms for anti-discrimination enforcement by the EEOC – fell from 1106 in 1975 to just 51 in 1989 (Stainback and Tomaskovic-Devey Reference Stainback and Tomaskovic-Devey2012, citing Donohue and Siegelman Reference Donohue and Siegelman1991). The shift in the EEOC’s regulatory power coincided with a conservative shift on the Supreme Court along with changing practices in corporate governance that rebranded human resource departments away from enforcing equal opportunity to a logic of diversity management (Berrey Reference Berrey2015; Dobbin Reference Dobbin2009).
Thus, the EEOC and its actors have played an important, though mixed, role in the development of employment discrimination law. Following the creation of the Civil Rights Act of 1964 and the ambiguity around how best to achieve employment desegregation, the EEOC became the key actor in defining compliance to the newly enacted law through a broad interpretation of its mandate. After gaining more enforcement power, notably through the coercive use of litigation, courts began to defer to the EEOC and expand the boundaries of what counted as discrimination. Although its regulatory power declined considerably amid the conservative backlash to civil rights, the EEOC remains an important legal intermediary with the potential to shape EEO law.
The EEOC’s role in the development of EEO law took a new turn in 2012 when, in Macy v. Dept. of Justice, the agency issued a decision that gender identity discrimination was sex discrimination because it is based on gender non-conforming behavior that is tied to sex (Macy v. Dept. of Justice 2012). In Macy, the complainant, a transwoman, alleged that she had not been hired following a background check uncovering her change in name and gender. The Commission concluded that discrimination based on gender identity is prohibited under Title VII, citing the Supreme Court’s finding in Price Waterhouse v. Hopkins (1989) that discrimination based on gender stereotypes is sex discrimination. Subsequent to this initial determination, the EEOC issued decisions that denial of restroom access, refusing to allow an employee to dress in accordance with their gender identity, refusal to change a complainant’s name following a gender transition, and repeated and intentional use of pronouns inconsistent with one’s gender identity are prohibited discriminatory conduct (Hillier v. Dept. of the Treasury 2016; Lusardi v. Dep’t of the Army 2015; Eric S. v. Department of Veteran Affairs 2014).
The EEOC’s decision that sexual orientation discrimination was also prohibited was less clearly connected to prior precedent. In 2015, in Baldwin v. Dept. of Transportation, the EEOC determined that sexual orientation is sex discrimination on two grounds. First, when an employer treats an individual differently because of their sexual orientation then that employer is necessarily taking that person’s sex into account. If the individual, for example, was terminated based on a marriage, and different-sex couples were not terminated based on marriage, then the employer’s decision was based on the employee’s sex (and, by association, the sex of their partner). Second, as with discrimination based on gender identity, discrimination based on sexual orientation is often tied to stereotypes about sex and gender norms for men and women. It is the variation from expected norms about how men and women should express their sex and gender that results in discrimination. This gender-based argument is, once again, tied to the Court’s decision in Price Waterhouse v. Hopkins (1989).
After the EEOC began accepting SOGI charges, federal courts took varying positions on whether these constitute sex discrimination under Title VII. A district court specifically rejected the EEOC’s expansion of Title VII, explaining that the EEOC’s decisions were not binding on the court with respect to sexual orientation discrimination (Brown v. Subway Sandwich Shop of Laurel, Inc. 2016). Split decisions in the circuit courts pushed the Supreme Court to take up this question for both gender identity (R.G. and G.R. Harris Funeral Homes v. Stephens) and sexual orientation (Bostock v. Georgia; Zarda v. Altitude Express, Inc.).
In 2020, the Court held in Bostock v. Clayton County that both SOGI discrimination are sex discrimination given that individuals would not have experienced the discrimination but for their sex. In the majority opinion, the Court described other instances in which the EEOC’s position on the expanse of Title VII had changed, or where the EEOC had noted that Title VII has been expanded to cover areas unanticipated at the time of its drafting (e.g. sex-segregated job listings, discrimination against male employees). The Court did not, however, cite to the EEOC’s decisions or discussion regarding sexual orientation or gender identity within its opinion, suggesting that the Court’s decision was not overtly in deference to the EEOC’s determinations. Nonetheless, the evolution of federal SOGI discrimination law suggests that the EEOC played an important role in expanding legal protection through its initial decisions to accept SOGI charges. Its subsequent determinations regarding the specific conduct that constitutes a strong SOGI discrimination case provide another avenue by which the agency had the potential to further develop the law.
Our review of the literature establishes that the law is viewed by some to be a useful mechanism for facilitating social change, but that it often has more symbolic effects as its interpretation and application are often harnessed to maintain existing inequalities in power and rights. Although legal intermediaries have the capacity to produce positive social change (Talesh Reference Talesh2015), much of the literature in the area of employment discrimination has focused on legal intermediaries facilitating only symbolic compliance with the law (see, e.g., Dobbin Reference Dobbin2009; Edelman Reference Edelman2016; Edelman et al. Reference Edelman, Petterson, Chambliss and Erlanger1991; Reference Edelman, Uggen and Erlanger1999; Talesh and Pelisse Reference Talesh and Pelisse2019). In these contexts, legal intermediaries draw upon managerial logics to demonstrate legitimacy through superficial adherence with the law, while limiting legal constraints on their organization (Talesh and Pelisse Reference Talesh and Pelisse2019). The focus, then, is ultimately on resisting compliance with the law. In contrast, we examine the EEOC’s handling of SOGI discrimination charges to better understand the role that this legal intermediary has taken to expand the scope of existing law rather than facilitate noncompliance with the spirit of the law. Through analyzing charge processing categorization, we offer insight into the ways that the EEOC recognized new rights, and both expanded and offered more conservative interpretations of existing EEO law.
Data and methods
Charge data
Our analysis draws on confidential data from the U.S. EEOC. All coauthors applied and were granted access through the data access program in place at the time of application. Through this program, researchers were detailed to the EEOC through the U.S. Intergovernmental Personnel Act, which provided them with access to confidential charge data for the purposes of conducting research on SOGI discrimination. In 2017, we were granted access to 5 years (2012–2016) of SOGI discrimination data which included a total of 9262 charges from the EEOC and state Fair Employment Practices Agencies (FEPAs). In the years that followed our initial application, the EEOC underwent changes in its data access protocol and, due to processes being in flux, we were unable to obtain additional years of data.
Out of the 9262 charges we received, 5711 included EEOC processing categories and, accordingly, are included in the regression analyses of this study. Our comparison of charges with EEOC processing categories and those without such charges indicates that the key difference is whether the charge was filed initially with an EEOC or a state FEPA. Of the 3208 FEPA charges in the dataset, only 463 had an EEOC processing code, whereas 5248 out of the 5279 EEOC charges had a processing code. No notable differences emerged between those FEPA charges with an EEOC processing category compared to those without, e.g. there were no differences between states with state-level SOGI antidiscrimination laws and those without, or differences in distribution of charges across gender identity and sexual orientation charges.
When an individual alleges discrimination, they either file a charge online, mail a charge to the EEOC (or state FEPA if that state has a sharing agreement with the EEOC), or physically go into an office to file a charge. These charges are entered into a computer database – the IMS – which becomes a centralized repository of all charges tracked by the EEOC and FEPAs. These charge data provided to us by the EEOC include the employer’s name, address, industry and establishment size; the charging party’s age, race, national origin and sex; the basis for the charge (that is, the protected class, such as sex, sexual orientation, gender identity, race, national origin); the issue charged (e.g. discrimination in promotion, harassment, discharge, etc.); the processing of the charge (e.g. whether it goes to mediation and/or is investigated) and disposition of the charge.
In addition to the basic, quantitative information contained in the IMS, most charges include a narrative which describes the charging party’s allegations. At intake, individuals provide a description of the alleged discriminatory acts that is turned into a charge narrative by the intake official and entered into a “Form 5” document. The Form 5 constitutes the formal charge sent to responding employers. These narratives sometimes include accompanying information, such as investigatory notes and employer responses. Taken together, these narratives generate qualitative data describing the actual actions and processes that led to the employee’s perception of discrimination. For those charges with an SOGI allegation, we were provided all of the charge narrative data as a large exported text file from the IMS. In this article, we draw on both the quantitative IMS data and a unique dataset constructed from coding these charge narratives.
Not all eligible charges of sexual orientation or gender identity discrimination in our narrative database, however, included the Form 5 or contained enough information for us to code. To select sample narratives with “codeable” materials, we included narratives that contained strings of text (e.g. “Civil Rights Act” or “Disabilities Act”) that appear in the Form 5 allegations; these narratives typically included descriptions of discriminatory experiences, as opposed to other administrative notes that would not produce adequate data for analysis.
After excluding duplicates and those that did not have enough words to code, we were able to match 1663 charges to the quantitative charge database by charge number. Of those 1663 charges, 1516 contained valid values in the charge database (year and charge type) and EEOC processing codes for analysis.
Coding charge narratives
Our goal in coding the qualitative charge narratives was to capture more detailed information about the various discriminatory experiences alleged by employees than can be captured from the quantitative data alone. These codes could then be used to gain greater insight into the alleged issues that the EEOC identified as more likely to indicate discrimination. Notably, due to confidentiality restrictions, we are unable to quote narratives or describe potentially identifying scenarios about discrimination that are contained in the narratives; rather, we can only identify patterns that appear in the stories.
We created a codebook from an initial reading of a subsample of narratives. The coauthors regularly met to refine several iterations of testing the codebook. The generation of the section of the codebook on alleged issues was informed by empirical work on SOGI employment discrimination. Research on LGBT employment discrimination has found that discrimination is common, with 20–40% of LGBT individuals reporting these experiences (Sears et al. Reference Sears, Mallory, Flores and Conron2021; Pew Research Center 2013; NPR Robert Wood Johnson Foundation and Harvard School of Public Health Reference Harvard2017). Transgender individuals report higher rates of discrimination than LGB individuals, with 30% in one large community survey and 49% in the Employment Experiences Survey indicating that they had experienced discrimination (James et al. Reference James, Herman, Rankin, Keisling, Mottet and Anafi2016; Sears et al. Reference Sears, Mallory, Flores and Conron2021). These differences are particularly stark when it comes to hiring decisions. Audit studies found that transgender individuals are less likely to be hired than cisgender individuals, with one matched pair applicant study finding a 42% rate of discrimination (Make the Road New York 2010; Winter et al. Reference Winter2016). Similarly, survey data indicate that over twice as many transgender individuals report not being hired (44%) because of their identity, as compared to LGB individuals (22%) (Sears et al. Reference Sears, Mallory, Flores and Conron2021).
In addition to adverse hiring decisions, prior research has documented the pervasiveness of experiences such as antigay language and jokes, verbal harassment, physical harassment, pressure to conceal one’s LGBT status, denial of spousal benefits, being ignored or shunned, refusal to use proper pronouns or being fired (Connell Reference Connell2015; Schilt Reference Schilt2006; Sears et al. Reference Sears, Mallory, Flores and Conron2021; Woods Reference Woods1993). Harassment was the most common form of discrimination reported by LGBT employees in one survey, with 38% reporting experiencing some form of harassment; verbal harassment was the most common (31%), followed by sexual harassment (26%) and physical harassment (21%) (Sears et al. Reference Sears, Mallory, Flores and Conron2021). Notably, transgender employees reported greater harassment and discrimination across hiring, terminations and verbal and physical harassment than did sexual minorities but similar levels of sexual harassment (Sears et al. Reference Sears, Mallory, Flores and Conron2021). In addition, there are particular workplace challenges related to gender transition that are distinct from challenges for lesbian, gay and bisexual people. New names, new pronouns, dress codes, and restrooms and locker rooms generate vulnerabilities to discrimination for many transgender people (Brewster et al. Reference Brewster, Velez, Mennicke and Tebbe2014; Budge et al. Reference Budge, Tebbe and Howard2010; Dietert and Dentice Reference Dietert and Dentice2009).
This literature suggests that we will find that charges contain descriptions of discrimination and harassment based on gender nonconformity, sexual harassment often directed toward sexual minorities, and partner/spouse/marriage issues – important factors shaping the discriminatory experiences of employees that cannot be captured in the IMS dataset. The charge data include more specifics regarding actual discriminatory experiences of employees. For instance, if an employee was harassed on the job, the IMS charge dataset would only indicate that the employee alleged harassment but would have no details about the type of harassment. Our coding of the charge narratives captures more context about the alleged discriminatory experiences; this detail allows us to better understand how EEOC actors were making decisions about what constituted prohibited conduct by employers. While our coding began with a set of deductive codes from prior case law and reports, we also inductively identified a series of codes on discrimination experiences from the narratives themselves.
We trained four undergraduate coders on the codebook and coding process. To ensure the greatest level of consistency across coders, we met regularly to test the codebook and discuss any inconsistencies or questions that arose. Upon completion of the training, each coder received a subset of narratives to code. After completion of their initial set, they were asked to review and add any missing codes from another coder’s set. We asked that the coders not remove any discrepant codes but document their suggested changes. The Principal Investigatorsthen reviewed those notes and made changes, where necessary.
Variables
The dependent variable is the EEOC processing category for a charge. The EEOC assigns to each charge a processing category based on their initial assessment of the likelihood that discrimination occurred (McCann et al. Reference McCann, Tomaskovic-Devey and Lee Badgett2018). Processing categories affect recommendations for mediation and the depth of agency investigations. The categories also reflect stated EEOC strategic priorities. The EEOC assigns a processing category “A” if reasonable cause of discrimination is likely to be found. Charges are classified as “B” if the charge may have merit but more information is necessary. Charges are classified as “C” if “cause is unlikely” for a variety of reasons, ranging from lack of credibility of the allegation to being outside the EEOC’s jurisdiction. However, employees with charges categorized as such may still choose to file a lawsuit against their employer if they wish. We captured charge classification as a dichotomous variable, with charges assigned “A” coded as 1 and charges coded as “B” or “C” coded as 0. Thus, we are predicting the charges that the EEOC identified as likely to show reasonable cause of discrimination. It is important to note that the EEOC’s reasoning for categorizing these charges as “A” could be varied, in that it could reflect specific EEOC priorities and/or attempts to broaden the scope of Title VII coverage, it could reflect how likely the agency thinks the claim is to succeed in being classified as sex discrimination if it proceeds to litigation (regardless of the EEOC’s own agenda), or it could reflect pragmatic concerns in terms of being a more easily established charge.
Focusing on the specific charge allegations, we include a dummy variable to indicate whether the claimant is alleging sexual orientation (coded 1) or gender identity (coded 0) discrimination (i.e. “Sexual orientation charge”) to assess whether the EEOC was more likely to assign an A to gender identity charges, in accordance with the prior caselaw on gender stereotypes (Price Waterhouse v. Cooper). We then include the issues of discrimination alleged in the charge narrative, which are the range of discriminatory behavior alleged against the employer, to determine the type of adverse employment actions and discriminatory conduct that the EEOC was prioritizing for SOGI charges. In our first set of analyses that use only the IMS quantitative data (presented in Table 2), we include the following issue codes from the dataset (0 indicating not present and 1 indicating present): benefits, constructive discharge, demotion, discharge, discipline, harassment, hiring, intimidation, layoffs, promotion, sexual harassment, suspension, terms and conditions and wages.
In our second set of analyses (presented in Table 3), we focus on the more fine-tuned coding from the charge narrative data to assess the EEOC’s processing decisions for items that are more relevant for SOGI claims and are not captured in the IMS charge data. These issues are coded as a set of dummy variables, with 1 indicating that the issue was mentioned in the charge narrative. For each of these issues, we include our fine-grained codes and an “other” code; the “other” code is coded as 1 if the charge included the issue in the IMS dataset, but the narrative did not include a description that fell into one of our fine-grained codes. For example, if the charge was coded as “harassment” in the IMS but did not contain descriptions of one of the specific harassment actions that we identified in our codebook for the content analysis, the charge is coded as “Harassment, other.”
We included fine-grained codes for the following issues: benefits, hiring, terms and conditions and harassment. For benefits, we included variables indicating whether the charge includes claims relating to benefits for a same-sex partner or for medical benefits related to gender-affirming care, including surgery or medication. For hiring, we included variables indicating the point in the hiring process when an adverse employment action occurred, including failure to interview, failure to hire, failure to rehire, or whether an offer was rescinded. Terms and conditions variables include more typical discriminatory conduct related to scheduling, as well as issues that are potentially more likely to surface in SOGI charges, including dress codes (which are often gender-restrictive), professional development opportunities and restrictions on restroom or locker room access dependent on gender.
For charges that allege harassment, we included variables for a range of particular harassing behaviors, including verbal, nonverbal and sexual harassment. We captured 11 types of harassing behaviors. Verbal harassment included the following variables: individual disparagement such as harassing comments that targeted the charging party specifically; general disparagement focused more on making jokes or expressing stereotypes about LGBT people; disparagement related to a relationship, including a spouse or partner, such as comments about one’s marriage or dating preferences; other verbal harassment focused on gendered issues, including gender disparagement related to the way an individual dressed or behaved; misgendering, such as using the wrong name or pronouns; and verbal threats. Nonverbal harassment variables included property damage, such as defacing an office or locker; acts of physical violence; and physical aggression, such as blocking an individual’s exit. Sexual harassment variables included verbal sexual harassment, such as sexual overtures, jokes or negative comments about sex; or physical sexual harassment, including unwanted kisses, touching of breasts or genitals or other sexual advances. Employees could allege multiple types of harassing behavior within a single charge.
Analytical approach
Within all analyses, we analyze SOGI charges as a whole rather than dividing SOGI charges. We do this because of the smaller number of gender identity charges which, when analyzed on their own, obscure many of the overall patterns in the EEOC’s categorization. Further, our questions are centered around how the EEOC is expanding or limiting federal SOGI discrimination law, which is highlighted by looking at all SOGI charges.
We first provide an overview of the percentage of charges alleging each basis and issue that fall into the EEOC’s A processing category; this includes a breakdown across our fine-grained categories of issues alleged. For each, we run z-tests of proportions to determine whether the percentage in category A for each issue listed is significantly different from the percentage in category A for those not in that issue (approximately 30–34% of those not in each issue were categorized as A, in accordance with the pattern in the overall IMS and charge narrative datasets). Next, we draw upon the IMS quantitative dataset to predict the EEOC’s processing category using the issues contained within the dataset (Table 2). Because our outcome measure is binary, we use logistic regression to analyze which issues increase the odds of the EEOC classifying a charge as likely to show reasonable cause of discrimination (i.e. classified as an A). We then focus on the more SOGI-specific issue codes from the content analysis by using logistic regression to predict the EEOC’s processing categories for the 1516 charges in the charge narrative dataset (Table 3). These analyses provide insight into how the EEOC was limiting or expanding SOGI law through its processing categorization decisions.
Results
In the following section, we describe the issues alleged in SOGI charges that were most likely to receive a processing category of “A,” indicating that they were understood to present stronger evidence of sex discrimination. Overall, our findings demonstrate that gender identity charges and issues focused on gender stereotypes received higher categorization by the EEOC during this time period, while charges focused more on sexual relationships or identity were less prioritized.
The EEOC’s categorization of the alleged issues
By examining the distribution of the EEOC processing categories across allegation characteristics, we can see that there are some elements of allegations that appear to be more likely to be classified as “A” by the EEOC, indicating stronger likelihood of demonstrating cause (Table 1). Overall, approximately 34% of the 5711 SOGI charges in the IMS dataset were classified as an “A” by the EEOC. For comparison, approximately 23% of all sex-based Title VII charges were categorized as an “A” during the same time period and 33% of all sexual harassment charges (McCann et al. Reference McCann, Tomaskovic-Devey and Lee Badgett2018). There was some variation across the 4 years in our dataset, with 33% of charges in 2013 (368), 33% of charges in 2014 (431), 37% of charges in 2015 (621) and 31% in 2016 (499) categorized as “A.” In 2015, the EEOC issued its decision that sexual orientation charges constituted sex discrimination, contributing to the increase in charges in that year (although the agency was already accepting sexual orientation charges several years before this decision). The increase in 2015 could also be partially attributable to the Supreme Court’s decision in Obergefell during that year, legalizing same-sex marriage, which could have made the extension of other legal rights to LGBTQ individuals more salient.
Percentage of charges in processing code A, by charge issue (IMS dataset and charge narratives)
Notes: z-tests of proportions were run to assess statistical significance of the bivariate relationships.
*** p ≤ .01;
** p ≤ .05;
* p ≤ .10.
There are notable differences between gender identity and sexual orientation charges, with 54% of charges categorized as “A” for gender identity in the IMS dataset compared to 30% for sexual orientation. This provides preliminary evidence that the EEOC viewed gender identity charges as priorities or as stronger sex discrimination cases during this period. Nonetheless, the EEOC was expanding Title VII law by coding any sexual orientation charges as category “A” prior to the Supreme Court determining that these were sex discrimination, given that sexual orientation was less clearly tied to the gender stereotyping body of cases than was gender identity.
In Table 1, we also present the key issues alleged in the SOGI charges categorized as an “A.” First, we list the codes of issues alleged that are contained within the IMS quantitative dataset, indicating those with statistically significant bivariate relationships with the processing categorization variable. Benefits (p ≤ .01) allegations clearly stand out, with 60% categorized as an “A.” Wages (p ≤ .10) and sexual harassment (p ≤ .01) follow at approximately 40% categorized as “A,” terms and conditions at 37% (p ≤ .05) and harassment at 36% (p ≤ .01). There appears to be a pattern of issues related to the terms of employment (wages, benefits, terms and conditions) and harassing behaviors (harassment and sexual harassment) that the EEOC was more likely to assign to processing category “A.”
Turning to the more fine-tuned, qualitative coding of issues, more detail emerges regarding the EEOC’s decision-making. The EEOC categorized many of the issues directly related to gender identity as an “A,” such as medical benefits (largely related to gender affirming care) (78%) (p ≤ .01), misgendering (61%) (p ≤ .01) and restroom and locker room access (50%) (p ≤ .01). These distinctions point toward possible EEOC priorities related to gender identity discrimination. This finding is bolstered by the EEOC’s decision in Lusardi v. Dep’t of the Army (2015) that restroom access and misgendering are evidence of prohibited conduct for gender identity discrimination claims.
In addition, the EEOC appears to have drawn distinctions across types of harassment, emphasizing cases that involve physical harassment and verbal harassment involving gender stereotyping. Specifically, processing category “A” was assigned to 48% of property damage issues (p ≤ .10), 44% of physical violence issues (p ≤ .01) and 44% of physical sexual harassment (p ≤ .05) charges. Our findings regarding higher categorization of sexual harassment charges are similar to what has been found for all sex-based Title VII charges, suggesting that sexual harassment was a priority for the EEOC during this period (McCann, Tomaskovic-Devey, and Badgett Reference McCann, Tomaskovic-Devey and Lee Badgett2018).
With respect to verbal harassment, 37% of gender stereotyping charges (p ≤ .01) and 35% of individual disparagement (p ≤ .05) charges were assigned an “A.” And 54% of charges related to dress code (p ≤ .01), which were primarily related to gender non-conforming dress, were categorized as “A.” This mirrors the EEOC’s higher categorization of charges for gender identity issues that are tied to gender stereotyping, reinforcing the idea that the EEOC regularly extended the Price Waterhouse v. Hopkins (1989) finding in evaluating SOGI charges.
Analysis of IMS quantitative data
We first report results of the logistic regression analysis drawing upon the quantitative IMS dataset’s issues to predict EEOC processing categories. These are the EEOC data typically used in employment discrimination analyses. Table 2 presents the odds ratios from the analysis. As anticipated, sexual orientation charges have 75% lower odds than gender identity charges of being categorized as an “A” by the EEOC, holding the issues alleged constant (p ≤ .01).
Odds ratios predicting EEOC processing category, by issue (IMS dataset)
*** Note: p ≤ .01;
** p ≤ .05;
* p ≤ .10.
Turning to the issues contained in the charges, many of the same patterns observed in the bivariate comparisons are present. Charges that allege benefits issues stand out, with 109% higher odds of being categorized as an “A” by the EEOC (p ≤ .05). Charges that include harassment allegations have 70% higher odds (p ≤ .01), and those with sexual harassment issues have 65% higher odds (p ≤ .05), of being categorized as an “A.” Issues related to specific terms of employment also increase the odds of the EEOC categorizing the charge as an “A,” including terms and conditions charges having 31% higher odds (p ≤ .05) and those with wages having 70% higher odds (p ≤ .05). Finally, allegations of suspension and discipline are marginally significant (p ≤ .10), with discipline decreasing the odds of being categorized as an “A” by 33% and suspension increasing the odds by 47%.
The analysis of the IMS quantitative data provides insight into the EEOC’s decision-making process, reinforcing that gender identity charges and those containing allegations related to benefits, harassment and terms and conditions are important predictors of the EEOC’s categorization of a charge. The categorizations of these issues, however, do not provide information about what aspects of a benefits or harassment charge, for example, draw higher classification from the EEOC. In the next section, we present finer-grained coding of benefits, terms and conditions, hiring and harassment issues from the charge narratives to further develop a picture of the role of the EEOC in developing this new area of federal law.
Analysis of charge narrative data
In Table 3, we present the odds ratios from the logistic regression analyses predicting the EEOC processing category “A” for the fine-grained issue codes. For benefits, medical benefits for gender-affirming care is marginally significant (p ≤ .10) and has odds 324% higher of being categorized as an “A.” It is notable, given the small number of charges that contain this issue (see Table 1), that there is marginal significance; the finding emphasizes the importance the EEOC was placing on gender-related medical benefits claims.
Odds ratios predicting EEOC processing category, by issue (charge narratives)
*** Note: p ≤ .01;
** p ≤ .05;
* p ≤ .10.
Next, we more closely considered the varying stages of the hiring process. Although the offer to interview, the refusal to hire, or refusal to rehire were not statistically significant, charges that described rescinding an offer to hire had 149% higher odds of being categorized as an “A” (p ≤ .10). This could be indicative of the greater evidence of discriminatory motivation in such cases, where there is a documented offer and withdrawal of the offer that could occur after discovering information related to sexual or gender identity. Indeed, the decision in Macy v. Dept. of Justice (2012), where the EEOC first stated that gender identity discrimination is sex discrimination, involved a hiring decision that was allegedly affected by the discovery of the complainant’s name and gender changes.
Harassment based on SOGI is widely reported in survey data (Sears et al. Reference Sears, Mallory, Flores and Conron2021; Pew Research Center 2013; NPR Robert Wood Johnson Foundation and Harvard School of Public Health Reference Harvard2017), emphasizing its commonality as an experience of workplace discrimination. Our findings indicate, however, that some types of harassment receive higher categorization by the EEOC. First, charges that contain descriptions of physical violence have 47% higher odds of being categorized as an “A” (p ≤ .10). Similarly, charges that detail physical sexual harassment had odds 72% higher of being categorized as an “A” (p ≤ .05). These results indicate that the EEOC perceived physical forms of harassment as more likely to show cause, which is reflected within prior caselaw as well as early EEOC guidance on sexual harassment (see, e.g., EEOC 1990).
In addition, two forms of verbal harassment had higher odds of being categorized as an A. Charges that contained descriptions of general LGBTQ stereotypes directed at the charging party had odds 75% higher of being categorized as an “A” (p ≤ .05), and those that contained personal disparaging comments directed at the charging party have odds 32% higher of being categorized as an “A” (p ≤ .05). These comments were often tied to gender-based assumptions about appropriate behaviors and presentation. In contrast, charges that included descriptions of relationship disparagement (such as same-sex marriage) had odds 51% lower of being categorized as an “A” (p ≤ .10). These findings suggest that, when it comes to SOGI-based harassment, the EEOC perceived comments directed at individuals related to gender stereotypes surrounding identity as stronger cases of discrimination than anything relationship-based.
Finally, we looked more closely at the types of terms and conditions that were likely to produce a higher categorization by the EEOC. Charges alleging discrimination based on restroom or locker room access or those tied to access to professional development were not statistically significant in the regression analysis. However, those that included discrimination related to dress codes, such as that related to gender non-conforming dress, had 88% higher odds of being classified as an “A” by the EEOC (p ≤ .05).
Discussion
When evaluating the utility of the law to advance social change, particularly when it comes to the acquisition of new rights and protections for disadvantaged groups, attention is often focused on lawmakers and the courts. Advocates seek to appeal to traditional rule-makers to advance their cause, maintaining a relatively narrow vision of the possible “branches” in which to pursue disputes (Albiston et al. Reference Albiston, Edelman and Milligan2014). Our findings suggest that legal intermediaries, like the EEOC, can play a strong role in determining how boundaries of the law are to be defined – with the potential to recognize and expand rights not yet established by the traditional rule-makers. Through our analysis of which charges received higher processing categorization, we can see how one legal intermediary’s decision making resulted in shifting the availability and scope of legal rights. Thus, unlike many other legal intermediaries within the employment discrimination context, the EEOC’s actions worked to expand Title VII law rather than to facilitate symbolic compliance by organizations (see, e.g., Dobbin Reference Dobbin2009; Edelman Reference Edelman2016; Edelman et al. Reference Edelman, Petterson, Chambliss and Erlanger1991; Reference Edelman, Uggen and Erlanger1999; Talesh and Pelisse Reference Talesh and Pelisse2019).
Our results also confirm that the EEOC was strongly guided by existing, external sex discrimination precedent in the initial recognition of gender identity discrimination as prohibited by Title VII and its ensuing higher categorization of gender stereotype charges. But our findings point toward a subsequent expansion of this role, given the agency’s later decision to classify sexual orientation as sex discrimination and higher categorization of charges that were tied less directly to the gender stereotype body of cases. In the following discussion, we trace patterns in our findings that illustrate the ways that the EEOC’s processing categorization resulted in both the expansion and continued restraint on different SOGI employment rights.
The EEOC more readily extended Title VII sex discrimination protection to gender identity claims than to sexual orientation ones. This is reflected by its earlier decision on gender identity than sexual orientation (Baldwin v. Foxx 2015; Macy v. Holder 2012) but also in its processing categorization of SOGI charges. Our results indicate that the odds of being categorized as an “A” charge were about 70% lower for sexual orientation charges, as compared to gender identity. In many respects, this is unsurprising given that the agency had more clear external guidance regarding gender-related charges through the gender-stereotyping body of sex discrimination cases following Price Waterhouse v. Hopkins (1989). Although discrimination based on sexual orientation is frequently aligned with gender stereotypes, gender identity is almost necessarily so. As a result, the higher categorization of gender identity charges suggests a more constrained expansion of legal rights by the EEOC, given that the agency is guided by external prior precedent even as it expanded the scope of what constitutes sex discrimination.
Although gender identity charges had greater odds of being categorized as an “A,” the EEOC’s decision to categorize any sexual orientation charges as “A” reflects a more significant departure from external precedent. The EEOC’s strategic enforcement plan reflects that this was an agency-directed priority (EEOC 2012), indicating an internal push for expansion of sex discrimination law that was likely informed by political objectives. During this period, our findings indicate that a greater proportion of sexual orientation charges were categorized as an “A” at 30% than the percentage of all sex-based charges at 23% (McCann et al. Reference McCann, Tomaskovic-Devey and Lee Badgett2018), providing further evidence that sexual orientation charges were a priority for the agency.
Our analysis of the processing categorization of fine-grained charge allegations, however, reveals that the gender stereotyping body of caselaw appeared to have a powerful influence on how the EEOC was shaping SOGI discrimination law. Dress code violations – which were almost always described as being related to gender norm violations, such as men wearing makeup or feminine clothing, or women preferring the male-gendered uniform – had greater odds of being categorized as “A” by the EEOC. Discrimination based on gendered dress has a strong foundation within the gender stereotyping sex discrimination caselaw, with this being one of the issues alleged in Price Waterhouse v. Hopkins (1989). Thus, extension of the principles to SOGI charges did not require a large departure from settled law for the EEOC.
In addition, for both SOGI charges, harassment charges that contained descriptions of targeted verbal harassment directed toward the charging party based on their identity and those that involved leveling LGBTQ stereotypes against the charging party were more likely to be categorized as an “A.” The descriptions of these types of harassment within the charges focused primarily on gender-related comments, such as criticizing the way individuals and/or LGBTQ individuals dressed, styled their hair or make up, expressed themselves through mannerisms or exhibited what was perceived to be emotions or characteristics that were not in alignment with their sex assigned at birth. At first glance, these charges are grounded in gender stereotypes and, thus, require the more modest extension of Price v. Waterhouse to include SOGI-related allegations of this nature. At the same time, the EEOC’s handling of sexual orientation related charges involving these types of verbal harassment required the agency to recognize the gender-based nature of charges related to LGBTQ stereotypes as part of affording higher protection. Thus, there was an additional step taken by the EEOC in its classification of these charges as “A” as compared to similar gender identity-based claims, suggesting the agency was taking a broader interpretation of sex discrimination law.
Gender-based charges containing issues more commonly experienced by gender minorities also received higher categorization by the EEOC. In particular, those reporting benefits issues related to gender affirming care were substantially more likely to be categorized as “A,” signaling this as an important EEOC priority. Access to gendered spaces such as bathrooms and locker rooms was not statistically significant in the regression analysis, but about 50% of all of these charges were categorized as “A” (Table 1). Similarly, misgendering was also not significant in the regression analysis, but 60% of these charges were categorized as “A” (Table 1). Unlike some of the previously mentioned issues, these issues are tied less to gender stereotypes and more to gender embodiment. Although the EEOC’s handling of these charges is likely still connected to its extension of protection from gender stereotyping caselaw, its higher categorization of these charges did require more direct protection of gender as an identity category than the dress code and verbal harassment issues.
Two issues that were more commonplace for sexual orientation charges involved sexual harassment and relationship-based charges. The salience of sexual identity in the workplace for LGB individuals appears to produce more sexual harassment for sexual minorities (15% of charges) as compared to gender minorities (9% of charges). Charges alleging physical sexual harassment (e.g. touching, grabbing, kissing, etc.) had higher odds of being categorized as an “A” by the EEOC. The physicality of these charges could be an important aspect of the higher categorization, given that non-sexual physical harassment also had higher odds of categorization as an “A.” Outside of SOGI claims, however, sexual harassment charges tend to receive higher categorization by the EEOC than other sex discrimination charges, suggesting this was more generally a priority for the EEOC (McCann et al. Reference McCann, Tomaskovic-Devey and Lee Badgett2018). Thus, with respect to sexual harassment charges, the EEOC was likely drawing upon its internal practices regarding higher categorization of sexual harassment allegations in expanding coverage to SOGI-based claims.
In addition, harassment related to relationships, such as about one’s marital or dating partner, was more commonplace for sexual orientation charges (3%) than gender identity charges (1.5%). Unlike the verbal harassment tied to individual identity or group stereotyping, verbal harassment related to relationships reduced the odds of the EEOC classifying the charge as an “A.” Descriptions of this type of harassment often involved employees being instructed not to talk about their partners or spouses, harassment regarding employees’ relationships with other employees of the same sex or comments (often religious in nature) about the offensiveness of same-sex relationships or sexual activity. The distinction between more gender-based harassment and relationship-based harassment is similar to prior findings regarding Title VII caselaw related to sexual orientation charges. In cases handled prior to Bostock, Title VII was sometimes extended to cover gay individuals who openly appeared to be violating gender norms through their gender presentation but almost never covered those who violated gender stereotypes through sexual or dating behaviors (Soucek Reference Soucek2014). Our findings, similarly, suggest that the EEOC found gender-based SOGI claims as more likely to be successful extensions of sex discrimination law but exhibited more caution surrounding less visible aspects of sexual identity. Relationship-based protection requires a greater leap from the Price Waterhouse and other sex discrimination caselaw, and the agency might have been more hesitant to assume Title VII coverage in these cases.
Finally, the EEOC also seems to have been guided by pragmatic concerns in its categorization of some charges as “A.” Our findings indicate that both rescinding an offer of employment and items related to wages were more likely to receive higher prioritization by the EEOC. In some respects, these are issues that are more readily established, given that there tends to be more record-keeping and correspondence with respect to offers made and rescinded or wage and hours issues. This finding is in line with prior research that has found that more clear-cut charges tied to wages, for example, are preferred by both employees and administrators over more difficult to establish claims (Gleeson Reference Gleeson2016). This is not unique to SOGI charges (see, e.g., Berrey et al. Reference Berrey, Nelson and Beth Nielsen2017) but is more indicative of the way some charges fail to progress in the dispute process due to obstacles in establishing their facts. Nonetheless, the fact that prior research has established that gender identity discrimination is particularly prominent in the hiring phase (Make the Road New York 2010; Sears et al. Reference Sears, Mallory, Flores and Conron2021; Winter et al. Reference Winter2016) indicates that the EEOC’s categorization of these charges is important in addressing a form of discrimination that is serving as a barrier to access for employment for gender minorities.
Conclusion
Overall, our analysis of processing categorization suggests that EEOC actors have had a complicated role as legal intermediaries, both expanding rights and practicing conformity. Specifically, our findings demonstrate that the EEOC has been a key player in the period of rights expansion for LGBT individuals over the past two decades. Since 2012, the EEOC has appeared to take on a more lawmaking orientation toward SOGI rights, through its identification of SOGI charges as a strategic priority (EEOC 2012), its acceptance and processing of SOGI charges in the absence of a Supreme Court ruling on SOGI discrimination, its decisions in 2012 and 2015 holding that SOGI are sex discrimination (Baldwin v. Foxx 2015; Macy v. Holder 2012) and its subsequent guidance regarding what constitutes discriminatory conduct in this area (see, e.g., EEOC 2021). This behavior is reminiscent of the EEOC’s more activist orientation from the 1970s (Stainback and Tomaskovic-Devey Reference Stainback and Tomaskovic-Devey2012), signaling the agency could be influenced in part by both external and internal political changes that have prioritized LGBT rights. Although the agency was proactive in expanding sex discrimination to include gender identity and sexual orientation, its categorization of charges suggests that priorities were at least initially placed largely on gender stereotypes. This could have resulted in less protection for individuals who are, for example, fired or not hired based on sexual identity but where there is no clear link to gender stereotyping. In other words, during this period of EEOC charge processing, protection based on the identity category itself appears to have been more limited than protection that was tied more directly to gender expression or gender stereotypes.
Notably, the fact that the EEOC is doing some of this “fit” work of evaluating SOGI cases before lawyers necessarily get involved, means there may be downstream effects in how these cases are seen by lawyers and the likelihood they will advance certain claims in the courts. In other words, the EEOC is not only a gatekeeper but also does quite a bit of channeling and signaling work on what “real” and “legal” discrimination looks like in this emerging area of claiming. In this respect, there could be an adoption by other legal actors of their interpretation of Title VII’s protection of SOGI discrimination, in much the same way as we have seen other intermediaries influence developing caselaw (e.g. Edelman et al. Reference Edelman, Uggen and Erlanger1999; Talesh and Pelisse Reference Talesh and Pelisse2019).
It is difficult to directly measure the effect that the EEOC’s SOGI-related decisions, guidances and charge categorization have had on the development of SOGI protection within the courts. Undoubtedly, its acceptance of SOGI charges and decisions regarding Title VII’s coverage served as impetus for split decisions in the courts and, ultimately, the Supreme Court taking up the issue in 2020. Following the Bostock decision, the EEOC took actions that signaled a continued push to expand SOGI rights and there has been some evidence of judicial pushback. In June 2021, the EEOC issued a technical assistance document explaining the Court’s decision in Bostock and outlining the EEOC’s position regarding specific conduct that would constitute SOGI discrimination (EEOC 2021). The conduct includes, among others, requiring employees to dress in accordance with their sex assigned at birth, restricting access to a bathroom that aligns with the employee’s gender identity, or using pronouns or names inconsistent with the employee’s gender identity – reinforcing our findings of what appeared to be several EEOC priorities. Although the summary notes that the document explains the “EEOC’s established legal positions” on SOGI issues, the agency also includes a disclaimer that the document itself does not have the force and effect of law.
Nonetheless, this technical assistance document has been challenged twice in federal courts. In both cases, the courts based their decisions in part on the argument that the EEOC’s technical assistance document could not be binding on the plaintiffs because it presented positions on employer conduct that had not yet been ruled upon by the guiding circuit court or the Supreme Court (Tennessee et al. v. U.S. Dep’t of Educ. et al. 2022; Texas v. EEOC et al. 2022). In Texas v. EEOC, the court drew a distinction between protection against discrimination based on status (e.g. being gay or being transgender) and protection against conduct correlated with that status (e.g. dress codes, restroom access, pronoun use). In vacating the EEOC’s document, the court stated that the EEOC speculated as to prohibited conduct when the Bostock decision only spoke to status – appearing to narrow protection to the identity category, rather than the EEOC’s (and earlier courts’) path of primarily protecting gender-related characteristics tied to identity.
This dispute signals that the EEOC could be experiencing restraints on its behavior, much as it did during prior periods of conservative pushback on civil rights agendas (Stainback and Tomaskovic-Devey Reference Stainback and Tomaskovic-Devey2012). Similar backlash against LGBTQ rights achievements have been seen in the recent passage of state-level laws restricting access to gender affirming care, instruction on topics related to sexual and gender minorities within schools, the ability of transgender athletes to play on sports aligned with their gender identity and related rights (Baumle et al. Reference Baumle, Miller and Gregory2023). Given these changes, it is possible that the agency will experience greater restrictions on its ability to expand Title VII protection for SOGI discrimination.
Nonetheless, the EEOC’s handling of SOGI charges during this period has already left its mark. The primary focus on the courts in analyses of legal mobilization often masks the significant work that is being done in other branches of the dispute tree (Albiston et al. Reference Albiston, Edelman and Milligan2014). Our analyses support that, in processing SOGI charges, the EEOC has likely drawn upon judicial precedent on sex discrimination, external and internal political directives and pragmatic considerations regarding claim processing. Although we cannot definitively tie processing categorization to an EEOC agenda to foster the development of a new area of Title VII claiming, its effects have nonetheless opened a door for additional SOGI discrimination charges to be processed. While the breadth of SOGI protection remains unsettled by the courts, this outcome indicates the power of bodies other than the courts and the legislature in expanding rights (Edelman and Stryker Reference Edelman and Stryker2005; Talesh and Pelisse Reference Talesh and Pelisse2019).
The analyses in this paper are important for shedding light on the role that the EEOC, as an administrative actor, plays in the development of a new area of claiming. Nonetheless, the decision by the Supreme Court in 2020 to recognize SOGI discrimination as sex discrimination likely influences the way that EEOC actors understand and process newer SOGI cases. Further, there has been additional time for the development of caselaw at the federal level. Future studies in this area should examine whether there have been any resulting changes in the types of charges that have been classified as “A” by the EEOC given these notable changes in the law. In particular, there could have been an expansion away from the initial focus on gender-based charges.
In addition, our use of a novel dataset derived from coding the qualitative charge narratives demonstrates that reliance upon the IMS quantitative codes masks some of the nuance in the EEOC’s decision-making. Without the fine-grained coding of harassment, for example, we could not identify that harassment charges tied to gender stereotypes were more likely to be coded as showing cause than those tied to relationships. These findings provide strong evidence that analyses of employment discrimination outcomes could benefit from capturing greater detail about the charge’s allegations to best understand progression within the dispute process.
Furthermore, our reliance on charge processing categories provides us with an indirect measure of the EEOC’s priorities or rationale for its decisions. The categories indicate charges that fit within priorities and/or are likely to be successful in demonstrating discrimination. Some of the agency’s determinations are undoubtedly shaped by what charges it viewed likely to succeed in court, despite any EEOC-specific agenda. Our analysis, alongside EEOC-issued priorities and guidances, provides a strong indication of EEOC priorities, but we are limited in our ability to draw definitive conclusions. Our discussion of the types of charges brought alleging SOGI discrimination is also limited in the amount of detail we are permitted by the EEOC to disclose about the content of the charge narratives. Further research could draw on interviews or other sources of data to reveal more about the stories of individuals alleging SOGI discrimination.
Financial support
This material is based upon work supported by the National Science Foundation under Grant No. 1823815.
Conflict of interests
The authors declare none.
Amanda K. Baumle is Professor of Sociology at the University of Houston and a Research Affiliate at the Center for Employment Equity. Her research centers on the demography of sexuality and sociology of law, with a particular focus on workplace disparities. Her articles have appeared in outlets such as Social Forces, Social Science Research, Journal of Marriage and Family, Population Research and Policy Review, Journal of Homosexuality and others. She recently co-edited the Handbook on Transgender, Nonbinary, and Gender Minority Populations, with Sonny Nordmarken (Springer Press, 2022).
Steven A. Boutcher is Research Associate Professor at the University of Massachusetts-Amherst and is a Research Affiliate at the Center for Employment Equity. His research explores a wide range of topics, including employment discrimination, social movements, sexuality and the legal profession. Recent publications have appeared in Social Networks, Law & Social Inquiry, Socius, Journal of Empirical Legal Studies, Social Science Research and Mobilization. He also recently co-edited the Research Handbook on Law, Movements, and Social Change, with Corey Shdaimah and Michael Yarbrough (Edward Elgar, 2023).
M. V. Lee Badgett is the Chief Economist and co-founder of Koppa: The LGBTI+ Economic Power Lab, and Professor Emeritx of Economics and former director of the School of Public Policy at the University of Massachusetts Amherst. She is a Williams Distinguished Scholar at UCLA’s Williams Institute, where she was a co-founder and the first research director. She is a pioneering researcher of the global cost of homophobia and transphobia, economic empowerment of LGBTI+ people and LGBTI+ economic inequality. Her latest book is The Economic Case for LGBT Equality: Why Fair and Equal Treatment Benefits Us All (Beacon Press, 2020).
Donald Tomaskovic-Devey is Professor of Sociology at the University of Massachusetts-Amherst and the Director of the Center for Employment Equity. Recent publications have appeared in Nature, Human Behaviour, PNAS, Socio-Economic Review, Socius, the American Sociological Review and the American Journal of Sociology. His most recent monograph, with Dustin Avent-Holt, is Relational Inequalities: An Organizational Approach (Oxford, 2019).
