Gender inequality: Nonbinary transgender people in the workplace

Abstract This study uses the National Transgender Discrimination Survey to evaluate the employment outcomes of nonbinary transgender people (those who identify as a gender other than man or woman). Regression analyses indicate that being out as a nonbinary transgender person has different effects on nonbinary transgender people based on sex assigned at birth, with those assigned male at birth tending to be discriminated against in hiring but those assigned female at birth more likely to experience differential treatment once hired. Transgender women tend to have worse employment experiences than nonbinary transgender people and transgender men, the latter two tending to have similar outcomes.


Introduction
Most research on employment gender inequality focuses on the distinctions between men and women, reinforcing a binary conception of gender. Even the United States Equal Employment Opportunity Commission (EEOC) includes only male and female as gender options, meaning that the EEOC cannot identify who is transgender and that nonbinary transgender people (those whose gender identity is something other than only man or only woman) are not acknowledged and Recently, there has been increased coverage of transgender people in the mainstream media. However, mainstream media sources sometimes provide narrow or inaccurate information about transgender people, sometimes ignoring the existence of nonbinary transgender people (those who identify as a gender other than man or woman) entirely. This study uses information from the National Transgender Discrimination Survey to explore the employment outcomes of nonbinary transgender people. This study finds that being out as a nonbinary transgender person has different effects on nonbinary transgender people based on sex assigned at birth, with those assigned male at birth tending to be discriminated against during the hiring process but those assigned female at birth more likely to experience differential treatment once hired. Transgender women tend to have worse employment experiences than nonbinary transgender people and transgender men, the latter two tending to have similar outcomes.

Gender inequality
A variety of theories about inequality between men and women posit that categorical distinctions allow for the generation of inequality (e.g. Ridgeway, 2011;Tilly, 1998). Nonbinaries, however, do not inhabit a widely known and understood category, and transgender men and transgender women transition between categories. Because the norm in Western society is to view gender as a binary biological construct, transgender and gender nonconforming people challenge the categorical norms about gender and sexuality (Monro, 2003). This study builds upon research about gender categorization and inequality by exploring what happens to people in a group that is likely to be miscategorized or not categorized.

What is a transgender person? What are nonbinary genders?
Because transgender terminology is complex and rapidly changing, this section begins with definitions of important terms. "Transgender" is an umbrella term that refers to people whose gender identity differs from the sex they were assigned at birth at least some of the time, and the term "cisgender" refers to people whose gender identity corresponds to the sex they were assigned at birth. Thus someone who is not transgender is cisgender. Sex is a biological category: designations of male, female, or intersex are based on a number of factors, including chromosomes, hormones, and genitalia. Though it is typical to view sex as a binary of male and female, the way in which this is done has varied across time and place (Fausto-Sterling, 2000). Gender is distinct from sex, but related: it is the translation of biological realities into social expectations for "men" and "women" (Beemyn & Rankin, 2011;Sausa, 2002). As with sex, it is common to view gender as a binary and marginalize expression that does not fit within this binary.
Gender identity refers to individual people's sense of their own sex and/or gender, which may differ from their gender expression and from the way other people perceive their sex and/or gender (Beemyn & Rankin, 2011). People may consider themselves transgender because of their feelings about their biological sex, because of their feelings about gender roles, or both. Gender identity is distinct from sexual orientation, which is the pattern of a person's attraction to others (Sausa, 2002). Both transgender and cisgender people may identify with any sexual orientation, including but not limited to heterosexual, gay, lesbian, bisexual, queer, or asexual (Beemyn & Rankin, 2011). Gender identity is also distinct from gender expression, which is the degree to which someone expresses masculinity, femininity, both, or neither.
People's gender identity may be man or woman, or it may be something else, such as both man and woman, neither man nor woman, or a unique identity. For example, Beemyn and Rankin (2011) conducted a survey in 2005 and 2006, which was open to anyone who considered themselves part of the umbrella term "transgender." As in the National Transgender Discrimination Survey, there were varied written responses, some of which were unique (2011). Some gender identities that fall under the umbrella term "nonbinary gender" are genderqueer, agender, androgynous, Two-Spirit, gender nonconforming, gender variant, third gender, genderfluid, and bigender.
Gender dysphoria refers to transgender people's feelings of distress because of the mismatch between their sex assigned at birth and their gender identity. Gender dysphoria can take the form of physical dysphoria, which is distress regarding sex characteristics such as genitals, breasts and facial hair; social dysphoria, which is distress regarding social interactions such as being perceived as the incorrect gender or being forced to wear clothing associated with the incorrect gender; or both physical and social dysphoria. All three transgender groups (transgender men, transgender women, and nonbinaries) can experience gender dysphoria. In other words, in contrast with some misleading information in the mass media, nonbinary genders are identities comparable to those of "man" and "woman" rather than fashion choices.

Overview of workplace inequality for transgender people
Transgender people have reported difficulty securing and maintaining employment as a result of their gender identity and expression. Unemployment rates for transgender people are approximately twice as high as those for cisgender people (Grant et al., 2011), about the same difference as between whites and blacks (Bureau of Labor Statistics, 2015). About half of transgender people have reported adverse job outcomes, such as being fired, not hired, or denied a promotion as a result of their gender identity or expression (Grant et al., 2011). This is higher than the rates for cisgender people; for example, 5.6% of cisgender people report being fired because of discrimination, 16.0% report not being hired because of discrimination, and 12.7% report being refused a promotion because of discrimination (Kessler, Mickelson, & Williams, 1999). Being transgender can also influence someone's salary; Schilt and Wiswall (2008) found that while transgender women on average lose approximately a third of their salary after transitioning, transgender men on average see no change in their salary or a slight increase. This finding relates to the wage gap between men and women more generally. Transgender people of color, particularly African Americans, report poorer employment outcomes than white transgender people (Grant et al., 2011). The literature has established evidence of employment difficulty for transgender people. However, most studies on transgender people have used small qualitative samples without comparison baselines, so the literature has not established the scope of the problem or made comparisons of inequality between different categories of transgender people.

Varieties of discrimination in the workplace
Transgender people have identified a number of issues in their workplaces that influence their ability to feel comfortable in their work environment, demonstrating the variety of experiences transgender people have. For transgender people who are transitioning, one issue is workplaces' lack of procedures for ensuring that others in the workplace are aware of how to treat a transgender person who is transitioning (Whittle, Turner, & Al-Alami, 2007). Hierarchical workplaces can exacerbate this issue, because transgender people must carefully determine the degree to which they can be out at the workplace to avoid harassment or job loss (Dietert & Dentice, 2009). They may only be able to be out to some people-possibly only authority figures, and possibly never authority figures (Dietert & Dentice, 2009). Some transgender people who have already completed their transition attempt to avoid disclosing their transgender status (Dietert & Dentice, 2009).
Bathrooms are a common source of difficulty for transgender people in the workplace. Many report being refused access to bathrooms or being verbally or physically attacked in bathrooms (Herman, 2013;Nadal, Skolnik, & Wong, 2012). Another problem transgender people experience in a variety of settings is verbal harassment. People may call a transgender or gender nonconforming person by incorrectly gendered terminology, whether intentionally or not, and possibly in a public setting that causes embarrassment (Nadal et al., 2012). This is especially a problem for transgender women, who may be stereotyped as sexual deviants (Nadal et al., 2012).
Transgender and gender non-conforming people report that gender-specific dress codes cause them difficulty, because they require them to dress in a manner inconsistent with their identity (Levi, 2007). People may lose their jobs for disregarding dress code rules (Levi, 2007). In addition, conflicts between someone's gender identity or expression and official identity documentation can lead to confusion or unintended outing. Changing one's name and gender on official identity documents can be difficult, because changing one document is sometimes dependent on changing another, which is in turn dependent on another (Nadal et al., 2012). In most places, it is impossible to obtain official identity documents listing a nonbinary gender. Governments may not have explicit procedures for changing official identity documents (Whittle et al., 2007), and some transgender people are not interested in obtaining surgery, which may be necessary to change official documents. Employers who are unfamiliar with difficulties involved in changing identity documents may be confused and not know how to react (Whittle et al., 2007).

Hypotheses
There is a lack of data on nonbinary transgender people in the workplace; however, prior research regarding transgender and gender-nonconforming people in general has demonstrated that they tend to encounter hostile or confused reactions from people in everyday situations. Employment policies acknowledging and protecting transgender people are often lacking, and when they do exist, they may elicit confusion and negative reactions. In many situations, nonbinaries' identities are not acknowledged, and nonbinaries are forced to affiliate with a binary gender option. Thus in many employment situations, nonbinaries find it difficult to fit in, to be acknowledged and accepted by coworkers. This study explores differences in outcomes that can occur as a result of hostile treatment on the part of employers or coworkers.
Hypothesis 1: Nonbinaries who are out about their gender identity will encounter more negative employment outcomes compared to nonbinaries who are not out about their gender identity.
Hypothesis 2: Nonbinaries who were assigned male at birth will encounter more negative employment outcomes than nonbinaries who were assigned female at birth.
As described in the literature, transgender women (assigned male at birth) tend to experience worse outcomes than transgender men (assigned female at birth). Prior research suggests that people assigned male at birth who deviate from masculinity (a valued characteristic) will encounter negative outcomes for doing so, and people assigned female at birth who deviate from femininity (a less valued characteristic) will encounter less hostility for doing so.
Hypothesis 3: Nonbinaries of color will encounter more negative employment outcomes than white nonbinaries.
Gender identity intersects with race to create additional challenges for people of color.
Hypothesis 4: Transgender women will encounter more negative employment outcomes than nonbinaries. Transgender men will encounter better employment outcomes than nonbinaries.
Prior research on transgender women describes them as overwhelmingly experiencing negative outcomes. In contrast, there are sometimes benefits for transgender men, such as an increase in salary. Employers, customers, and coworkers may be more likely to react to nonbinaries with confusion rather than either hostility directed toward a highly devalued category (that of transgender woman) or the respect that is often granted to men.

Overview of the NTDS
This study uses data from the National Transgender Discrimination Survey (NTDS). The NTDS met the ethical standards of the Pennsylvania State University Institutional Review Board with regard to confidentiality and humane treatment of research participants. The researchers obtained participants through about 800 transgender organizations and about 150 online listserves, and the total number of respondents in this survey was 6,456. The NTDS follows the precedent of Blumstein and Schwartz's (1983) American Couples study in size and methods. For that study, they surveyed about 6,000 couples in order to investigate experiences of marriage and cohabitation across sexual orientation, demonstrating the value of large-sample non-random surveys for studying LGBT people.
In contrast to much prior research on transgender people, based on small samples, often from clinical settings (Kuper et al., 2012), the NTDS was designed to be large and as representative as possible of the transgender and gender nonconforming population of the United States. Though it is not a random sample and probably has some representational bias, perhaps underrepresenting racial and ethnic minorities and overrrepresenting highly educated people (Harris, 2015), it does represent a demographically diverse population, including substantial variation along the lines of race, education, and age.

Coding decisions
The question about sex assigned at birth had two responses, "male" and "female." The question about gender had four responses, "male/man," "female/woman," "part time one gender, part time another," and "other." The NTDS also had questions in which respondents could express the degree to which they identify "not at all," "somewhat," or "strongly" with certain transgender terms.
Write-in responses were coded as nonbinaries, regardless of whether they were assigned male or female at birth, for a total of 859. Added to the respondents who wrote in their gender are the 695 respondents who chose "part time as one gender, part time as another" who identified strongly with the terms gender nonconforming or gender variant, genderqueer, androgynous, third gender, Two-Spirit, and other, which are all terms that fall under the umbrella category of "nonbinary gender." Thus there is a total of 1,554 nonbinaries.
The category of transgender women includes those people who chose "female/woman" as their gender identity but did not choose "female" as their sex assigned at birth (2,273 people). In addition, the category of transgender women is composed of those people who chose "part time as one gender, part time as another" and also stated that they identified strongly with the term "male to female" (679 people), for a total of 2,952 transgender women. Similarly, the category of transgender men contains those people who chose "male/man" as their gender identity but did not choose "male" as their sex assigned at birth (1,319 people). In addition, this category includes those people who chose "part time as one gender, part time as another" and also stated that they identified strongly with the term "female to male" (119 people), for a total of 1,438 transgender men. Respondents who fit none of these criteria, including cross-dressers and drag queens (1,214), were dropped.

Variables
The employment outcome dependent variables are whether the respondent is currently unemployed and whether as a result of being transgender the respondent has been underemployed, lost a job, been denied a promotion, and been removed from direct contact with clients, customers, or patients. Both those respondents who are currently looking for a job and those who have stopped looking count as unemployed. The other four dependent variables are yes/no questions.
The outness and transgender appearance variables are ordinal. The question regarding outness ("How many people know or believe you are transgender/gender nonconforming on the job?") has these responses: none, a few, some, most, and all. For simplicity, these responses were collapsed into three categories: none, some (composed of "a few" and "some" from the previous question), and most/all (composed of "most" and "all" from the previous question). The three transgender groups are out at similar rates in the workplace (34% of nonbinaries, 34% of transgender men, and 44% of transgender women report being out to most or all people). Outness may reflect different choices for nonbinaries compared to the other two transgender groups, however, because nonbinary genders are not mainstream categories. It is possible for a transgender man or a transgender woman to appear cisgender, so some transgender men and transgender women are able to present themselves as their gender identity without calling attention to their transgender status. In contrast, a nonbinary transgender person must identify themselves as transgender in order for their gender identity to be recognized.
The question about transgender appearance ("People can tell I'm transgender/gender non-conforming even if I don't tell them") has these responses: always, most of the time, sometimes, occasionally, and never. This question was collapsed into three responses: always/most of the time (composed of "always" and "most of the time" from the previous question), sometimes (composed of "sometimes" and "occasionally" from the previous question), and never. The survey did not have the means to compare nuances of appearance, such as whether people know whether someone is transgender based on clothing, secondary sex characteristics, mannerisms, a combination of these, or something else. Table 1 descriptively compares nonbinaries, transgender women, and transgender men in terms of labor market outcomes. Compared to nonbinaries, transgender women have a higher unemployment rate and experience more underemployment, job loss, denial of promotions, and removal from contact with clients, customers, or patients. Nonbinaries are more likely to have experienced underemployment, been denied a promotion or been removed from direct contact with clients or customers than transgender men. Like transgender men, nonbinaries have better outcomes than transgender women on all measures. All groups of transgender people have a higher unemployment rate than the general population. (The unemployment rate, including discouraged workers, for the United States as a whole in 2008 was 10.5% Bureau of Labor Statistics, 2011). The following regression analyses explore the magnitude of these differences, controlling for relevant demographic and human capital factors.

Regression models
Models for all dependent variables use logistic regression. This study reports odds ratios (a measure of the likelihood of an event occurring). Each model includes outness, appearance (appearing transgender or gender nonconforming), and sex assigned at birth. For sex assigned at birth, 0 stands for male and 1 stands for female; for outness, "not out" is the reference group, and for appearance, "not visibly transgender" is the reference group. There is only a moderate correlation (0.38) between outness and appearance. In addition to these variables, each model also contains several demographic and human capital variables available in the NTDS data-set that typically influence employment outcomes: education (an ordinal measure including below high school, high school, associate's degree/technical school/some college, bachelor's degree, and graduate or professional degree), race (including white, black, Latino, Asian, and other/mixed, with white as the reference group), age, and disability (yes/no). There are also models for nonbinaries that add an interaction effects between sex assigned at birth and outness.

Results and discussion
This section proceeds in the order of the hypotheses. All significance values are for two-tailed tests.

Hypothesis 1
Hypothesis 1 was: nonbinaries who are out about their gender identity will encounter more negative employment outcomes compared to nonbinaries who are not out about their gender identity. There is some evidence that nonbinaries' outness influences their employment outcomes, although not all of it supports Hypothesis 1. As Table 2 shows, on average, the odds ratio of being unemployed is 0.64 (corresponding to a log odds of −0.44) for each increase of one value on the three-value scale of outness, but the odds ratio of being denied a promotion is 1.27 (corresponding to a log odds of 0.24) for each increase of one value on the three-value scale of outness. This may mean that when employers recognize nonbinaries as a category, they are willing to employ them, though they may discriminate against them in job assignment. It may also mean that nonbinaries conceal their gender identity when applying for jobs and are penalized if they come out later on. This would concur with the fact that there was no statistically significant relationship between outness and underemployment or job loss, and with Dietert and Dentice's (2009) explanation of how transgender people carefully determine when and how to come out in the workplace.

Hypothesis 2
Hypothesis 2 was: nonbinaries who were assigned male at birth will encounter more negative employment outcomes than nonbinaries who were assigned female at birth. There is conflicting evidence for this hypothesis. As Table 2 shows, the odds ratio of being removed from direct contact with clients, customers, or patients for those assigned female at birth, compared to those assigned male at birth, is 0.66 (corresponding to a log odds of −0.41). In an attempt to further explore the impact of outness, one model investigates the interaction between sex assigned at birth and outness, as shown in Table 3. Employers' reactions to nonbinaries are highly conditioned by sex assigned at birth. Outness does not increase the odds of unemployment for nonbinaries assigned male at birth, but it is associated with substantially lower unemployment among those who were assigned female at birth. In contrast, outness does not increase the odds of underemployment, job loss, denial of promotion, or removal from contact with clients, customers, or patients for nonbinaries assigned male at birth but is associated with higher levels of those four negative employment outcomes for nonbinaries assigned female at birth. Thus outness appears to primarily contribute to discrimination in hiring for nonbinaries assigned male at birth and to discrimination while on the job for nonbinaries assigned female at birth. Specifically, the odds ratio for female*outness for unemployment is 0.58 (corresponding to a log odds of −0.55); for underemployment, 1.52 (corresponding to a log odds of 0.42); for job loss, 1.65 (corresponding to a log odds of 0.50); for denial of promotion, 1.68 (corresponding to a log odds of 0.52); and for removal from contact, 1.70 (corresponding to a log odds of 0.53).
These interactions may mean that while employers are inclined to resist employing nonbinaries assigned male at birth outright, employers are inclined to police what they perceive as a rejection of femininity in a person assigned female at birth among people currently in their employment. This would be consistent with Schilt and Connell's (2007) analysis of how people police gender expression in the workplace. Though Schilt and Connell focused on transgender men and women, their analyses of masculinity and femininity parallel nonbinaries' experiences. They provided examples of transgender women whose coworkers perceived them as losing competence in their work upon their transition to female, which parallels the experiences of unemployed nonbinaries assigned male at birth.

Table 3. Regression models for nonbinaries only, including interactions
Notes: All coefficients were derived from logistic regression; the columns display log odds. Standard errors are in parentheses. They also provided examples of transgender men who reported that women coworkers communicated with them as they do with other women. This parallels the salience of gender in the workplace for nonbinaries assigned female at birth.

Hypothesis 3
Hypothesis 3 was: nonbinaries of color will encounter more negative employment outcomes than white nonbinaries. Consistent with prior research on transgender men and women and providing support for Hypothesis 3, nonbinaries of color sometimes experience worse outcomes compared to white nonbinaries. As Table 2 shows, on average, the odds ratio of being unemployed for blacks compared to whites is 1.99, corresponding to a log odds of 0.69; for having lost a job, 2.18, corresponding to a log odds of 0.78; and for having been removed from direct contact with customers, clients, or patients, 2.51, corresponding to a log odds of 0.92. On average, the odds ratio for having been underemployed for people of mixed race or ethnicity or people of a race or ethnicity other than white, black, Latino, or Asian compared to whites is 1.70, corresponding to a log odds of 0.53; for having lost a job, 3.56, corresponding to a log odds of 1.27; for having been denied a promotion, 3.90, corresponding to a log odds of 1.36; and for having been removed from direct contact with customers, clients, or patients, 2.86, corresponding to a log odds of 1.05. These coefficients suggest that black and mixed race nonbinaries experience racial bias. Gender may intersect with race to emphasize the stereotypes of people of color, particularly African Americans, as threatening, angry, or violent (Schilt, 2010).

Hypothesis 4
Hypothesis 4 was: Transgender women will encounter more negative employment outcomes than nonbinaries. Transgender men will encounter better employment outcomes than nonbinaries. There is evidence that some outcomes are worse for transgender women. Specifically, as shown in Table 4, on average, the odds ratio for having lost a job for transgender women compared to nonbinaries is 2.16, corresponding to a log odds of 0.77, and the odds ratio for having been removed from direct contact with clients, customers, or patients for transgender women compared to nonbinaries is 1.43, corresponding to a log odds of 0.36. There are no statistically significant differences between the outcomes of transgender men and nonbinaries, suggesting again that being a woman intersects with being transgender to lead to negative outcomes. This result corroborates prior research on transgender women's experiences, such as Westbrook and Schilt's (2013) evaluation that women's spaces are highly policed and thus potentially hostile to transgender women.

Conclusion
The results from this study suggest that being out at the workplace has negative effects for nonbinaries. Nonbinaries who are more out at the workplace are less likely to be unemployed but more likely to have been denied a promotion, indicating that although employers may be willing to work with nonbinaries, some are inclined to discriminate against them in access to authority. While employers appear more inclined to avoid hiring nonbinaries assigned male at birth compared to those assigned female at birth, they appear more inclined to discriminate against nonbinaries assigned female at birth than those assigned male at birth once hired, suggesting that employers are inclined to police perceived femininity. The results also indicate that nonbinaries of color, particularly African Americans, tend to face major challenges. The fact that nonbinaries of color have the worst outcomes confirms that an intersectional analysis of transgender issues is necessary in order to understand how transgender people are treated.
The fact that the three transgender groups sometimes experience the same challenges, particularly if they are people of color, indicates that some similar interventions may improve their workplace outcomes. Broader education about and recognition of nonbinaries could help put into practice effective transgender-inclusive policies. Following from Dietert and Dentice's (2009) emphasis on the importance of upper management in protecting transgender people, businesses can do a variety of things to improve transgender people's experiences in the workplace.
Businesses can be inclusive of transgender people by evaluating their methods of recording gender, choosing to record it only when necessary and permitting people to respond with gender identity (including the option to write in a response) rather than sex when possible (Grant et al., 2011;Miller & Weingarten, 2005;Sausa, 2002). In the case of businesses with more than 100 employees (50 if federal contractors), the EEOC currently mandates that they report employees' gender, but the EEOC only permits the options of male and female. If this were changed to provide transgender-inclusive options (adding transgender men, transgender women, and nonbinaries), researchers would be able to investigate transgender employment inequality with more precision. If the EEOC did provide this option, it would educate employers that the category exists and is protected. Businesses could permit transgender people to change their name and gender on that institution's documentation, even if they have not done so legally (Beemyn, 2005). Because the number of people who are likely to use transgender-specific options is small, the inclusion of these options should not pose a burden for data collection (Miller & Weingarten, 2005). Businesses can include gender identity and expression on antidiscrimination policies and implement standards for reporting transphobic incidents (Sausa, 2002).
Finally, there are some things researchers can do in the future to improve our understanding of transgender people's employment outcomes. Though the NTDS was intended to provide comprehensive information on transgender people, it was not derived from a random sample. Institutions that have the resources for large-scale or random sampling could include transgender-inclusive questions about gender (Grant et al., 2011). The size of the transgender population has been estimated at 0.3% of people (Gates, 2011), so a large-scale survey with transgender-inclusive questions would be needed to provide an adequate sample size of transgender men, transgender women, and nonbinaries. Although coefficients for most control variables in my analyses were nonsignificant, the intersections between transgender status and education or disability would be useful topics for future research. Because transgender people with disabilities are at a high risk of developing psychological difficulties as a result of the multiple types of discrimination and oppression they experience, it is essential for social scientists to understand their needs better (Ballan, Romanelli, & Harper, 2011).
The NTDS had only one question about the degree to which others can tell whether the respondent was transgender or gender nonconforming. Because the NTDS is unable to determine nuances of appearance, researchers should also investigate how different types of gender presentation, such as masculine, feminine, mixed, and neutral, influence how transgender people are treated. Future studies should explain nuances of appearance, such as perceptions of transgender people's clothes, mannerisms, and sex assigned at birth. Researchers could also investigate how transgender people alter their gender expression for work environments. Finally, surveys regarding transgender people in the workplace should ask for information on the industry, occupation, and characteristics of employers and workplaces, such as transphobic employers or the existence of an antidiscrimination statement. This would allow us to understand the degree to which transgender people attempt to sort themselves into more supportive work environments.