The Social Costs of Gender Nonconformity for Transgender Adults: Implications for Discrimination and Health.

Research suggests that transgender people face high levels of discrimination in society, which may contribute to their disproportionate risk for poor health. However, little is known about whether gender nonconformity, as a visible marker of one's stigmatized status as a transgender individual, heightens trans people's experiences with discrimination and, in turn, their health. Using data from the largest survey of transgender adults in the United States, the National Transgender Discrimination Survey (N = 4,115), we examine the associations among gender nonconformity, transphobic discrimination, and health-harming behaviors (i.e., attempted suicide, drug/alcohol abuse, and smoking). The results suggest that gender nonconforming trans people face more discrimination and, in turn, are more likely to engage in health-harming behaviors than trans people who are gender conforming. Our findings highlight the important role of gender nonconformity in the social experiences and well-being of transgender people.

Introduction

Gender theorists have long suggested that no social space exists in Western societies for individuals who deviate from binary gender systems (Lorber 1994). Indeed, transgender people are systematically oppressed and experience high rates of discrimination and violence in the United States (Clements et al. 1999; Factor and Rothblum 2008b; Lombardi et al. 2002). For example, transgender people frequently face subtle, day‐to‐day expressions of discrimination, including others' expressed discomfort, as well as physical threats, and harassment (Nadal, Skolnik, and Wang 2012). However, researchers have not examined whether gender nonconformity—as a visible and known marker of one's stigmatized status—shapes trans people's experiences with discrimination and health. This is particularly surprising, given that gender theorists have posited that transgender people primarily face social penalties due to perceptions that they are not normatively “doing gender” (West and Zimmerman 1987).

In this article, we use data from the landmark 2008 National Transgender Discrimination Survey (NTDS) to examine whether gender nonconforming transgender people face more discrimination and worse health than their gender conforming counterparts. In particular, we examine three research questions. First, is perceived gender nonconformity associated with exposure to day‐to‐day and major events of transphobic discrimination? Second, is more exposure to transphobic discrimination associated with greater likelihood of engaging in health‐harming behaviors (i.e., attempted suicide, drug/alcohol abuse, and smoking)? Finally, does self‐reported transphobic discrimination mediate the relationships between gender nonconformity and these health‐harming behaviors? In other words, does transphobic discrimination help to explain why gender nonconforming transgender people may have worse health problems? Taken together, our analyses highlight the role that gender nonconformity plays in transgender people's experiences with discrimination and poor health.

Background

The Lives of Transgender People

We use the terms transgender people and trans people to refer to individuals whose gender identity and expression do not normatively align with their assigned sex. The term gender nonconformity is used as a way to signify that one's gender expression breaks cultural expectations for normatively “doing gender” (West and Zimmerman 1987). In addition, the term gender transitioning is used to signify a social process wherein modifications are made to one's appearance, style of dress, hair, body, hormones, physical anatomy, and pronoun/name usage. The decision to undergo gender transitioning is typically motivated by one's desire to affirm one's gender identity (Mason‐Schrock 1996). Rather than a singular event, transitioning is generally a process that unfolds over time, taking anywhere from several months to several years.

Transgender people are stigmatized in Western societies that are characterized by a binary gender system (Lorber 1994). These societies do not offer a social space or societal recognition for individuals who identity with a sex other than the one assigned to them at birth (Gagne and Tewksbury 1998; West and Zimmerman 1987). Because of this, trans people who decide to undergo gender transitioning often face major social, economic, and legal risks, including discrimination in both institutional and interactional arenas (Gagne and Tewksbury 1998). Particularly alarming is that the majority of trans people report that they face discrimination within a number of social institutions (e.g., the workplace and the medical establishment), in addition to daily prejudiced encounters and violence in social interactions (Clements et al. 1999; Nadal et al. 2012; Stotzer 2009). For instance, trans people frequently encounter the use of incorrect gender terminology, denial of bodily privacy, exoticization, and assumptions of sexual pathology (Nadal et al. 2012).

The Doing and Undoing of Gender

Although the majority of trans people may face stigma in society, we do not yet know whether certain members of this population face more transphobic discrimination than others. Transphobia takes the general form of prejudice and hostility toward the existence of transsexuality, but it may also be heightened in some social contexts if a person is more readily and frequently read as gender nonconforming. It is important to note, however, that the social penalties for gender nonconformity do not stem from the individual failings of transgender people; rather, it is a social problem that takes its root in structures that do not permit gender nonconformity on the part of social actors.

Relying in part on theories of “doing gender,” we ask whether transgender people face more severe social consequences if they are read by others as gender nonconforming. Symbolic interactionists have argued that all social actors are accountable for normatively “doing gender” in ways that are consistent with the sex assigned to them at birth (West and Zimmerman 1987). Specifically, one may face ostracism in society if others assume that one's sex (assignment at birth as male or female), gender identity (subjective sense and labeling of one's own gender), and gender expression (social presentation of gender) do not align (see Pfeffer 2010 for a review). Others have importantly referenced this as a moment of “misrecognition” that carries material and social consequences (R. Connell 2009; Pfeffer 2014). This problem is perhaps best exemplified when transgender people enter bathrooms that are not considered consistent with their perceived sex by social onlookers, which often leads to trans people experiencing harassment and bodily harm (Halberstam 1998; Herman 2013).

Our article is well suited to make contributions to debates in gender scholarship about the “doing” and “undoing” of gender. Within recent years, some scholars have argued that an overemphasis has been placed upon gender conformity, resulting in a tendency to ignore instances of gender resistance (Deutsch 2007; Risman 2009). Those on this side of the debate argue that there is already evidence of degendering in society, offering the existence of transgender and genderqueer people as one such example (Lorber 2005; Risman, Lorber, and Sherwood 2012). Other scholars maintain that gender can never truly be “undone,” only “redone,” because people are still held accountable for adhering to normative gender expectations even as its possibilities and meanings change (West and Zimmerman 2009). Despite theoretical disagreements, scholars maintain that these questions would be best explored and resolved with further research on the everyday lives of transgender people (Risman et al. 2012; Vidal‐Ortiz 2009; Williams, Weinberg, and Rosenberger 2013). What is largely missing from this literature is an understanding of whether transgender people face more severe social consequences for being read as gender nonconforming. We thus provide an empirical investigation of this and also offer an understanding of the health consequences of these dynamics.

Gender Nonconformity, Discrimination, and Health

Minority Stress Model and Transgender Health

Transphobic discrimination has consequences for many aspects of transgender people's lives, including their health and well‐being. We draw on minority stress theory to help explain the potential linkages among gender nonconformity, discrimination, and health‐harming behaviors. The primary theoretical aim of the minority stress framework is to illustrate how social environments contribute to the relatively poor health status of lesbian, gay, and bisexual people (Meyer 1995). In particular, sexual minorities face disproportionate exposure to unique, group‐specific stressors (i.e., prejudice and discrimination), which contribute to sexual orientation disparities in health (Hatzenbuehler 2009; also see Thoits 2010). Individuals who face discrimination are more likely to engage in health‐harming behaviors (e.g., self‐harm, drug use, and smoking), particularly because they have fewer resources and less energy to cope and make healthy behavioral choices (Cochran and Mays 2000; Pascoe and Richman 2009).

Preliminary research using convenience samples of trans people documents the applicability of the minority stress model for understanding transgender health (Bockting et al. 2013), wherein evidence suggests that transphobic discrimination is linked with health problems (e.g., attempted suicide, depression, and substance abuse) (Bradford et al. 2013; Clements‐Nolle, Marx, and Katz 2006; Hendricks and Testa 2012). There is also evidence that trans people experience worse health than their cisgender (i.e., non‐transgender) counterparts (IOM 2011). However, most of these studies have relied on fairly limited measures of discrimination, focusing primarily on workplace discrimination. Despite the possibility that gender nonconformity may exacerbate transphobic discrimination (Gordon and Meyer 2007), none of these studies have examined the role that gender nonconformity plays in minority stress processes for trans people.

Gender Nonconformity as a Form of Stigma Visibility

We extend the aforementioned line of research by considering whether transgender people who are more visibly stigmatized, by virtue of being read as transgender and gender nonconforming, face more discrimination and, in turn, worse health. In doing so, we build upon prior research on other stigmatized populations that shows that the visibility of one's stigmatized status can make one vulnerable to discriminatory treatment and psychological distress (Frable, Wortman, and Joseph 1997; Stutterheim et al. 2011). We define stigma visibility as the extent to which one holds a known, visible, conspicuous, and discredited stigmatized status. The concept builds upon Goffman's (1963) work on “discredited stigmas” (i.e., known and visible to others) and “discreditable stigmas” (i.e., hidden and invisible). Most centrally, stigma visibility as a concept highlights the fact there are sometimes visible, conspicuous, and known markers on the body that reveal a person's stigmatized status to others.

In this article, we conceive of gender nonconformity as a form of stigma visibility, signifying whether others are able to “read” that a trans individual's sex, gender identity, and gender expression do not align (C. Connell 2010; Kando 1972). In this way, gender nonconformity serves as a visible and known marker of one's status as a trans person, and, most importantly, it can shift one's status as transgender from concealable to conspicuous. Preliminary evidence suggests that stigma visibility is particularly consequential for trans people (Beemyn and Rankin 2011; Levitt and Ippolito 2014). For instance, prior qualitative studies show that trans people who are routinely “read” as transgender and gender nonconforming face more discrimination in the workplace and their everyday lives than those who are not (C. Connell 2010; Dozier 2005). Although these studies serve as an important starting point, further work is needed to assess whether these patterns hold for larger populations of trans people, and within other contexts. We advance this line of work by examining the association between gender nonconformity and transphobic discrimination, as well as the health consequences of these experiences.

Methods

Data

We use cross‐sectional data from the NTDS (Grant et al. 2011), the largest and highest‐quality survey of trans people in the United States to date. The survey covers a wide array of substantive areas, including health‐harming behaviors and exposure to discrimination. The NTDS was designed and disseminated by the National Center for Transgender Equality and the National LGBTQ Task Force, fielded in 2008 with the assistance of Pennsylvania State University's Consortium on Higher Education. Noninstitutionalized adults age 18 and older who identify as transgender, transsexual, or gender nonconforming were eligible to participate. Respondents were recruited through advertisements distributed via 800 transgender‐serving community organizations and 150 active transgender community e‐mail listservs.

It is important to note that the NTDS sample is large and diverse, but it is not a nationally representative sample of the transgender population in the United States. Indeed, the size and geographic distribution of this hard‐to‐reach population remains unknown, decreasing the feasibility of collecting representative data (Meier and Labuski 2013). However, additional steps were taken in the NTDS to reduce sampling bias. In addition to the online survey, paper versions of the survey were administered at homeless shelters, legal aid clinics, and mobile health clinics. This yielded an additional 500 respondents from the most disadvantaged segments of the population, including those living in rural areas, experiencing poverty and homelessness, and trans people of color (Reisner et al. 2014). This offers an advantage over prior surveys of trans people that rely exclusively on Internet‐based recruitment, thus underrepresenting these disadvantaged communities (Beemyn and Rankin 2011; Kuper, Nussbaum, and Mustanski 2012; Rosser et al. 2007). In our analyses, we focus exclusively on the 4,790 transgender respondents in the NTDS. We exclude the 1,660 genderqueer and other respondents who do not identify as transgender because their sense of gender identity and expression—namely, their greater emphasis on gender nonconformity—is qualitatively different than those of trans people (Factor and Rothblum 2008b; Gagne and Tewksbury 1998; Harrison, Grant, and Herman 2011–2012).4 Our final sample includes 4,115 trans adults after using listwise deletion for missing data.5 Data were systematically missing in one way only: survey version (online or paper), wherein respondents who completed the paper version of the survey are more likely to be missing information. However, supplemental analyses excluding the 152 respondents who completed the paper version of the survey yield similar results to those presented (available upon request).

Measures

Gender Nonconformity

We focus on respondents' perceived gender nonconformity. Perceived gender nonconformity is measured by responses to the following statement: “People can tell I'm transgender/gender nonconforming even if I don't tell them.” Responses ranged from “never” (0) to “always” read as transgender or gender nonconforming (4); higher values represent greater gender nonconformity. It should be noted that this item likely reflects current and/or recent visibility of one's transgender identity.

Transition Status

Because the extent to which trans people have transitioned influences their level of perceived gender nonconformity (analyses available upon request), we also examine respondents' transition status. Medical transition status is a measure of the medical procedures that respondents have pursued to transition their physical sex to match their gender identity. We include separate dichotomous variables for respondents who have only received hormonal therapy (1 =  yes) and those who have undergone any surgical procedures (regardless of use of hormones) (1 =  yes) compared to respondents who have not pursued any medical procedures (1 =  yes).6 We acknowledge, however, that receiving hormones or undergoing surgery is only one step in the transition process, and that some transgender people live full time as their desired gender irrespective of whether they have completed medical procedures (Factor and Rothblum 2008b).7 Thus, similar to other scholars (Bauer et al. 2012; Bradford et al. 2013), we also include a measure of social transition status—a binary indicator for whether respondents report currently living full time in their desired gender (1 =  yes).

Transphobic Discrimination

NTDS respondents were asked a series of questions about their experiences with discrimination. Using these data, we assess two types of self‐reported transphobic discrimination: major discrimination and everyday discrimination (Thoits 2010). To measure major discrimination, we created a scale of discriminatory events (1 =  yes to each event) in which respondents were denied equal treatment or access to opportunities and services necessary for their livelihood and well‐being. The scale includes 26 types of discriminatory events (e.g., fired, not hired, denied health care, harassed by police) and ranges from 0–26 (M = 3.20). To measure everyday discrimination, we created a scale of discriminatory events (1 =  yes to each event) in which respondents were harassed or treated unfairly by coworkers, staff at public accommodations, or personnel in legal or medical settings. The scale includes 11 types of events (e.g., referred to by wrong pronoun intentionally, denied service at a hotel or restaurant, harassed by airport security) and ranges from 0–11 (M = 2.48).8 The major discrimination and everyday discrimination scales represent the number of distinct types of events of discrimination respondents have experienced. However, it should be noted that these measures do not allow us to determine how frequently each distinct event occurred.

Health‐Harming Behaviors

We examine two health‐harming behaviors that are associated with experiences of interpersonal discrimination (Cochran and Mays 2000; Pascoe and Richman 2009): drug/alcohol abuse and smoking. We measure drug/alcohol abuse as a binary indicator of whether respondents answered yes (1) or no (0) to the following statement: “I drink or misuse drugs to cope with the mistreatment I face or faced as a transgender person.” Smoking is a binary indicator of whether respondents report that they currently smoke cigarettes (1 =  yes). In addition, we assess a third, more extreme form of self‐harm—attempted suicide—that has received less attention in prior research on discrimination outside of broad measures of depressive symptoms (Williams and Mohammed 2009). Attempted suicide is a binary measure of whether respondents have ever attempted to commit suicide in their lifetime (1 =  yes).

Social Status

Prior research suggests that trans people's experiences vary by race and ethnicity, social class, and gender identity (Butler 2004; C. Connell 2010; Dozier 2005; Schilt 2006). As such, we include a number of variables to assess the associations among social status, discrimination, and health‐harming behaviors. We measure gender identity using a dichotomous variable (1 =  trans woman; 0 =  trans man). We include dichotomous variables to measure race and ethnicity, compared to non‐Hispanic whites (1 =  yes for each): non‐Hispanic Black, Latina/o, American Indian or Alaska Native, Asian/Pacific Islander, and multiracial. Education is a measure of the level of education respondents have completed (M = 2.53), ranging from less than high school (0) to a graduate degree (4). We measure income using the natural logarithm of respondents' annual household income. Age is measured in years (M = 37.76 years). Sexual identity is a dichotomous variable indicating respondents' self‐reported sexual identity (1 =  sexual minority; 0 =  heterosexual). Finally, all multivariate analyses control for survey version—a binary indicator for the version of the survey respondents completed (1 =  paper; 0 =  online).

Analysis Plan

To assess whether perceived gender nonconformity is associated with transphobic discrimination and health‐harming behaviors among transgender adults, our analyses proceed in a series of steps. We begin by providing descriptive statistics for the NTDS sample. Next, using negative binomial regression modeling, we estimate the association between perceived gender nonconformity and the number of types of events of major and everyday discrimination reported. We use negative binomial regression modeling to account for overdispersion in reports of transphobic discrimination. Then, using binary logistic regression modeling, we estimate the relationship between gender nonconformity and the three health‐harming behaviors: attempted suicide, drug/alcohol abuse, and smoking. In these analyses, we account for major and everyday transphobic discrimination in the final model to assess whether exposure to discrimination mediates the relationship, if any, between gender nonconformity and health‐harming behaviors. In particular, this final step will determine whether trans people who are gender nonconforming face more discrimination and, in turn, are at greater risk for engaging in the three unhealthy behaviors compared to their gender conforming counterparts. Although our cross‐sectional data do not allow us to assess causality or the temporal ordering, they offer an important starting point for examining the associations among these variables.

Results

Table 1 presents the descriptive statistics of independent and dependent variables for the NTDS sample (N = 4,115). Four percent of respondents completed the paper version of the NTDS. Overall, the sociodemographic profile of the NTDS sample mirrors that of other surveys of trans adults (Kuper et al. 2012; Rosser et al. 2007). On average, respondents have received some college (M = 2.53 [3 =  college degree]), but have an average annual household income slightly under $35,000 (logged M = 10.42). The majority of the sample is non‐Hispanic white (78%) and nonheterosexual (79%). Respondents report being read as transgender/gender nonconforming occasionally or sometimes, on average (M = 1.49).

Table 1

National Transgender Discrimination Survey Descriptive Statistics (N = 4,115)

	SAMPLE		GENDER
	M	(SD)	Trans Men (n = 1,601)	Trans Women (n = 2,514)
Survey Version (1 =  paper; 0 =  online)	.04	–	.02	.05***
Social Status
Race and Ethnicity
Non‐Hispanic White (1 =  yes)	.78	–	.75	.79***
Asian/Pacific Islander (1 =  yes)	.02	–	.02	.02
Non‐Hispanic Black (1 =  yes)	.03	–	.04	.03
Latina/o (1 =  yes)	.04	–	.04	.04
American Indian (1 =  yes)	.01	–	.01	.02
Multiracial (1 =  yes)	.11	–	.14	.10***
Income (logged)a	10.42	(1.04)	10.29	10.50***
Education (4 =  graduate school)	2.53	(.99)	2.63	2.47***
Age (in years)	37.77	(12.96)	31.03	42.07***
Sexual Minority (1 =  yes)	.79	–	.77	.80*
Perceived Gender Nonconformity (4 =  always)	1.49	(1.13)	1.34	1.58***
Transition Status
Medical Transition
Any Surgical Treatment (1 =  yes)	.43	–	.50	.38***
Hormones Only (1 =  yes)	.37	–	.23	.45***
No Medical Treatment (1 =  yes)	.20	–	.26	.17***
Social Transition (1 =  lives full time)	.74		.78	.71***
Self‐Reports of Transphobic Discrimination
Major Transphobic Discrimination (0–26 events)	3.20	(4.01)	3.05	3.29
Everyday Transphobic Discrimination (0–11 events)	2.47	(2.46)	2.73	2.31***
Health‐Harming Behaviors
Attempted Suicide (1 =  yes)	.44	–	.47	.42**
Drug/Alcohol Abuse (1 =  yes)	.27	–	.29	.26*
Smoking (1 =  yes)	.30	–	.34	.28***

*p < .05, **p < .01, ***p < .001 compared to trans men.

A logged annual household income of 10.46 is equivalent to $35,000.

Over two‐thirds of the sample report experiencing any exposure to major and everyday transphobic discrimination (70% and 71%, respectively) (analyses available upon request). Respondents, on average, report 3.20 types of events of major discrimination and 2.47 types of events of everyday discrimination. A substantial number of respondents report engaging in health‐harming behaviors, including 44% who have attempted suicide, 27% who have abused drugs/alcohol, and 30% who currently smoke. These rates of suicidality, drug/alcohol abuse, and smoking are substantially higher than those among the cisgender population, as well as among sexual minorities (CDC 2011, 2012; Cochran and Mays 2000).

Table 1 also presents the descriptive statistics by gender identity. There are a number of significant bivariate differences between transgender men (n = 1,601) and transgender women (n = 2,514). Trans women are significantly more likely than trans men to have completed the paper version of the NTDS (5% and 2%, respectively; p < .001). On average, trans men (M = 2.63) report significantly higher levels of education than trans women (M = 2.47); however, trans women report significantly higher levels of income than trans men (p < .001). Trans women are significantly older than trans men (M = 42.07 compared to M = 31.03, respectively). In terms of medical transition status, trans men are significantly more likely than trans women to have undergone surgery or no medical procedures to transition, while trans women are significantly more likely to have received hormones only (p < .001). Trans men are significantly more likely to report living full time in their desired gender than trans women (p < .001). Finally, trans women (M = 1.58 [2 =  sometimes]) are also significantly more likely than trans men (M = 1.34) to be perceived as gender nonconforming by others.9

Transphobic Discrimination

Table 2 presents the associations between perceived gender nonconformity and major discrimination (Models 1–3) and everyday discrimination (Models 4–6) using negative binomial regression modeling. Models 1 and 4 display the incidence risk ratios (IRRs) for transphobic discrimination on gender nonconformity at the bivariate level. Models 2 and 5 add respondents' medical and social transition status. Finally, Models 3 and 6 add controls for gender identity, race and ethnicity, income, education, age, and sexual identity.

Table 2

Incidence Risk Ratios (IRRs) for Self‐Reported Transphobic Discrimination on Gender Nonconformity (N = 4,115)

	Major Transphobic Discrimination			Everyday Transphobic Discrimination
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Gender Nonconformity	1.06**	1.14***	1.12***	1.13***	1.20***	1.20***
	(1.02–1.09)	(1.10–1.18)	(1.08–1.16)	(1.09–1.16)	(1.17–1.24)	(1.17–1.24)
Hormones Only		1.22***	1.27***		1.07	1.22***
		(1.08–1.37)	(1.13–1.43)		(.97–1.18)	(1.11–1.35)
Had Surgery		1.33***	1.55***, a		1.22***, a	1.42***, a
		(1.17–1.50)	(1.37–1.75)		(1.10–1.36)	(1.28–1.58)
Lives Full Time		1.69***	1.50***		1.53***	1.42***
		(1.52–1.88)	(1.35–1.66)		(1.40–1.67)	(1.30–1.54)
Trans Woman			1.28***			.98
			(1.18–1.40)			(.91–1.05)
Asian/Pacific Islander			.74*			.87
			(.56–.97)			(.69–1.08)
Black			1.00			.81*
			(.80–1.24)			(.68–.98)
Latina/o			1.25*, b			1.03c
			(1.03–1.52)			(.88–1.21)
American Indian			1.67**, b, c			1.30b, c
			(1.22–2.28)			(1.00–1.69)
Multiracial			1.61***, b, c, d			1.26***, b, c, d
			(1.43–1.80)			(1.14–1.38)
Income (Logged)			.80***			.92***
			(.77–.83)			(.89–.95)
Education			1.06**			1.06***
			(1.01–1.10)			(1.03–1.10)
Age			.99***			.99***
			(.98–.99)			(.98–.99)
Sexual Minority			1.01			1.11**
			(.92–1.11)			(1.03–1.20)
AIC	18,874.93	18,697.38	18,372.10	17,042.03	16,872.76	16,667.94
BIC	18,900.22	18,741.64	18,479.58	17,067.32	16,917.01	16,775.42

Exponentiated coefficients from negative binomial regression; 95% CIs in parentheses. *p < .05, **p < .01, ***p < .001. All models control for survey version. No medical procedures is the reference medical transition category. Non‐Hispanic whites are the reference racial/ethnic group.

Significantly differs from hormonal treatment only (p < .05).

Significantly differ from Asians/Pacific Islanders (p < .05).

Significantly differ from Blacks (p < .05).

Significantly differ from Latina/os (p < .05).

At the bivariate level, gender nonconformity is significantly associated with exposure to major transphobic discrimination (IRR: 1.06; CI: 1.02–1.09) in Model 1 and everyday transphobic discrimination (IRR: 1.13; CI: 1.09–1.16) in Model 4. That is, trans people who are more frequently read as transgender or gender nonconforming report facing more types of transphobic discrimination compared to those who are read as such less frequently. In Models 2 and 5, these patterns hold, net of medical and social transition statuses. Finally, in Models 3 and 6, gender nonconformity is still significantly and positively associated with exposure to major (IRR: 1.12; CI: 1.08–1.16) and everyday discrimination (IRR: 1.20; CI: 1.17–1.24), net of sociodemographic controls. These results suggest that transgender adults who are more gender nonconforming face significantly more types of both major and everyday discrimination than those who are less gender nonconforming.

There are significant differences in self‐reports of transphobic discrimination by medical and social transition statuses in Table 2 as well. For example, in Models 3 and 6, respondents who have undergone any surgical procedures to transition report facing the greatest number of events of discrimination, followed by those who have received hormones only, and then those who have received no medical treatment. In addition, respondents who live full time in their desired gender report significantly more major discrimination (OR: 1.50; CI: 1.35–1.66 [Model 3]) and everyday discrimination (OR: 1.42; CI: 1.30–1.54 [Model 6]) than those who do not. These patterns appear to operate in the opposite direction of what one might expect. Trans people who have undergone the most extensive medical treatment to transition (i.e., surgery) and those who live full time in their desired gender report the most exposure to discrimination. However, supplementary analyses suggest significant interactions between gender nonconformity and transition status, wherein the discrimination consequences of gender nonconformity are greatest for trans people who have received no medical treatment and those who do not live full time in their desired gender (available upon request).10

The patterns regarding social status differences in transphobic discrimination are also somewhat complex (Models 3 and 6). Regarding gender identity, trans women (IRR: 1.28; CI: 1.18–1.40) report significantly more types of major events of discrimination than trans men, although no significant gender difference exists for everyday discrimination. Some racial and ethnic minority groups report experiencing more discriminatory events than their non‐Hispanic white counterparts—in particular, multiracial (IRR: 1.61 [major] and 1.26 [everyday]), American Indian (IRR: 1.67 [major]), and Latina/o (IRR: 1.25 [major]) respondents. However, Asian/Pacific Islander (IRR: .74 [major]) and Black (IRR: .81 [everyday]) respondents report experiencing fewer discriminatory events than non‐Hispanic white trans people. A mixed pattern emerges for socioeconomic status differences in reports of discrimination. More highly educated (IRR: 1.06 for both) respondents report significantly more discrimination than those with lower levels of education. However, lower income respondents report more types of transphobic discrimination than higher income respondents (IRR: .80 [major] and .92 [everyday]). Young respondents report more types of events of major and everyday discrimination than older respondents (IRR: .99 for both). Finally, trans people who are sexual minorities (IRR: 1.10; CI: 1.03–1.20) report significantly more everyday (but not major) discrimination than heterosexual trans people. Thus, while self‐reports of transphobic discrimination vary by gender nonconformity and medical and social transition status, there are noteworthy sociodemographic differences, as well.

Health‐Harming Behaviors

The remaining analyses estimate the association between perceived gender nonconformity and attempted suicide (Table 3), drug/alcohol abuse (Table 4), and smoking (Table 5) using binary logistic regression modeling. For each table, Model 1 presents the odds for perceived gender nonconformity at the bivariate level. Model 2 adds respondents' medical and social transition status. Models 3 adds controls for gender identity, race and ethnicity, income, education, age, and sexual identity. Finally, Models 4 adds major and everyday discrimination.

Table 3

Odds Ratios for Attempted Suicide on Gender Nonconformity (N = 4,115)

	Model 1	Model 2	Model 3	Model 4
Gender Nonconformity	1.06*	1.08**	1.07*	1.02
	(1.00–1.12)	(1.02–1.15)	(1.01–1.14)	(.95–1.08)
Hormones Only		1.05	1.31**	1.19
		(.87–1.26)	(1.07–1.59)	(.97–1.46)
Had Surgery		.85a	1.26*	1.04
		(.69–1.03)	(1.01–1.56)	(.84–1.31)
Lives Full Time		1.74***	1.49***	1.32**
		(1.47–2.06)	(1.25–1.78)	(1.10–1.58)
Trans Woman			.96	.87
			(.83–1.11)	(.74–1.01)
Asian/Pacific Islander			1.14	1.29
			(.73–1.79)	(.82–2.04)
Black			.95	.97
			(.66–1.38)	(.67–1.42)
Latina/o			.99	.90
			(.71–1.39)	(.63–1.28)
American Indian			1.61	1.30
			(.92–2.84)	(.72–2.36)
Multiracial			1.54*** b c	1.27*
			(1.26–1.89)	(1.02–1.57)
Income (Logged)			.80***	.86***
			(.75–.85)	(.80–.93)
Education			.77***	.74***
			(.72–.83)	(.69–.80)
Age			.99***	.99*
			(.98–.99)	(.99–1.00)
Sexual Minority			1.13	1.12
			(.96–1.33)	(.95–1.32)
Major Discrimination				1.12***
				(1.09–1.15)
Everyday Discrimination				1.05*
				(1.01–1.09)
AIC	5,647.36	5,601.01	5,405.53	5,192.64
BIC	5,666.32	5,638.95	5,506.69	5,306.44

Exponentiated coefficients from binary logistic regression; 95% CIs in parentheses. *p < .05, **p < .01, ***p < .001. All models control for survey version. No medical procedures is the reference medical transition category. Non‐Hispanic whites are the reference racial/ethnic group. aSignificantly differs from hormonal treatment only (p < .05). bSignificantly differ from Blacks (p < .05). cSignificantly differ from Latina/os (p < .05).

Table 4

Odds Ratios for Drug/Alcohol Abuse on Gender Nonconformity (N = 4,115)

	Model 1	Model 2	Model 3	Model 4
Gender Nonconformity	1.12***	1.12***	1.10**	1.01
	(1.06–1.19)	(1.05–1.19)	(1.03–1.18)	(.94–1.08)
Hormones Only		1.17	1.39**	1.26*
		(.95–1.43)	(1.12–1.73)	(1.00–1.57)
Had Surgery		.92a	1.17a	.93a
		(.74–1.15)	(.93–1.49)	(.73–1.20)
Lives Full Time		1.17	1.02	.86
		(.97–1.40)	(.84–1.23)	(.70–1.04)
Trans Woman			.90	.84*
			(.76–1.06)	(.70–.99)
Asian/Pacific Islander			.48*	.50*
			(.26–.89)	(.27–.93)
Black			.84	.90
			(.56–1.26)	(.59–1.37)
Latina/o			1.12b	1.05b
			(.78–1.61)	(.72–1.52)
American Indian			.89	.66
			(.48–1.67)	(.34–1.29)
Multiracial			1.55*** bc	1.28* b
			(1.26–1.91)	(1.02–1.59)
Income (Logged)			.85***	.91*
			(.79–.91)	(.84–.98)
Education			.91*	.87***
			(.85–.99)	(.80–.94)
Age			.99**	1.00
			(.98–1.00)	(.99–1.00)
Sexual Minority			1.18	1.13
			(.98–1.41)	(.94–1.37)
Major Discrimination				1.07***
				(1.04–1.09)
Everyday Discrimination				1.17***
				(1.12–1.22)
AIC	4,785.83	4,780.29	4,699.15	4,448.93
BIC	4,804.79	4,818.22	4,800.31	4,562.73

Exponentiated coefficients from binary logistic regression; 95% CIs in parentheses. *p < .05, **p < .01, ***p < .001. All models control for survey version. No medical procedures is the reference medical transition category. Non‐Hispanic whites are the reference racial/ethnic group. aSignificantly differs from hormonal treatment only (p < .05). bSignificantly differ from Asians/Pacific Islanders (p < .05). cSignificantly differ from Blacks (p < .05).

Table 5

Odds Ratios for Smoking on Gender Nonconformity (N = 4,115)

	Model 1	Model 2	Model 3	Model 4
Gender Nonconformity	1.16***	1.10**	1.10**	1.05
	(1.09–1.23)	(1.03–1.17)	(1.03–1.18)	(.98–1.13)
Hormones Only		.82*	1.05	.98
		(.68–.99)	(.86–1.30)	(.80–1.21)
Had Surgery		.55***, a	.82a	.71**, a
		(.45–.68)	(.65–1.03)	(.57–.90)
Lives Full Time		1.20*	1.01	.92
		(1.00–1.42)	(.84–1.22)	(.76–1.11)
Trans Woman			.73***	.68***
			(.62–.85)	(.58–.80)
Asian/Pacific Islander			1.10	1.17
			(.68–1.77)	(.72–1.91)
Black			1.49*	1.56*
			(1.03–2.17)	(1.07–2.29)
Latina/o			.99	.94b
			(.69–1.41)	(.65–1.35)
American Indian			1.24	1.06
			(.70–2.21)	(.59–1.92)
Multiracial			1.21	1.05
			(.97–1.49)	(.85–1.31)
Income (Logged)			.83***	.87***
			(.77–.89)	(.81–.94)
Education			.66***	.64***
			(.61–.71)	(.59–.69)
Age			.99	1.00
			(.99–1.00)	(.99–1.01)
Sexual Minority			1.02	1.00
			(.86–1.21)	(.84–1.19)
Major Discrimination				1.05***
				(1.03–1.08)
Everyday Discrimination				1.07**
				(1.03–1.11)
AIC	5,007.81	4,976.97	4,763.64	4,677.17
BIC	5,026.77	5,014.91	4,864.80	4,790.97

Exponentiated coefficients from binary logistic regression; 95% CIs in parentheses. *p < .05, **p < .01, ***p < .001. All models control for survey version. No medical procedures is the reference medical transition category. Non‐Hispanic whites are the reference racial/ethnic group.

Significantly differs from hormonal treatment only (p < .05). bSignificantly differ from Blacks (p < .05).

In Model 1 of Table 3, gender nonconformity is significantly and positively associated with attempted suicide (OR: 1.06, CI: 1.00–1.12). Trans people who are more frequently read as transgender or gender nonconforming are more likely to report having attempted suicide than those read as such less frequently. Perceived gender nonconformity is significantly associated with attempted suicide even upon controlling for medical transition status (Model 2) and social statuses (Model 3). However, gender nonconformity becomes nonsignificant in Model 4, which adds controls for major and everyday discrimination. These patterns suggest that gender‐nonconforming trans people are more likely to have attempted suicide than their gender conforming counterparts—until accounting for transphobic discrimination.

Indeed, in Model 4, self‐reported major discrimination (OR: 1.12; CI: 1.09–1.15) and everyday discrimination (OR: 1.05; CI: 1.01–1.09) are significantly and positively associated with attempted suicide. Trans people who report more types of events of transphobic discrimination are more likely to have attempted suicide than those who report fewer or no discriminatory events. Further, post hoc Sobel (1982) tests suggest that major discrimination (Z = 5.43; p < .001) and everyday discrimination (Z = 8.57; p < .001) significantly mediate the relationship between gender nonconformity and attempted suicide (available upon request). That is, self‐reported major and everyday discrimination partially explain the relationship between gender nonconformity and attempted suicide.

In Tables 4 and 5, gender nonconformity is significantly and positively associated with the likelihood of drug/alcohol abuse and smoking, respectively. Across Models 1–3, trans people who are frequently read as transgender or gender nonconforming are significantly more likely to report abusing drugs/alcohol and smoking than those who are less frequently read as such. Similarly, major and everyday transphobic discrimination are associated with these unhealthy behaviors. However, the associations between gender nonconformity and these two health‐harming behaviors become nonsignificant in Models 4, which controls for transphobic discrimination. Post hoc Sobel tests suggest that exposure to transphobic discrimination partially explains the association between gender nonconformity and drug/alcohol abuse (Z = 5.45 [major] and Z = 9.70 [everyday]; p < .001), and between gender nonconformity and smoking (Z = 4.87 [major] and Z = 7.09 [everyday]; p < .001).11 Taken together, these findings suggest that the more frequently trans people are read as transgender or gender nonconforming, the more types of major and everyday discriminatory events they face and, in turn, the more likely they are to engage in these health‐harming behaviors.

Tables 3, 4, 5 also demonstrate significant associations between transition status and social status with the three health‐harming behaviors. However, there are few consistent differences in these health behaviors by transition status. There is only one consistent pattern for social transition status: trans people who live full time in their desired gender are significantly more likely to report having attempted suicide than those who do not (OR: 1.32; CI: 1.10–1.58 [Model 4]). Trans people who have received hormones are significantly more likely to report abusing drugs/alcohol than those who have undergone surgery or no medical procedures. In addition, trans people who have undergone surgery are significantly less likely to smoke than those who have received hormones only or no medical treatment. In sum, the patterns for the effect of transition status on health‐harming behaviors are mixed.

Trans women are less likely than trans men to smoke (OR: .68; CI: .58–.80 [Model 4]). The associations between socioeconomic status and the three health‐harming behaviors are consistent: higher income and more highly educated respondents are less likely than those with lower incomes and less education to report engaging in these unhealthy behaviors, respectively. Older respondents are less likely than younger respondents to have attempted suicide and to abuse drugs/alcohol (Model 3 only). Similar to self‐reports of discrimination, the patterns for race and ethnicity are mixed. Compared to their non‐Hispanic white counterparts, multiracial trans people are significantly more likely to have attempted suicide (OR: 1.27; CI: 1.02–1.57 [Model 4]) and to abuse drugs/alcohol (OR: 1.28; CI: 1.02–1.59 [Model 4]), and Blacks are significantly more likely to smoke (OR: 1.56; CI: 1.07–2.29 [Model 4]). In contrast, Asians and Pacific Islanders are significantly less likely than their non‐Hispanic white counterparts to abuse drugs/alcohol (OR: .50; CI: .27–.93 [Model 4]).

Discussion

In this article, we sought to highlight the important role of gender nonconformity in the social experiences and well‐being of transgender people. Using data from the NTDS, we assessed whether gender nonconformity—a visible and known marker of one's stigmatized status—heightens trans people's exposure to discrimination and, in turn, their likelihood of engaging in health‐harming behaviors. In doing so, we offered a starting point for considering the role that stigma visibility plays in minority stress processes. Consistent with prior research on trans people, the majority of transgender adults (70%) in the NTDS sample reported experiencing transphobic discrimination (Clements et al. 1999; Factor and Rothblum 2008a; Lombardi et al. 2002).

Our study offers two major findings. First, consistent with prior research, transgender people who face more everyday and major discrimination are more likely to engage in health‐harming behaviors (i.e., attempted suicide, drug/alcohol abuse, and smoking) (Clements‐Noelle et al. 2006; Pascoe and Richman 2009). These findings offer further evidence that the minority stress framework (Meyer 1995) can be used to understand the aspects of the social environment that shape and constrain the health status of the transgender population—in this case, widespread exposure to transphobic discrimination.

Our second, more central finding is that gender nonconformity may heighten trans people's exposure to discrimination and health‐harming behaviors. Gender nonconforming trans adults reported more events of major and everyday transphobic discrimination than their gender conforming counterparts. That is, the more frequently trans people are read as transgender or gender nonconforming by others, the more they are subject to major and day‐to‐day discriminatory treatment. In addition, gender nonconformity predicted greater likelihood of attempted suicide, drug/alcohol abuse, and smoking—a relationship that was partially mediated by major and everyday discrimination. In other words, one plausible interpretation of these patterns is that gender nonconforming individuals are at greater risk for poor health as a consequence of facing more transphobic discrimination.

A secondary finding concerns the association between transition status and discriminatory treatment. Our findings regarding medical transitioning (i.e., none, hormonal therapy only, and any surgical treatment) and social transitioning (i.e., living full time in one's desired gender) may seem paradoxical. Trans people who live full time in their desired gender and those who have received surgical treatment reported the most transphobic discrimination. In two ways, these findings may reflect the fact that our data are cross‐sectional. First, it is possible that trans people who have undergone surgery to transition experienced more discrimination prior to transitioning—a decision made, in part, to be read as their desired gender to avoid any future discrimination. Alternatively, because being read as transgender and gender nonconforming can vary at different points during a person's life, it is possible that those who have lived longer in their desired gender have had more time and opportunities to face discrimination. Certainly, even trans people who transition still report minority stressors, especially if their legal documents do not reflect their present gender identity (Levitt and Ippolito 2014). Thus, these findings may not be nearly as counterintuitive as they initially appear. Indeed, our findings are consistent with prior studies that examine the relationship between transition status and transphobic discrimination (Bradford et al. 2013).

Another secondary finding concerns sociodemographic differences in trans people's exposure to discriminatory treatment. Our results are mixed, but we do find that some individuals who belong to multiple disadvantaged groups (i.e., multiracial and lower income trans people) faced more transphobic discrimination and were more likely to engage in health‐harming behaviors. We also found important gender differences in exposure to major discrimination: trans men reported that they face fewer events of major discrimination than trans women in a number of arenas beyond the workplace. These findings also speak to prior discussions in qualitative studies about whether trans men benefit from male privilege (C. Connell 2010; Dozier 2005; Meier and Labuski 2013; Schilt 2006). Future work should consider accounting for other forms of discrimination (e.g., race‐ and class based) in order to more clearly highlight social status differences in discrimination and health among trans people (Grollman 2014).

Our study offers important insights to recent theoretical debates about the “doing” and “undoing” gender (Deutsch 2007; Risman et al. 2012; West and Zimmerman 2009), including an empirical assessment of the social consequences of gender nonconformity that transgender people face (Vidal‐Ortiz 2009). Indeed, we find that transgender people who are read as gender nonconforming faced more discrimination than their gender conforming counterparts. In other words, our findings are consistent with assertions by West and Zimmerman (2009) that accountability structures are still in place, which serve to punish those who deviate from binary understandings of gender. Our findings suggest that gender still remains highly salient in the lives of transgender people. In particular, the social costs of gender nonconformity for trans people remain high, and the consequences are far reaching. However, rather than it being trans people's responsibility to conform to society's gender expectations, we note that the onus is ultimately upon society to expand the possibilities of gender so that transgender individuals may lead more livable lives (Butler 2004; Lorber 2005). Indeed, as others suggest, it is more productive to consider the ways in which misrecognition is an inherent social process, not an individual failing of transgender persons (Pfeffer 2014).

Additionally, our findings make major theoretical contributions to the minority stress model, primarily through highlighting the role that stigma visibility plays in minority stress processes. Within the context of our study, gender nonconformity serves as a form of stigma visibility, wherein transgender people who are more frequently read as trans and gender nonconforming experienced more discrimination and worse health. In documenting this pattern, we also fill key gaps in the literature through illustrating the critical role that gender nonconformity plays in minority stress processes for transgender people (Gordon and Meyer 2007). These results suggest that it would be fruitful to integrate the concept of stigma visibility into minority stress research in order to document within‐group variation in exposure to discrimination among marginalized populations. Our work also builds upon prior studies that show that stigma visibility makes one more vulnerable to discriminatory treatment and subsequent poor health (Frable et al. 1997; Stutterheim et al. 2011).

Our study faces a few methodological limitations. One major data limitation is that we do not know whether or why people intentionally try to conceal their status as transgender, or the internal costs of doing so (see Meier and Labuski 2013 for a review). Studies suggest that some transgender people report that they feel pressured to be more gender conforming, partially as a way to avoid job discrimination and physical violence (Gagne and Tewksbury 1998). However, other research suggests that the main reason trans people adopt gender conforming expressions of gender is to affirm their “true self” (Mason‐Schrock 1996). Both motivations are likely valid; it is thus important to acknowledge that attempting to avoid discrimination only partially explains why some trans people may desire to be gender conforming.

A second major limitation is the cross‐sectional nature of the NTDS. Without longitudinal data, we are unable to make causal inferences about our variables, especially the measure of lifetime suicidality. Instead, our data allow us to show correlations and associations between variables. Regarding the temporal ordering of the variables, we also cannot conclude that discrimination precedes participation in health‐harming behaviors. However, prior longitudinal research on discrimination confirms that reports of discrimination are associated with later poor health, but poor health does not predict later reports of discrimination (Williams and Mohammed 2009). In addition, other studies find that transphobic discrimination is linked with suicide attempts, even when controlling for individual risk factors (e.g., age, depression, substance abuse, and histories of sexual assault) (Clements‐Nolle et al. 2006). Thus, our findings are at least consistent with prior evidence that discrimination and other minority stressors exacerbate LGBT people's likelihood of engaging in unhealthy behaviors.

A related limitation is our reliance on self‐report measures—a common approach for survey research. A number of scholars have raised concerns, in particular, regarding perceptions of discrimination, which may be over‐ or understated due to hypervigilance or minimization, respectively (Williams and Mohammed 2009). However, there is evidence that self‐reports reflect actual experiences of discrimination (NRC 2004). It is also important to note that perceptions of discrimination have consistently been linked to observable health problems, regardless of their basis in “objective” events. Additionally, our measure of gender nonconformity is based upon self‐reports; more information is needed on how outsiders view transgender people's level of gender conformity. One possible solution for future research is to collect data on siblings' evaluations of how gender nonconforming they perceive their transgender brothers and sisters to be (Factor and Rothblum 2008a).

A number of other limitations should also be noted. First, the NTDS is a large, yet nonrepresentative survey of transgender adults; thus, we are unable to make conclusions regarding the generalizability of our findings to the larger transgender population in the United States. Nevertheless, the NTDS was able to avoid some sampling biases by relying on a range of sampling strategies to capture diverse segments of the transgender population. Second, we considered a limited range of health outcomes due to the small amount of available indicators of respondents' health status in the NTDS. Finally, the NTDS did not ask respondents how often they experience discrimination, thus we cannot assess whether there are differences among trans people in the frequency of exposure to discrimination—a factor that is particularly important for experiences of everyday discrimination (Grollman 2014). Despite these limitations, the NTDS remains the highest quality and largest data set available to test the associations among gender nonconformity, transphobic discrimination, and health.

Our findings make major contributions to understanding the social consequences of gender nonconformity for transgender people. Our analyses suggest that gender nonconformity—as a visible and known marker of one's stigmatized status—plays an important role in trans people's experiences with discrimination and poor health. Specifically, we find that trans people who are frequently read by others as transgender and gender nonconforming faced more major and day‐to‐day events of discrimination and, in turn, were more likely to engage in health‐harming behaviors compared to more gender conforming trans people. As such, the results suggest important variations in the experiences of transgender people. More generally, our findings suggest that researchers would benefit from an integration of the concept of stigma visibility into the minority stress model, and that it would be fruitful to consider external factors that exacerbate minority groups' exposure to discrimination and its associated health consequences.