Educational Attainment of Transgender Adults: Does the Timing of Transgender Identity Milestones Matter?

Abstract

Adolescence is a difficult life stage in which to navigate a transgender identity, yet adolescence plays a key role in shaping educational trajectories. While transgender-related stigma and victimization within secondary schools persists, the social climate in which transgender adolescents navigate their identity has changed over time. Analyzing data from the National Transgender Discrimination Survey, a national, non-probability sample of U.S. transgender adults, we address the following research questions: 1) Is experiencing transgender identity milestones in adolescence associated with educational attainment?; 2) Does this association vary by birth cohort? We find that those who first experienced transgender identity milestones in adolescence attained less education than those who first experienced milestones in other life stages. This association is larger among younger birth cohorts, pointing to the adolescent years as a particularly difficult time to navigate a transgender identity, even in the midst of increased transgender awareness and resources within schools and society.

1. Introduction

Normative expectations of gender and sexuality are heightened in adolescence (Pascoe 2007; Eder, Evans, and Parker 1995), a key period of identity development and preparation for adulthood. Those first recognizing themselves as different due to gender, first identifying as transgender¹, or first living as transgender in adolescence often experience additional stress during an already stressful developmental stage. Yet there is little research examining the association between experiencing transgender identity milestones in adolescence and later life outcomes. Transgender students in middle schools and high schools face transgender-related harassment and discrimination at alarming rates (McGuire and Conover-Williams 2010) and have described attending school as one of the most traumatic experiences of growing up (Grossman and D’Augelli 2006). Young people who first experience transgender identity milestones during adolescence, within compulsory cisnormative secondary schools where peer networks are highly salient, often confront stigma and stress (Russell and Toomey 2012). Adolescents may also have less developed cognitive resources for coping with stress and access to fewer identity-based resources, relative to adults. Consequently, individuals who first experience transgender identity milestones in adolescence may experience more disruption in educational trajectories and lower levels of educational attainment than those who experience milestones in other life stages (Barrett, Pollack, and Tilden 2002; Ueno, Roach, and Pena-Talamantes 2013).

Activism and scholarship by lesbian, gay, bisexual, transgender, and queer (LGBTQ) rights advocates have contributed to dramatic societal shifts in awareness and perceptions of LGBTQ people (Schilt and Lagos 2017). The lived experiences of sexual and gender minorities are thus vastly different today than in the past, and while numerically small, the transgender population is growing in both size and visibility in the U.S. (Flores et al. 2016; Hartocollis 2015; Minter 2012; Zucker et al. 2008). Transgender identities and gender variance are more openly acknowledged and discussed in society today, and transgender individuals are identifying, coming out, and living as transgender at younger ages today than in the past (Beemyn and Rankin 2011a; Hendricks and Testa 2012; Vanderburgh 2009). Much of the awareness and advocacy aimed at transgender issues has occurred in schools, with the emergence of the Gay, Lesbian, Straight Education Network (GLSEN), Gay Straight Alliances (GSAs), school anti-bullying policies (E. J. Meyer 2015; Kull, Kosciw, and Greytak 2015), and transgender rights advocacy in public schools (Hirschfeld and Apuzzo 2016). This is significant given the critical role schools play, particularly middle schools and high schools, in conveying norms about gender and sexuality and preparing youth for post-secondary education. Additionally, increased advocacy on college campuses in recent decades means that transgender students who do enter college are likely to find more supportive spaces and policies (Beemyn 2015; Beemyn and Rankin 2011b). Given these changes, experiencing transgender identity milestones for the first time in adolescence may impact educational attainment differently today than in the past.

Integrating the minority stress and life course perspectives and analyzing data from a national non-probability sample of transgender individuals from the National Transgender Discrimination Survey, we address two primary research questions: 1) Is experiencing transgender identity milestones for the first time in adolescence associated with educational attainment?; and 2) Does this association vary by birth cohort? While results may not be generalizable to the broader U.S. transgender population, results have implications for policies and programs aimed at transgender youth in schools and add to our knowledge of the transgender population, one of the least studied and understood segments of the LGBTQ population.

2. Background

2.1. Minority stress and educational attainment

The minority stress paradigm has been developed to address the additional stresses that accrue to minorities due to higher rates of stigma and discrimination (Lick, Durso, and Johnson 2013; Meyer 2003) and has more recently been applied to understanding the outcomes of transgender individuals (Hendricks and Testa 2012; I. H. Meyer 2015). According to the minority stress theory, a transgender identity is a socially stigmatized status associated with greater exposure to prejudice and discrimination (Meyer 2003). The stress associated with stigma, prejudice, and discrimination due to one’s transgender status has been linked to higher rates of psychological distress experienced both through external processes, including actual experiences of rejection and discrimination, and through internal processes, such as perceived rejection and expectations of being stereotyped or discriminated against (Bockting et al. 2013; Goffman 1963; Seelman, Woodford and Nicolazzo 2017). However, stigmatized identities can also provide access to identity-based resources used to combat such risks (I. H. Meyer 2015). Experiences of stress and distress, as well as the risk and resilience demonstrated by individuals confronting minority stress, can have long-term consequences for status attainment processes such as educational attainment (Pearson and Wilkinson 2016).

Transgender youth face gender-related harassment and discrimination at alarming rates: over 80% of transgender students report some form of harassment at school, and transgender youth are likely to report their first experiences of gender-related victimization at school (D’Augelli, Grossman, and Starks 2006). LGBTQ youth consistently report higher rates of victimization than their non-LGBTQ peers (Greytak, Kosciw, and Diaz 2009; Kosciw, Palmer, and Kull 2015), with transgender students reporting more harassment and victimization than their cisgender LGBQ peers (Kosciw, Diaz, and Greytak 2008). Transgender students frequently report experiencing homophobic language and derogatory comments about gender expression from both students and staff (Greytak et al. 2009; Clements-Nolle, Marx, and Katz, 2006; Grossman and D’Augelli, 2007). This harassment may continue into college, as many transgender college students also report harassment, derogatory remarks, exclusion, and violence based on their gender identity (Griner et al. 2017; Rankin et al. 2010).

The experience of minority stress due to transgender-related stigma can create feelings of distress and isolation for transgender students, impacting their ability to engage in and succeed in school (Greytak et al. 2009; Grossman and D’Augelli 2006; Rosenberg 2002). As a result of minority stress, transgender youth are often at greater risk for emotional distress (D’Augelli et al. 2006; Rivers 2001; Rosenberg 2002; Seelman et al. 2017) and risky behaviors (Garofalo et al. 2006), which may impact academic performance (Kosciw et al. 2015) and the ability to complete high school (Grossman and D’Augelli 2006; Sausa 2005). In addition, unwelcoming school environments may reduce feelings of connectedness or attachment to school and teachers (Poteat and Espelage 2007; Rostosky et al. 2003; Kosciw et al. 2015) leading students to disengage from academics (Akerlof and Kranton 2002; Finn 1989). Transgender college students may disengage from supportive spaces and leave college altogether if confronted by a chilly campus climate and perceived lack of safety (Rankin et al. 2010). Given the importance of belonging and connectedness for educational success (Finn 1989; Libbey 2004; Tinto 1993), minority stress may disrupt the educational trajectories of transgender students.

Of course, not all sexual and gender minorities perform poorly or attain less education in response to transphobic and heteronormative contexts (Watson and Russell 2014; Russell 2005), and not all social contexts are equally transphobic and heteronormative (Hatzenbuehler et al. 2014; Kosciw et al. 2015; Wilkinson and Pearson 2009). Experiences of minority stress and disrupted educational trajectories among transgender individuals vary by social factors (Kosciw et al. 2015). Below we consider how the timing of transgender identity milestones is associated with educational attainment, and how this association may vary across multiple dimensions of life course timing.

2.2. A life course perspective on minority stress

An important element of minority stress models is the identify formation and coming out process, as identity milestones may lead to both risk and the capacity to be resilient in the face of risk (Meyer 2015; Pollock and Eyre 2012). A life course perspective emphasizes the importance of timing, duration, and social context of milestones (Elder, Johnson, and Crosnoe 2003) and points to three dimensions of timing that shape life course trajectories: chronological timing, social timing, and historical timing. Chronological timing (age) emphasizes developmental processes associated with aging, such as puberty; social timing emphasizes the social construction of age categories and the social roles and statuses accompanying these categories; and historical timing emphasizes the role of historical processes and macro-level social changes that impact life course trajectories. The timing of transgender identity milestones, both within an individual’s life course as well as within historical time, may impact exposure and responses to minority stress, with implications for educational attainment.

2.2.1. Transgender identity milestones

The various experiences associated with transgender identity development are not monolithic; rather, they vary from one person to another and may occur non-linearly, simultaneously, or not at all (Beemyn and Rankin 2011a; Bilodeau 2005; Diamond, Pardo, and Butterworth 2011). Yet various models of sexual and gender identity development (Devor 2004; D’Augelli 1994; Heistand and Levitt 2005; Pollock and Eyre 2012) suggest at least three common milestones associated with the development of a transgender identity: recognition of self as different due to gender, identification of self as transgender, and first living as transgender. Recognition of difference occurs as individuals experience feelings of discomfort with their assigned gender identity and as self-authenticity conflicts with social expectations of gendered behavior. This often occurs early in life as individuals are socialized to enact an assigned binary gender within families, schools, and other social institutions and as individuals feel a sense of conflict with gendered expectations (Beemyn and Rankin 2011a). Feelings of difference and the potential confusion, isolation, and shame are often followed by an identification process in which individuals come to learn about transgender and non-binary gender identities and, for female-to-male and male-to female identified individuals, about the possibilities of transition (Beemyn and Rankin 2011a). The process of identity development often involves testing different identities, including LGBQ identities (Bilodeau 2005), and adopting one that feels authentic and manageable given one’s social context and access to resources (Bilodeau 2005; Devor 2004; Levitt and Ippolito 2014). Putting a name and label to feelings of gender confusion and difference by identifying with a transgender identity can provide greater understanding, pride, and access to resources, but it can also provoke feelings of stigma and stress associated with a minority identity (Denny 2004; Diamond et al. 2011; Meyer 2015).

Acceptance of a transgender identity prompts decisions about how to integrate one’s identity into daily life, or how to express one’s transgender identity in various contexts. Similar to other identity milestones, there is great variability in the experience of living as transgender, yet across the spectrum of transgender identities a core experience involves the intentional enactment of a gender (binary or non-binary) different from that assigned at birth through, for example, changes in dress, speech, pronoun usage, or name. Living as transgender often, but not always, involves coming out as transgender in various contexts such as work, school, and family (Devor 2004) and may take many forms: for some it may involve the decision to be read as the opposite gender and to begin transitioning medically (Bilodeau 2005; Diamond et al. 2011), while for others it may involve the complete rejection of the gender binary and an attempt to transgress the gender binary (Bornstein 1994; Girchick 2008).

As transgender identities become more visible in society, individuals have begun to experience these identity milestones at earlier ages (Diamond et al. 2011), and contextual factors may hasten or delay these milestones (D’Augelli 1994; Bilodeau 2005; Levitt and Ippolito 2014). Yet the implications of the timing of milestones for life course outcomes such as educational attainment remain unclear.

2.2.2. Social and chronological timing of milestones

Given adolescence is often a stressful yet critical period in the life course (Arnett 1999) in which feelings of difference associated with gender are heightened (Pascoe 2007), experiencing transgender identity milestones in adolescence may have a greater impact on educational attainment than experiencing them in childhood or in adulthood. Minority stressors, including potentially negative responses from peers, school personnel, and family to a non-normative gender identity and presentation may exacerbate stress already present in adolescence, a time when not “fitting in” is particularly salient (Anderson and Teicher 2008; Crosnoe 2011). It is during adolescence that peer groups become a central socializing agent for youth, within which youth spend a significant portion of their time building relationships and developing social identities (Eccles and Roeser 2011). Homophobic attitudes, which are often closely tied to transphobic attitudes (Hill and Willoughby 2005), and rates of harassment and bullying are highest in early adolescence (Horn 2006; Poteat, Espelage, and Koenig 2009; Smyser and Reiss 2002; Unnever and Cornell 2004) as youth’s social identities and sense of self are increasingly tied to interactions within peer groups (Hill and Lynch 1983; Eccles and Roeser 2011) and when grades and course-taking become central to educational trajectories (Adelman 2006).

For transgender youth, adolescence is a developmental period in which bodily changes and the emergence of secondary sex characteristics heighten visible gender non-conformity and associated gender dysphoria, or negative feelings towards one’s gendered body (Vanderburgh 2007). Adolescents have had less time than adults to learn adaptive coping strategies (Anderson and Teicner 2008) and may lack access to LGBTQ resources (Fields 2004; Greytak et al. 2009). As a result, adolescents coping with minority stress due to gender-related stigma may be more likely, relative to adults, to respond to minority stressors in maladaptive ways (e.g., avoidance via dropping out of school; low self-esteem; social anxiety) that are disruptive to educational achievement and attainment (Grossman and D’Augelli 2006; Toomey et al. 2010). In addition, students who experience these milestones in high school and enter college as a transgender or gender non-conforming student often face additional hurdles as they navigate highly gendered structures (i.e. student housing, public facilities, legal records) and unwelcoming classroom climates (Beemyn and Brauer 2015; Garvey and Rankin 2015; Pryor 2015; Rankin et al. 2010). Alternatively, first experiencing transgender identity milestones in adulthood, when trans-specific resources and adaptive coping strategies are more accessible, may lessen experiences of and impact of minority stressors, reducing the presence of maladaptive coping strategies that interfere with educational attainment.

Relative to those first identifying and first living as transgender in adolescence, experiencing transgender identity milestones in childhood may be associated with characteristics and resources that promote educational attainment. Those who first identify as transgender in childhood rather than in adolescence are more likely to have significant adults in their lives who intervene early to initiate transition and to provide support (Ehrensaft 2013; Kennedy and Helen 2010), more often come from two-parent families of higher socioeconomic status (Vanderburgh 2007), and report more positive self-perceptions and mental health (Kennedy and Helen 2010; Vanderburgh 2009). Those experiencing these milestones in childhood are also doing so before puberty, which may enable transgender individuals to be read as cisgender by peers in adolescence and thus receive less negative reaction from peers (Pusch 2005). Research has found that transgender students who “pass” and fit into the gender binary are less likely to be victimized by peers (Grossman et al. 2006). These youth may also find strength and support in transgressing the gender binary at an early developmental age (Ehrensaft 2013).

Taken together, we expect that:

Hypothesis 1: First experiencing transgender milestones in adolescence, in comparison to in adulthood or in childhood, will be negatively associated with educational attainment.

2.2.3. Historical timing of milestones

In addition to the social and chronological timing of transgender identity milestones, historical timing of milestones may impact exposure and response to minority stress and educational attainment. Historical timing addresses how historical changes, including macro-level institutional and cultural changes, impact life course trajectories and outcomes. Historical timing is often measured by changes across birth cohorts as each birth cohort encounters a unique historical context. Transphobia is still widespread within larger society and within schools, as transgender youth are at greater risk for negative educational outcomes compared to cisgender LGB youth (Kosciw et al. 2008), yet transgender social justice issues are garnering greater attention and provoking legal and social changes (Hirschfeld and Apuzzo 2016). Transgender issues have also become more prominent in the media, raising awareness of transgender students within schools (Case and Meier 2014; Schulman 2013). Though there is still progress to be made, a growing number of states, colleges, and universities include gender identity and/or expression in their non-discrimination policies (Beemyn 2015, Human Rights Campaign 2017). Such changes have been accompanied by greater acceptance of transgender individuals within LGBQ communities, greater diversity of transgender identities and experiences, and more access to transgender medical services (Denny 2004).

The experiences of transgender persons who identify and live as transgender for the first time in adolescence thus likely varies by birth cohort. Millennial transgender individuals (those born after 1981) experienced adolescence during the 1990s as many schools began creating and implementing policies and programs to protect and promote the well-being of gender and sexual minority students (Fetner and Kush 2008; Kosciw et al. 2015; Kull et al. 2015). Indeed, the 1990s is often considered a transitional decade, leading to increased awareness, attention, and resources devoted to LGBQ students in the 2000s (Meyer 2015). Though often ignoring the “T,” LGBQ services positively impact transgender students (Greytak, Kosciw, and Boesen 2013). Generation X individuals (those born between 1965 and 1981) experienced adolescence before this transitional period. Baby Boomers (born between 1946 and 1964) and members of the Silent Generation (born before 1946) experienced adolescence prior to the post-Stonewall gay identity movement and prior to the first wave of transgender activism, which emerged in the 1970s (Escoffier 1985). Among younger birth cohorts, identifying and living as transgender may provide opportunities for increased self-esteem, sense of purpose, and access to social support and other resources from the LGBTQ and medical communities (Denny 2004; Kosciw et al. 2015; Levitt and Ippolito 2014; Pinto, Melendez, and Spector 2008; Riggle et al. 2011), with implications for long-term educational attainment.

Taken together, we expect that:

Hypothesis 2: Any observed negative association between first experiencing transgender identity milestones in adolescence and educational attainment will be weaker among younger birth cohorts, relative to older birth cohorts.

3. Data and methods

We used data from the National Transgender Discrimination Survey (NTDS), a non-probability sample of transgender adults fielded by the National Center for Transgender Equality and the National Gay and Lesbian Task Force between September 2008 and March 2009. While non-probability samples limit researchers’ ability to make generalizations about populations and raise issues of selection bias, these approaches are often necessary to find “hidden populations,” or populations for which no sampling frame exists and for which privacy concerns related to stigmatized identities or behaviors are of concern (Heckathorn 1997). Traditional methods of survey design are often inefficient at collecting data from hidden populations given they often result in small samples, providing little opportunity for quantitative analysis.

While limited due to generalizability, the NTDS is the first and only large-scale, national sample of transgender people in the U.S. that includes questions about the timing of transgender identity milestones, birth cohort, and educational attainment. Using venue-based sampling and snowball sampling the NTDS was announced through a network of more than 800 transgender-led or transgender-serving community-based organizations in the U.S. and through 150 active online community listservs. The organizations used to recruit respondents were from across the U.S., serving a variety of purposes and demographic groups, including but not limited to college students, specific race-ethnic and religious groups, and those receiving transgender related health care. In addition, two thousand paper surveys were made available to organizations that serve hard-to-reach populations—including rural, homeless, and low-income transgender people. The NTDS sampling design resulted in a large and diverse sample of transgender individuals from across all 50 U.S. states and territories.

The final NTDS sample included 6,456 respondents who identified as transgender or gender non-conforming, or “people whose gender identity or expression is different, at least part of the time, from the sex assigned to them at birth” (Grant et al. 2010, p. 183). While NTDS sampled those who identified as transgender, gender non-conforming (GNC)², or cross-dressers, our analytic sample includes only those who identified as transgender or gender non-conforming (N=5,547). We further limited our sample to respondents who were 25 or older at the time of the survey (N=4,226) as those younger than 25 may not have had adequate time to attain a four-year college degree. Finally, we excluded those who were missing (<3%) on any of our analytic variables, resulting in a final analytic sample of 3,796 transgender and GNC-identified respondents ages 25-98.

3.1. Dependent variable: educational attainment

We measure educational attainment using the survey question asking, “What is the highest degree or level of school you have completed? If you are currently enrolled, please mark the previous grade or highest degree received.” We collapsed the eleven response categories into a dichotomous variable where “1” represents attaining a four-year degree or higher (bachelor’s degree, master’s degree, professional degree, doctorate degree) and “0” indicates attaining less than a four-year college degree. In analyses not shown (available upon request), we used a continuous measure of educational attainment, years of schooling; these results revealed no substantive differences.

3.2. Independent variable: Timing of transgender identity milestones

Respondents were asked to identify the age they first recognized they were different due to their gender; age they first recognized their transgender/GNC identity; and the age they first began living as transgender/GNC. For each of these three transgender identity milestones, we constructed three mutually exclusive dummy variables representing 1) first experienced milestone in childhood (younger than twelve), 2) first experienced milestone in adolescence (between twelve and seventeen), and 3) first experienced milestone in adulthood (eighteen or older). Given our focus on adolescence, experiencing a milestone in adolescence was used as the reference category in all models. For age first began living as transgender/GNC, respondents were asked two questions: 1) age first began living part-time as transgender/GNC and 2) age first began living full-time as transgender/GNC. For the majority of respondents, we used age first lived part-time; for respondents missing on age first lived part-time, we used age first lived full-time when available.

3.3. Independent variable: Birth cohort

We used respondents’ self-reported age and survey year to construct three dichotomous birth cohort variables representing Millennials (those born after 1981), Generation Xers (those born between 1965 and 1981), and Baby Boomers or pre-Baby Boomers (those born before 1965). We combined Baby Boomers and pre-Baby Boomers (sometimes referred to as the Silent or Great Generation) into one group (Baby Boomers+) given the small number of pre-Baby Boomers (n=145) in our sample and the small number of pre-Baby Boomers who first lived as transgender in childhood or adolescence. Baby Boomers+ were used as the reference category given our interest in the outcomes of younger birth cohorts relative to older birth cohorts.

In our sample, Millennials were between the ages of 25-26, Generation Xers were 27-43, and Baby Boomers+ were 44-98 at the time of the survey. Approximately 12% (n=442) of our sample respondents were Millennials, 51% (n=1,928) were Generation Xers, and 38% (n=1,426) were Baby Boomers+ (see Table 1). Although not shown, the majority of Baby Boomers+ (90%) were Baby Boomers, with only 10% of the Baby Boomers+ group born before 1946 (pre-Baby Boomers). Among the 10% of pre-Baby Boomers in the sample (n=145), none first began living as transgender in childhood, and only three (2.1%) first began living as transgender in adolescence. Substantive results did not change when Baby Boomers and pre-Baby Boomers were included as distinct categories.

Table 1.

Means and Proportions of Analytic Variables by Birth Cohort

Birth Cohort	All Cohorts		Millennials		Gen Xers		Baby Boomers+
Variables	Mean/Prop	95% CI	Mean/Prop	95% CI	Mean/Prop	95% CI	Mean/Prop	95% CI
Educational Attainment
College Degree or Higher (0/1)	0.50	[0.48, 0.51]	0.41	[0.37, 0.46]	0.50	[0.48, 0.52]	0.52	[0.50, 0.55]
Some College (0/1)	0.41	[0.39, 0.42]	0.44	[0.39, 0.48]	0.40	[0.38, 0.43]	0.41	[0.38, 0.43]
High School (0/1)	0.07	[0.06, 0.08]	0.12	[0.09, 0.15]	0.07	[0.06, 0.08]	0.06	[0.04, 0.07]
Less than High School (0/1)	0.02	[0.02, 0.03]	0.03	[0.02, 0.05]	0.03	[0.02, 0.03]	0.01	[0.01, 0.02]
Age and Life Course Stage First Recognized Different Due to Gender
Age First Recognized	8.03	[7.83, 8.23]	8.64	[8.15, 9.12]	7.89	[7.65, 8.14]	8.03	[7.64, 8.43]
First Recognized In Childhood	0.80	[0.79, 0.82]	0.72	[0.68, 0.76]	0.79	[0.77, 0.81]	0.84	[0.82, 0.86]
First Recognized In Adolescence	0.13	[0.12, 0.14]	0.21	[0.17, 0.25]	0.14	[0.12, 0.16]	0.10	[0.08, 0.11]
First Recognized In Adulthood	0.07	[0.06, 0.07]	0.07	[0.05, 0.09]	0.07	[0.06, 0.08]	0.06	[0.05, 0.07]
Age and Life Course Stage First Identified as Transgender
Age of Identification	18.04	[17.67, 18.41]	15.58	[15.07, 16.10]	17.02	[16.64, 17.40]	20.19	[19.37, 21.00]
First Identified in Childhood	0.30	[0.28, 0.31]	0.20	[0.16, 0.23]	0.26	[0.24, 0.28]	0.37	[0.35, 0.40]
First Identified in Adolescence	0.27	[0.25, 0.28]	0.37	[0.32, 0.41]	0.26	[0.24, 0.28]	0.24	[0.21, 0.26]
First Identified in Adulthood	0.44	[0.42, 0.45]	0.44	[0.39, 0.48]	0.47	[0.45, 0.50]	0.39	[0.37, 0.42]
Age and Life Course Stage First Lived as Transgender
Age Began Living as Transgender	28.73	[28.31, 29.15]	18.33	[17.94, 18.73]	23.77	[23.42, 24.12]	38.66	[37.92, 39.41]
First Lived in Childhood	0.05	[0.04, 0.06]	0.05	[0.03, 0.08]	0.05	[0.04, 0.06]	0.05	[0.04, 0.06]
First Lived in Adolescence	0.13	[0.12, 0.14]	0.29	[0.25, 0.33]	0.14	[0.12, 0.16]	0.06	[0.05, 0.07]
First Lived in Adulthood	0.82	[0.81, 0.84]	0.66	[0.61, 0.70]	0.81	[0.79, 0.83]	0.89	[0.88, 0.91]
Covariates
Non-Hispanic White	0.78	[0.77, 0.80]	0.71	[0.66, 0.75]	0.74	[0.72, 0.76]	0.87	[0.85, 0.89]
Non-Hispanic Asian	0.03	[0.02, 0.03]	0.05	[0.03, 0.06]	0.04	[0.03, 0.05]	0.01	[0.01, 0.02]
Other Non-White	0.19	[0.17, 0.20]	0.25	[0.21, 0.29]	0.22	[0.20, 0.24]	0.12	[0.10, 0.13]
Age at Survey	40.06	[39.68, 40.44]	25.50	[25.45, 25.55]	33.62	[33.39, 33.84]	53.29	[52.94, 53.64]
Assigned Female Sex at Birth	0.38	[0.37, 0.40]	0.66	[0.62, 0.71]	0.50	[0.48, 0.52]	0.14	[0.12, 0.16]
Non-Binary Gender Identity	0.08	[0.07, 0.09]	0.17	[0.14, 0.21]	0.11	[0.09, 0.12]	0.02	[0.02, 0.03]
Heterosexual	0.20	[0.19, 0.22]	0.17	[0.13, 0.20]	0.20	[0.18, 0.22]	0.22	[0.20, 0.24]
New England	0.09	[0.08, 0.10]	0.12	[0.09, 0.15]	0.09	[0.07, 0.10]	0.08	[0.07, 0.09]
Mid-Atlantic	0.20	[0.19, 0.22]	0.21	[0.17, 0.25]	0.20	[0.19, 0.22]	0.20	[0.18, 0.22]
South	0.19	[0.18, 0.20]	0.17	[0.13, 0.20]	0.18	[0.16, 0.20]	0.21	[0.19, 0.23]
Mid-West	0.21	[0.20, 0.22]	0.21	[0.17, 0.25]	0.21	[0.19, 0.23]	0.21	[0.19, 0.23]
West	0.17	[0.16, 0.19]	0.17	[0.13, 0.20]	0.17	[0.15, 0.19]	0.18	[0.16, 0.20]
California	0.13	[0.12, 0.15]	0.12	[0.09, 0.15]	0.15	[0.13, 0.17]	0.12	[0.10, 0.14]
N	3796		442		1928		1426

Source: National Transgender Discrimination Survey

3.4. Control variables

In all multivariate models we controlled for socio-demographic variables associated with timing of transgender identity milestones and educational attainment. Because our sample included those with a non-binary gender identity and those with a binary gender identity, we included a dummy variable indicating non-binary gender identity (1= yes; 0 = no). Research suggests non-binary, gender non-conforming individuals are at heightened risk of discrimination that may be negatively associated with educational attainment (Miller and Grollman 2015) and that the process of transgender identity development may be different for those with non-binary gender identities (Beemyn and Rankin 2011). Assigned female at birth is a dummy variable indicating respondent was assigned a female sex/gender at birth, with assigned male sex/gender at birth as the reference. We included this as a control given the higher achievement and attainment of females, relative to males, among recent birth cohorts (DiPrete and Buchman 2013), and the role of gender socialization in educational outcomes (Morris 2012; Sadker, Sadker, and Zittleman 2009). In addition, research suggests that gender non-conformit displayed by males is more strictly policed than it is among females, with male-to-different transgender youth experiencing higher rates of gender-related discrimination and victimization (D’Augelli et al. 2006; Kosciw et al. 2008).

Age is a continuous variable measuring respondents’ age at the time of the survey and ranges from 18 to 98. Age is an important control given older individuals have had more time to complete a college degree and to experience transgender identity milestones. Race/ethnicity was measured with a set of mutually exclusive dichotomous variables, including non-Hispanic white (the reference), non-Hispanic Asian or Pacific Islander, and non-Hispanic other minority (black or African American, American Indian or Alaska Native, Hispanic or Latino, or multiracial). We separated Asian and other non-white individuals given the positive association between Asian racial identity and educational attainment and the negative association between other non-white identities (e.g., black, American Indian or Alaska Native, Hispanic/Latino, multiracial) and educational attainment (Kao and Thompson 2003). We controlled for current sexual orientation given the negative association between non-heterosexual identity and educational achievement and attainment (Pearson, Wilkinson, and Muller 2007; Pearson and Wilkinson 2016) and given the complex interplay between sexual orientation and gender identity (Dozier 2005; Levitt and Ippolito 2014). To measure sexual orientation, we created a dummy representing heterosexual with all other sexual identities included in the reference category. Finally, we controlled for respondents’ geographic region at the time of the survey given local context is associated with risk of victimization, access to resources, and educational attainment (Kosciw, Greytak, and Diaz 2009). Region is measured with a set of mutually exclusive dichotomous variables representing states in New England, the Mid-Atlantic, the South, the Mid-West (the reference), the West (including Alaska and Hawaii), and California.

3.5. Analytic approach

We first examined descriptive statistics for all analytic variables, across birth cohorts. Next, we examined bivariate associations between timing of transgender identity milestones and four-year college degree attainment and then examined these bivariate associations across birth cohorts. Finally, we ran binary logistic regression models to examine 1) whether timing of transgender identity milestones is associated with four-year college degree attainment, controlling for covariates and previous milestones, and 2) whether these associations vary across birth cohorts. For each transgender identity milestone, we started with a main effect model to test the general association between timing of milestone and college degree attainment, controlling for covariates. We then added interactions between timing of milestone and birth cohort in the next model. We included timing of antecedent milestones into models for later milestones because the antecedent milestones might be a confounder in the association between later milestones and educational attainment. For example, the timing of first feeling different due to one’s gender might be associated with both the timing of first identifying and living as transgender as well as educational attainment. Based on estimates from multivariate models, we calculated odds ratios and average marginal predicted probabilities (Mood 2010; Williams 2012) of four-year degree attainment for each milestone and birth cohort group. We ran all regression models in Stata and report 95% confidence intervals derived from robust standard errors. We do not report results of significance tests in Table 1–3, as tests of statistical significance cannot be interpreted in the conventional way given that the NTDS sample is not a population-based random sample. Our findings, therefore, cannot be assumed to be generalizable to the larger U.S. transgender population.

Table 3.

Odds Ratios from Logistic Regression Models Predicting 4-Year College Degree Attainment

	Model 1		Model 2		Model 3		Model 4		Model 5		Model 6
	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
First Felt Different Due to Gender (Ref: In Adolescence)
Childhood	1.17	[0.96, 1.42]	0.94	[0.65, 1.36]	1.17	[0.95, 1.44]	0.92	[0.63, 1.35]	1.14	[0.92, 1.40]	0.92	[0.63, 1.35]
Adulthood	1.44	[1.05, 1.98]	0.84	[0.48, 1.46]	1.25	[0.91, 1.73]	0.87	[0.49, 1.51]	1.20	[0.87, 1.66]	0.86	[0.49, 1.51]
First Identified as Transgender (Ref: Adolescence)
Childhood					1.39	[1.16, 1.67]	0.99	[0.74, 1.32]	1.31	[1.09, 1.59]	0.99	[0.74, 1.33]
In Adulthood					1.54	[1.30, 1.83]	0.92	[0.69, 1.23]	1.32	[1.10, 1.57]	0.88	[0.65, 1.18]
First Lived as Transgender (Ref: Adolescence)
Childhood									2.02	[1.40, 2.91]	1.40	[0.70, 2.78]
Adulthood									1.93	[1.52, 2.44]	1.69	[1.04, 2.72]
Birth Cohort (Ref: Baby Boomers)
Millennial	1.17	[0.78, 1.75]	0.78	[0.41, 1.48]	1.19	[0.79, 1.79]	0.40	[0.19, 0.83]	1.27	[0.85, 1.92]	0.43	[0.17, 1.06]
Gen X	1.47	[1.11, 1.95]	1.15	[0.70, 1.89]	1.45	[1.10, 1.93]	0.69	[0.39, 1.20]	1.47	[1.11, 1.95]	0.73	[0.35, 1.51]
First Felt Different x Birth Cohort (Ref: Adolescence, Baby Boomers+)
Childhood x Millennial			1.58	[0.86, 2.90]			1.90	[1.00, 3.61]			1.84	[0.96, 3.51]
Childhood x Gen X			1.29	[0.82, 2.03]			1.30	[0.81, 2.08]			1.23	[0.77, 1.98]
Adulthood x Millennial			2.69	[0.96, 7.53]			1.82	[0.64, 5.20]			1.79	[0.63, 5.05]
Adulthood x Gen X			2.12	[1.05, 4.29]			1.68	[0.82, 3.42]			1.60	[0.78, 3.27]
First Identified x Birth Cohort (Ref: Adolescence, Baby Boomers+)
Childhood x Millennial							1.38	[0.71, 2.68]			1.18	[0.57, 2.43]
Childhood x Gen X							1.76	[1.19, 2.61]			1.64	[1.10, 2.44]
Adulthood x Millennial							2.82	[1.62, 4.90]			2.12	[1.15, 3.92]
Adulthood x Gen X							2.09	[1.43, 3.03]			1.88	[1.28, 2.76]
First Lived x Birth Cohort (Ref: Adolescence, Baby Boomers
Childhood x Millennial											1.72	[0.52, 5.66]
Childhood x Gen X											1.58	[0.68, 3.69]
Adulthood x Millennial											1.20	[0.57, 2.52]
Adulthood x Gen X											1.05	[0.59, 1.84]
Covariates
Age	1.04	[1.03 1.05]	1.04	[1.03, 1.05]	1.04	[1.03, 1.05]	1.04	[1.03, 1.05]	1.04	[1.02, 1.05]	1.04	[1.02, 1.05]
Non-Binary Gender Identity	1.67	[1.29, 2.17]	1.67	[1.28, 2.17]	1.67	[1.29, 2.17]	1.70	[1.31, 2.21]	1.66	[1.28, 2.15]	1.67	[1.29, 2.18]
Assigned Female at Birth	2.34	[2.00, 2.17]	2.35	[2.00, 2.74]	2.23	[1.90, 2.62]	2.22	[1.89, 2.61]	2.32	[1.97, 2.74]	2.30	[1.95, 2.72]
Heterosexual	0.76	[0.64, 0.90]	0.77	[0.65, 0.91]	0.77	[0.65, 0.91]	0.78	[0.65, 0.92]	0.76	[0.64, 0.90]	0.77	[0.65, 0.91]
Race-Ethnicity (Ref: Non-Hispanic White)
Asian	1.14	[0.76, 1.70]	1.14	[0.76, 1.71]	1.14	[0.76, 1.70]	1.18	[0.78, 1.79]	1.16	[0.77, 1.75]	1.19	[0.79, 1.81]
Other Non-White	0.55	[0.46, 0.66]	0.55	[0.46, 0.66]	0.56	[0.46, 0.66]	0.56	[0.47, 0.67]	0.57	[0.48, 0.69]	0.57	[0.48, 0.69]
Region (Ref: Mid-West)
Mid-Atlantic	1.47	[1.19, 1.80]	1.47	[1.20, 1.81]	1.48	[1.21, 1.82]	1.48	[1.20, 1.82]	1.50	[1.22, 1.84]	1.50	[1.21, 1.84]
New England	1.47	[1.12, 1.93]	1.47	[1.12, 1.93]	1.47	[1.12, 1.93]	1.49	[1.13, 1.96]	1.47	[1.12, 1.93]	1.48	[1.13, 1.95]
California	1.34	[1.07, 1.69]	1.35	[1.07, 1.70]	1.35	[1.07, 1.70]	1.35	[1.07, 1.70]	1.36	[1.08, 1.72]	1.36	[1.07, 1.71]
West	1.11	[0.89, 1.37]	1.10	[0.89, 1.37]	1.11	[0.89, 1.37]	1.11	[0.90, 1.38]	1.11	[0.89, 1.38]	1.11	[0.89, 1.38]
South	0.89	[0.72, 1.10]	0.88	[0.71, 1.09]	0.90	[0.73, 1.11]	0.89	[0.72, 1.11]	0.91	[0.74, 1.13]	0.90	[0.73, 1.12]
Intercept	0.10	[0.05, 0.20]	0.12	[0.05, 0.26]	0.08	[0.04, 0.16]	0.14	[0.06, 0.31]	0.05	[0.02, 0.11]	0.09	[0.04, 0.22]
N	3796

Source: National Transgender Discrimination Survey

4. Results

4.1. Descriptive and bivariate results

Table 1 presents means and proportions of analytic variables across birth cohorts. Millennials in our sample are less likely to have attained a four-year college degree than those in older birth cohorts: 41% of Millennials, 50% of Generation Xers, and 52% of Baby Boomers+ had attained a four-year college degree at the time of the survey. While educational attainment in the U.S. population has increased over time (Ryan and Bauman 2016), older birth cohorts have had more time to complete a four-year degree. The high levels of education among older adults in our sample may also be a result of the sampling bias, as adults with higher levels of education may have greater awareness of transgender identities and greater access to transgender organizations, and they may therefore be more likely to appear in the NTDS sample.

Results in Table 1 suggest that respondents in younger birth cohorts first identified as transgender and first lived as transgender at younger ages than respondents in older birth cohorts. Millennials first identified as transgender at an average age of 15.6, compared to an average age of 17 among Generation X respondents and 20 among Baby Boomers+. Similarly, Millennials in our sample first lived as transgender at an average age of 18, compared to an average age of 24 among Generation Xers and 39 among Baby Boomers+. Related, a greater proportion of Millennial and Gen X respondents experienced these milestones in adolescence. The average age respondents first recognized themselves as different due to gender did not vary across birth cohort, however, with all cohorts in our sample experiencing this milestone around age eight. We suspect changes in timing of first identifying and first living as transgender, but not in first recognizing difference due to gender, is a result of increased social awareness of transgender identities over time.

Table 2 presents proportion attaining four-year college degree by timing of transgender identity milestones, across birth cohorts. Among Millennials and Gen Xers in our sample, a greater proportion of respondents who first recognized difference due to gender in adulthood attained a four-year degree, relative to those who first recognized this difference in adolescence. While 61% of Millennials who first recognized difference due to gender in adulthood attained a four-year degree, only 36% of Millennials who first recognized difference due to gender in adolescence attained a four-year degree. Similarly, while 62% of Gen Xers who first recognized difference due to gender in adulthood attained a four-year degree, only 48% of Gen Xers who first recognized this difference in adolescence attained a four-year degree. We see similar patterns among Millennials and Gen Xers when examining the association between first identified as transgender in adulthood (relative to in adolescence) and attaining a four-year degree. These associations between timing of milestones – timing of first recognized self as different and first identified as transgender – and four-year degree attainment, are not, however, observed among Baby Boomers+ in our sample. Finally, among all birth cohorts, a greater proportion of respondents who first lived as transgender in adulthood had attained a four-year degree, relative to those who first lived as transgender in adolescence. This association is, however, larger among Millennials and Gen Xers than among Baby Boomers+. Overall in our sample, respondents who experienced transgender milestones in adolescence have the lowest likelihood of attaining a four-year degree at the time of the survey, relative to those who first experienced milestones in adulthood. Finally, the strongest associations between timing of milestones and four-year degree attainment are observed among younger, not older, birth cohorts

Table 2.

Proportion with College Degree or Higher by Birth Cohort and Timing of Transgender Identity Milestones

Birth Cohort	All Cohorts		Millennials		Gen Xers		Baby Boomers+
Variables	Prop	95% CI	Prop	95% CI	Prop	95% CI	Prop	95% CI
Life Stage First Recognized Different Due to Gender
Childhood	0.50	[0.48, 0.51]	0.41	[0.36, 0.46]	0.49	[0.47, 0.52]	0.52	[0.49, 0.55]
Adolescence	0.48	[0.43, 0.52]	0.36	[0.26, 0.46]	0.48	[0.42, 0.54]	0.54	[0.46, 0.63]
Adulthood	0.59	[0.53, 0.65]	0.61	[0.44, 0.79]	0.62	[0.54, 0.71]	0.53	[0.43, 0.64]
Life Stage First Identified as Transgender
Childhood	0.49	[0.46, 0.52]	0.32	[0.22, 0.42]	0.49	[0.45, 0.53]	0.52	[0.48, 0.56]
Adolescence	0.41	[0.38, 0.44]	0.30	[0.23, 0.37]	0.37	[0.32, 0.41]	0.52	[0.46, 0.57]
Adulthood	0.56	[0.54, 0.59]	0.55	[0.48, 0.62]	0.58	[0.55, 0.61]	0.53	[0.49, 0.57]
Life Stage First Lived as Transgender
Childhood	0.56	[0.49, 0.63]	0.50	[0.30, 0.70]	0.61	[0.51, 0.70]	0.52	[0.40, 0.64]
Adolescence	0.32	[0.28, 0.36]	0.23	[0.16, 0.31]	0.34	[0.28, 0.39]	0.39	[0.29, 0.50]
Adulthood	0.52	[0.51, 0.54]	0.49	[0.43, 0.54]	0.52	[0.50, 0.55]	0.53	[0.51, 0.56]
N	3796		442		1928		1426

Source: National Transgender Discrimination Survey

4.2. Multivariate logistic regression results

Next we discuss results from binary logistic regression models predicting four-year college degree attainment focusing on estimated effects of timing of transgender identity milestones, birth cohort, and the interactions between timing of milestones and birth cohort (Table 3). Models 1-2 report results for timing of first recognized difference due to gender; Models 3-4 report results for timing of first identified as transgender, controlling for antecedent milestones; and Models 5-6 report results for timing of first lived as transgender, controlling for antecedent milestones.

After controlling for socio-demographic covariates, first recognizing difference due to gender in adolescence, relative to in childhood or in adulthood, is negatively associated with the likelihood of attaining a four-year college degree (Model 1, OR = 1.17 for childhood vs. adolescence; OR = 1.44 for adulthood vs. adolescence). Figure 1a presents average predicted probabilities of four-year degree attainment by timing of first recognized difference using estimates from Model 1: the predicted probability of four-year degree attainment among those who first recognized difference in adolescence is .46 (95% CI [.42, .50]), compared to .50 (95% CI [.48, .52]) among those who first recognized difference in childhood and .55 (95% CI [.49, .61]) among those who first recognized difference in adulthood.

Figure 1a.

Adjusted predictions of college completion by timing of first recognized self as different due to gender. Average predicted probabilities with 95% CIs derived from Model 1, Table 3.

Estimates from Model 2 indicate the association between timing of first recognized difference due to gender and four-year degree attainment varies by birth cohort in our sample. To illustrate this interaction, Figure 1b presents average predicted probabilities of four-year degree attainment by timing of first recognized difference due to gender, across birth cohorts, based on estimates from Model 2: among Millennials, the predicted probability of four-year degree attainment among those who first recognized difference in adolescence is .41 (95% CI [.32, .53]), compared to .50 (95% CI [.43, .57]) among those who first recognized difference in childhood and .59 (95% CI [.41, .77]) among those who first recognized difference in adulthood. Among Gen Xers, those who first recognized difference due to gender in adulthood also have a higher predicted probability of four-year degree attainment relative to those who first recognized difference in adolescence, yet the difference is smaller in magnitude. Among Baby Boomers+, those who first recognized difference in adulthood do not have a greater probability of attaining a four-year degree relative to those who first recognized difference in adolescence.

Figure 1b.

Adjusted predictions of college completion by timing of first recognized self as different due to gender, by birth cohort. Average predicted probabilities with 95% CIs derived from Model 2, Table 3.

In our sample, first identifying as transgender in adolescence, relative to in childhood or in adulthood, is also negatively associated with four-year degree attainment (Model 3, OR = 1.39 for adolescence vs. childhood; OR = 1.54 for adulthood vs. adolescence) even after controlling for socio-demographic covariates and antecedent milestones. Figure 2a presents average predicted probabilities of four-year degree attainment by timing of first identified as transgender using estimates from Model 3: the predicted probability of four-year degree attainment among those who first identified as transgender in adolescence is .43 (95% CI [.40, .46]), compared to a predicted probability of .51 (95% CI [.48, .54]) among those who first identified in childhood and .53 (95% CI [.51, .56]) among those who first identified as transgender in adulthood.

Figure 2a.

Adjusted predictions of college completion by timing of first identified as transgender. Average predicted probabilities with 95% CIs derived from Model 3, Table 3.

Estimates from Model 4 indicate the association between timing of first identifying as transgender and four-year degree attainment varies by birth cohort in our sample. Figure 2b presents average predicted probabilities of four-year degree attainment by timing of first identified as transgender, across birth cohorts, based on estimates from Model 4: among Millennials, the predicted probability of four-year degree attainment among those who first identified as transgender in adolescence is .38 (95% CI [.29, .47]), compared to a predicted probability of .45 (95% CI [.33, .57]) among those who first identified as transgender in childhood and .59 (95% CI [.51, .68]) among those who first identified as transgender in adulthood. Among Gen Xers, those who first identified as transgender in adolescence also have a lower predicted probability of four-year degree attainment (.43) relative to those who first identified as transgender in adulthood (.58). Among Baby Boomers+, those who first identified as transgender in adolescence do not have a lower probability of attaining a four-year degree relative to Baby Boomers+ who first identified as transgender in childhood or adulthood.

Figure 2b.

Adjusted predictions of college completion by timing of first identified as transgender, by birth cohort. Average predicted probabilities with 95% CIs derived from Model 4, Table 3.

Finally, after controlling for socio-demographic covariates and antecedent milestones, there remains a negative association between first lived as transgender in adolescence, relative to in childhood or in adulthood, and four-year degree attainment (Model 5, OR = 2.02 for childhood vs. adolescence; OR = 1.93 for adulthood vs. adolescence). Figure 3a presents average predicted probabilities of four-year degree attainment by timing of first lived as transgender using estimates from Model 5: the predicted probability of four-year degree attainment among those who first lived as transgender in adolescence is .37 (95% CI [.32, .42]), compared to a predicted probability of .53 (95% CI [.46, .60]) among those who first lived as transgender in childhood and .52 (95% CI [.50, .53]) among those who first lived as transgender in adulthood.

Figure 3a.

Adjusted predictions of college completion by timing of first lived as transgender. Average predicted probabilities with 95% CIs derived from Model 5, Table 3.

Estimated interactions between timing of first lived as transgender and birth cohort are in the same direction yet smaller in magnitude, relative to interactions from previously discussed timing by birth cohort interactions. Figure 3b presents average predicted probabilities of four-year degree attainment by timing of first lived as transgender, across birth cohorts, based on estimates from Model 6: among Millennials, the predicted probability of four-year degree attainment among those who first lived as transgender in adolescence is .37 (95% CI [.25, .48]), compared to a predicted probability of .56 (95% CI [.37, .75]) among those who first lived as transgender in childhood and .52 (95% CI [.45, 60]) among those first lived as transgender in adulthood. Among Gen Xers, those who first lived as transgender in adolescence also have a lower predicted probability of four-year degree attainment (.42) relative to those who first lived as transgender in adulthood (.55) and first lived as transgender in childhood (.60). Similarly, among Baby Boomers+, those who first lived as transgender in adolescence have a lower predicted probability of four-year degree attainment (.35) relative to those who first lived as transgender in adulthood (.47) and to those who first lived as transgender in childhood (.42). These results suggest that birth cohort plays less of a moderating role in the association between timing of first lived as transgender and four-year degree attainment relative to the associations between timing of other transgender identity milestones (first recognized difference due to gender; first identified as transgender) and four-year degree attainment.

Figure 3b.

Adjusted predictions of college completion by timing of first lived as transgender, by birth cohort. Average predicted probabilities with 95% CIs derived from Model 6, Table 3.

5. Discussion

In this study, we analyzed one of the very first and most comprehensive non-probability samples of transgender people in the U.S. to assess how the timing of transgender identity milestones is associated with educational attainment across birth cohorts—Millennials, Generation Xers, and Baby Boomers/Silent Generation. Integrating minority stress and life course theories, we argue that the social, chronological, and historical timing of milestones is important to consider when examining life course trajectories of transgender individuals. We focused primarily on the experience of transgender identity milestones in adolescence, a pivotal life stage in the educational attainment trajectory (Adelman 2006). Adolescence is a particularly stressful and critical life stage in which gender identities are highly salient (Crosnoe 2011; Pascoe 2007). Stressors associated with transgender identity milestones may exacerbate stressors already present in adolescence in ways that could have long-term consequences for educational trajectories. Because of increased awareness and resources available to gender and sexual minorities in secondary and post-secondary schools today (Beemyn 2015; Meyer E. 2015), we incorporated historical timing by assessing the association between timing of transgender identity milestones and educational attainment across birth cohorts. While data limitations prevent us from determining causality or generalizing results to the larger U.S. transgender population, results should inform future research exploring the experiences and educational outcomes of transgender individuals.

5.1. Does the timing of transgender identity milestones matter?

Minority stress theory suggests that risks and resources associated with a stigmatized identity are associated with educational trajectories (Meyer 2003; Bockting et al. 2013), while life course theory suggests that the timing of exposure to minority stress may moderate this association (Ueno et al. 2013). Our analysis of data from the National Transgender Discrimination Survey found that transgender persons who identified as transgender and lived as transgender for the first time in adolescence were less likely to attain a four-year degree than their counterparts who first experienced these milestones in adulthood or in childhood. Experiencing milestones in adolescence may be particularly difficult given the narrowly and rigidly defined expectations of doing gender and the increased pressure for “fitting in” during adolescence (Crosnoe 2011). Adults experiencing similar milestones may have greater ability to choose the contexts in which they interact and may have greater access to transgender-affirming and supportive resources that promote resilience (Beemyn and Rankin 2011b). Given social, psychological, and biological differences between adults and adolescents, adults may be better able to cope with and navigate a stigmatized identity (Anderson and Teichner 2008; Blakemore and Mills 2014) and may have greater ability to use their transgender identity as a resource (Nicolazzo 2016). Transgender individuals who first identify or live as transgender in childhood often have greater access to family and institutional supports to overcome cisnormative school contexts and may benefit from experiencing milestones prior to puberty (Vanderburgh 2007).

An alternative interpretation of findings is that a college education itself shapes the timing of developmental milestones among transgender individuals. College campuses and exposure to gender diversity may promote transgender identity development, as has been suggested in the literature on sexual minorities (Annes and Redlin 2012; Zemsky 2004), and research suggests that many transgender individuals identify college as the context in which they first recognize, accept, and express their authentic gender identity (Beemyn and Rankin 2011b). It is also possible that young people who recognize they are different due to gender may delay accepting or expressing a transgender identity until they can do so in a less hostile environment that provides supportive spaces and resources, which may be truer of college campuses relative to other contexts. Consistent with this idea, non-population based samples of transgender adults suggest transgender individuals have higher levels of educational attainment (Grant et al. 2010), and research on men reporting same-sex attractions and behaviors has found that those with higher levels of education are more likely to identify with a gay identity (Barrett and Pollack 2005). However, the challenges transgender students face on college campuses are unique relative to those of sexual minority students (McKinney 2005): individual and institutional discrimination against transgender college students is still pervasive, and many college campuses lack adequate resources for transgender students (Bilodeau 2005; Beemyn and Rankin 2016).

5.2. Birth cohort differences

Importantly, we find that the association between first experiencing transgender identity milestones in adolescence and educational attainment varies across birth cohorts in the NTDS sample. We expected variation across birth cohorts, yet the direction of the association was contrary to what we predicted: among younger birth cohorts, those who first recognized difference due to gender, first identified as transgender, or first lived as transgender in adolescence were less likely to attain a four-year degree than those who experienced these milestones in adulthood; however, this pattern was not observed among the oldest birth cohorts in the NTDS sample.

Increased attention and awareness to transgender students’ rights has increased the visibility of transgender students and transgender identities, potentially increasing the minority stressors associated with adolescent transgender identity development and disrupting educational trajectories among younger birth cohorts. In various parts of the country, there have been intense negative reaction from conservative groups and parents to the formation of GSAs as well as to schools’ support of transgender bathroom rights (Healy and Perez-Pena 2016; Schilt and Westbrook 2015; Wilson 2016). The Lieutenant Governor of Texas recently stated that providing transgender students access to bathrooms that fit their preferred sex/gender “is the biggest issue facing families and schools in America since prayer was taken out of public schools” (Healy and Perez-Pena 2016, para. 4). The Trump administration’s roll back of Title IX protections for transgender students (Kreighbaum 2017) represents another example of the backlash against transgender rights occurring at the national level.

An alternative perspective to understanding the unexpected birth cohort differences observed is variation in selection processes across birth cohorts. Given the dramatic paradigm shift that has occurred in society’s understanding of transgender identities, members of younger birth cohorts, relative to older birth cohorts, have a broader range of transgender identities to choose from and expanded options for how to “do” transgender (Denny 2004), and these options are available at younger ages today than in the past. Such changes have led to shifting characteristics of the transgender population, which we observe in our descriptive statistics and control for in multivariate models: Millennial respondents, relative to those of older birth cohorts, are more likely to identify as non-binary/GNC, to be assigned a female sex at birth, and to identify as non-white.

It is also possible that transgender individuals of older birth cohorts who first experienced transgender milestones in adolescence are unique from those of younger birth cohorts in unobservable ways that impact trends in the NTDS sample. In order to overcome increased barriers and fewer options for gender variant persons, transgender respondents of older birth cohorts who first felt different, first identified, or first lived as trans in adolescence may have been particularly resilient or may have had access to additional supports, either within their families or communities. Moreover, we have seen dramatic changes across generations in rates of college attendance and completion (Ryan and Bauman 2016) as well as gender gaps in educational attainment (DiPrete and Buchman 2013), both of which suggest that access to college and its social meaning vary by birth cohort. Finally, it is possible that differences by birth cohort could be a result of the sampling approach used by NTDS. For example, a college education may have had a greater impact on selection into the NTDS sample among older respondents: college-educated transgender adults may be more likely to identify as transgender and access transgender organizations and resources compared to those with lower levels of education.

5.3. Limitations

Our study has several limitations. First, although the National Transgender Discrimination Survey provides one of the very first national datasets to study transgender people in the U.S., the sample is not a nationally representative, population-based sample. Instead, the recruitment process was based on non-probability sampling techniques that targeted transgender people, including the use of venue-based sampling and snowball sampling. While the sponsors of NTDS attempted to identify and recruit a sample representative of the U.S. transgender population, this approach raises the possibility of sampling bias, bias that may or may not be associated with birth cohort, timing of milestones and educational attainment. For example, it may be that among older birth cohorts of transgender individuals those who experienced milestones in adolescence and have lower educational attainment have less access to transgender-led or transgender-serving organizations and are thus less likely to appear in the NTDS sample. Future research, preferably using population-based generalizable data, should further explore variation in transgender developmental milestones across birth cohorts in order to better understand how these processes are linked to life course trajectories such as educational attainment.

NTDS is also cross-sectional and retrospective, which raises additional limitations. NTDS respondents from older birth cohorts had more time than respondents from younger birth cohorts to attain a four-year degree. However, results from supplemental analyses not shown (but available upon request) that limited respondents to those thirty and older and thirty-five and older were substantively similar. Additional supplementary analysis also revealed that, within birth cohorts, those experiencing milestones in adulthood were not significantly older than those first experiencing milestones in adolescence and thus did not have additional time to complete a four-year degree. Finally, although we intentionally limited predictor variables to those that theoretically precede four-year degree attainment, we are unable to determine if educational attainment leads to transgender milestones in adulthood, rather than vice versa, and it is possible that educational attainment impacts perceptions of the timing of milestones given the retrospective nature of the data.

While we work from a minority stress and life course perspective to build our research hypotheses, we are unable to directly assess causal mechanisms related to stigma and minority stress or to determine the timing of minority stressors that lead to lower educational attainment. While the NTDS includes rich measures of perceived discrimination, we are unable to determine when in the life course perceived discrimination occurred, either before or after adolescence or before or after college degree attainment. Future research should assess the role of victimization and perceived discrimination in the educational attainment of transgender respondents. Finally, we are unable to measure the characteristics of respondents’ families, schools, or communities of origin, which has implications for causal mechanisms and potential selection processes across birth cohorts. To fully understand how the timing of transgender identity milestones impacts educational attainment across birth cohorts, population-based longitudinal datasets with additional information on family, school, and community contexts, across the life course, are required.

We acknowledge the analysis of quantitative data limits our ability to share the voices of transgender people, as we do not hear the first-hand accounts of their lived experiences. Yet our theoretical framework draws heavily on previous qualitative work, including in-depth interviews, and our findings encourage future qualitative work to examine how the timing of identity milestones shapes life course outcomes among diverse groups of transgender individuals. Despite these limitations, the NTDS is so far the most comprehensive national dataset that includes information on the timing of transgender identity milestones and educational attainment among several birth cohorts of transgender people across the U.S.

6. Conclusion

The U.S. has witnessed significant progress in promoting awareness and rights for transgender students in secondary schools through the emergence of GSAs, anti-bullying policies, and recent advocacy for transgender rights. Indeed, the 1990s, when the majority of Millennials experienced adolescence, is often considered a transitional decade in which the greatest increase in awareness, attention, and resources devoted to LGBT issues in schools emerged (E. Meyer 2015; Fetner and Kush 2008). Though many “LGBT” policies and programs are not specifically designed to address unique issues faced by transgender students (Dugan, Kusel, Simounet 2012), research has found that transgender students benefit socially and academically from these programs (Greytak et al. 2013). Similarly, many college campuses have worked to create safer and more inclusive environments for transgender students, taking steps such as adding gender identity and expression to non-discrimination policies, providing gender-inclusive housing, and allowing students to change campus records to reflect their identities (Beemyn 2015). At the same time that LGBT supportive programs and policies were emerging within schools, society’s understanding of gender variance was also changing as the medicalization and pathologization of transgender gave way to a social understanding of diverse transgender experiences and identities (Denny 2004; Schilt and Lagos 2017). Given these societal changes and progress, we would expect that experiencing transgender identity milestones in adolescence and transitioning into college as a transgender student would be less disruptive to educational attainment today than in the past, with younger cohorts of transgender adolescents better positioned to transform trans identities into resources (Nicolazzo 2016).

However, associations observed in the NTDS sample suggest that this is not the case, as the negative associations between first experiencing transgender identity milestones in adolescence and educational attainment are more pronounced among younger birth cohorts than among older birth cohorts. Transphobia is still highly pervasive in today’s society (Grant et al. 2010; Norton and Herek 2013), and increased attention to transgender rights and issues may increase the visibility and thus stigma faced by transgender students, making it more disruptive to explore and express gender variant identities in adolescence today than in the past. Our results from analysis of the NTDS sample suggest that secondary schools have more to do to provide environments that support the long-term educational success of transgender youth by reducing stigma and providing resources for students experiencing gender non-conformity and exploring trans identities and trans expressions. College campuses can also do more to provide inclusive learning environments that do not create additional barriers for transgender students (Beemyn 2003). The gendered structure of some aspects of campus life such as housing, student organizations, and athletic facilities often create unnecessary barriers for transgender students (Pryor 2015; Rankin et al. 2010), and being denied access to gender-segregated spaces has consequences for student well-being (Seelman 2016).

As transgender students continue to advocate for themselves, we anticipate that more schools will develop policies and practices that support transgender students throughout their educational career. We hope this effort remains inclusive of all transgender and gender non-conforming identities by recognizing the diverse needs and range of trans students who may feel stigma and stress associated with gender variance. Moreover, we emphasize an agenda that aims to undo transphobic and heteronormative social structures that place all students and faculty at risk due to narrow definitions of gender and strict policing of gender variance. Finally, we encourage programs and policies that recognize the resources trans students bring into classrooms and communities as well as their unique forms of resilience. Adolescent awareness of gender difference and adoption of transgender identities and expressions need not be seen or experienced as risk that leads to interrupted educational trajectories; rather, given a supportive context, transgender identities can serve as resources that promote resilience and long-term success while also serving to transgress the gender binary for the benefit of all students and faculty (Kosciw et al. 2015; Nicolazzo 2016). More qualitative work and more quantitative work based on population-based, longitudinal data should further explore these questions so we can better understand the changing landscape of transgender experiences and identities in our schools.