Abstract
Transgender (trans) people comprise about 2 million members of the U.S. population. While these communities are blueprints for culture and liberation, they face an acute threat of interpersonal violence, including domestic violence (DV) by intimate partners and family. Trans people are especially susceptible to polyvictimization: experiencing multiple types of violence, often across multiple settings. Furthermore, the negative impacts of violence are often compounded by institutional discrimination when seeking help, or sanctuary harm. For trans survivors of DV, it is crucial that researchers understand group differences in sanctuary harm during help-seeking—differences not only by identity and other biographical traits but also by sanctuary harm experiences in other institutional settings (e.g., healthcare, legal services). To explore associations between these factors, the current study presents a secondary analysis of data from the 2011 National Transgender Discrimination Survey, a large sample of trans and gender nonconforming adults living in the United States, Puerto Rico, the U.S. Virgin Islands, and Guam. Two binary logistic regressions model the relationship between several biographical and sanctuary harm predictors, and the study outcome of anti-trans sanctuary harm in DV services. Results indicate that sanctuary harm in four other settings—ambulance/emergency medical technician, mental health clinic, drug treatment program, and legal services—significantly predicts the same type of victimization in a DV services context. Some biographical factors were also significant: being more visibly nonconforming, being marginally housed, and ever engaging in sex work were all associated with higher odds of the dependent variable. Study limitations, suggestions for future research, and implications for comprehensive and intersectional DV services are discussed.
Nearly 2 million Americans 13 and older identify as transgender or trans (Herman et al., 2022). Trans individuals have a gender identity different from the sex assigned to them at birth, while cisgender (cis) people have a gender identity matching the sex assigned to them at birth (National Center for Transgender Equality, 2023). Trans people face disproportionately more victimization than their cis counterparts—for example, there are elevated rates of physical and sexual violence among trans women (Stotzer, 2009). Work on domestic violence (DV) among trans people suggests this type is quite prevalent (Peitzmeier et al., 2020; Scheer & Baams, 2021). A related area of research demonstrates that help-seeking trans violence survivors meet unique barriers or “sanctuary harm” (Bloom & Farragher, 2010) that arise from discrimination and violence (Kattari et al., 2017; Peterson et al., 2016). Using a national, all-trans, and gender nonconforming sample, this quantitative study highlights the impact of biographical and victimization factors on the odds of experiencing anti-trans sanctuary harm in DV services. Specifically, respondents are more at risk when they (a) are more visibly gender nonconforming; (b) are marginally housed; (c) have ever done sex work; and have experienced the same type of sanctuary harm in (d) an ambulance, (e) a mental health clinic, (f) a drug treatment program, or (g) a legal services clinic.
Domestic Violence and Group Differences
Broadly, DV is a pattern of harmful actions used to maintain power and control over an intimate, familial, or household relationship (Barocas et al., 2016; Berkshire District Attorney’s Office, n.d.; Peterson et al., 2016; United Nations, n.d.; Walters et al., 2011; Waters, 2017). While legal definitions vary across jurisdictions, DV includes physical, sexual, emotional, financial, and other behaviors (Barocas et al., 2016; Peterson et al., 2016; Walters et al., 2011; Waters, 2017). It can involve current or former romantic, dating, and sexual partners—intimate partner violence (IPV)—as well as parents and children, siblings, elders, and other relatives, known as family violence (FV; Barocas et al., 2016; Berkshire District Attorney’s Office, n.d.; United Nations, n.d.). Sociodemographic differences in DV are long-established, especially for IPV; for instance, early scholars highlighted the burden of IPV victimization placed on women (e.g., Dobash & Dobash, 1979). More recently, national studies like the National Intimate Partner and Sexual Violence Survey (NISVS) have examined IPV among LGBTQ+ subgroups, focusing on differences by sexual orientation and (binary) gender (Seelman, 2015). For example, compared to straight and lesbian women, bisexual women have significantly higher lifetime prevalence of physical IPV (Walters et al., 2011). Meanwhile, gay men have significantly higher lifetime prevalence of receiving psychological aggression than both bisexual and heterosexual men (Walters et al., 2011). However, while these assessments have certainly advanced DV scholarship beyond its traditional focus on heterosexual women as survivors and men as offenders, there is still limited work explicitly on DV among trans people (Etaugh, 2020). Surveys like the NISVS do not explicitly exclude trans Americans, but this group’s vulnerability and limited “outness” impede their adequate representation in probability-based sampling (Etaugh, 2020). As such, these findings predominantly represent cis experiences.
Trans Experiences of Domestic Violence
Quantitative scholarship on trans DV survivors is limited but growing. Of the existing studies, some explore HIV risk factors of this population, finding that physical IPV victimization is especially prevalent for trans women (Clements et al., 1999; Risser et al., 2005). Among 2015 U.S. Transgender Survey (USTS) respondents, 54% reported ever being victimized by a partner (James et al., 2016). Other IPV work has compared trans and cis experiences, with trans participants facing significantly more sexual and physical IPV (Peitzmeier et al., 2020; Valentine et al., 2017), threats of physical IPV (Landers & Gilsanz, 2009), and identity abuse (e.g., denial of a partner’s gender identity; Scheer & Baams, 2021). Regarding FV, trans people commonly report being kicked out of the home, assaulted, or being sent to anti-trans “therapy” while growing up (James et al., 2016). Other surveys have found that, of sexual assaults against trans adults, nearly one-quarter (24%) are family-perpetrated (Cook-Daniels & Munson, 2010). Of course, this area of victimization work is not without limitations. For one, studies with LGBTQ+ samples typically have small proportions of trans participants (Turell, 2000). Furthermore, very few studies of trans IPV examine differences across specific gender identities, as called for by recent theoretical work (Yerke & DeFeo, 2016); respondents are often aggregated into a single “transgender” category that erases important heterogeneity. That said, DV research on trans samples indicates this group’s high risk of both IPV and FV victimization.
Anti-Trans Challenges in Help-Seeking
As many trans people face increased risks of DV, they turn to others for help. Most often, these are informal social supporters (Guadalupe-Diaz & Jasinski, 2017; Kurdyla et al., 2021), but commonly pursued formal services include housing, healthcare, legal services, and emergency funds (Waters, 2017). There are also violence-specific initiatives; for example, about 38% of participants in the National Transgender Discrimination Survey (NTDS)—the same sample analyzed here—ever sought access to a DV shelter or program (Grant et al., 2011; Seelman, 2015). Unfortunately, while some survivors find safety, others face numerous barriers to effective help (Guadalupe-Diaz & Jasinski, 2017; Seelman, 2015). First, they often encounter gendered notions of victimization that make it difficult to identify their experiences as DV in the first place (Guadalupe-Diaz & Jasinski, 2017). Second, trans people may avoid approaching formal resources out of fear for transphobic discrimination (Messinger et al., 2022b); if they do choose to seek out these services, this fear is often realized (Guadalupe-Diaz & Jasinski, 2017). Many trans survivors experience unequal access to treatment, verbal harassment, and/or physical violence when accessing doctors’ offices, emergency rooms, drug programs, and mental health clinics (Kattari & Hasche, 2016; Kattari et al., 2015, 2017). Similar mistreatment has been reported in the case of DV programs, homeless shelters, rape crisis centers, and other social services (Kattari & Begun, 2017; Kattari et al., 2017; Seelman, 2015; Stotzer et al., 2013). In sum, formal organizations meant to foster healing can inflict harm on violence survivors (Bloom & Farragher, 2010; Peterson et al., 2016), especially when they are trans.
Conceptual Framework
Two concepts aid our understanding of the discrimination and violence that trans people face when accessing social services. The first, sanctuary harm, consists of the damaging impacts that institutions can have on members of marginalized groups (Bloom & Farragher, 2010). Different forms of structural oppression such as white supremacy and antiblackness, colonialism, and patriarchy are inherent in formal institutions and render them traumatogenic, especially when these long-standing dynamics are minimized (Peterson et al., 2016). One example of this process is in healthcare. Historically, the fields of medicine and psychiatry pathologized gender nonconformity and transition, classifying them as a mental disorder until 2013 (Peterson et al., 2016). Today, this legacy of transphobia is exemplified by interactions with health care professionals who are uninformed about trans needs at best, and outright discriminatory (or violent) at worst (Kattari & Hasche, 2016; Kattari et al., 2015, 2017). Regarding the U.S. criminal legal system, police raided bars and other businesses to arrest patrons for gender nonconforming behavior into the late-20th century (Buist & Stone, 2014). Scholars find that, troublingly, trans individuals still face increased police violence, especially on the basis of their gender expression (Grant et al., 2011; James et al., 2016; Waters, 2017). Overall, modern sanctuary harm toward trans survivors of violence follows the historical oppression of trans communities by myriad institutions. This harm can take many forms, from subtle microaggressions in organizational language to overt mistreatment (Peterson et al., 2016).
Second, while polyvictimization was initially conceptualized as experiencing multiple types of victimization, it also describes victimization in multiple contexts or settings, including the home, school, neighborhoods, the criminal legal system, and social services (Finkelhor et al., 2009; Wolfe, 2018). Understanding the diversity of victimization across contexts is crucial for studying its consequences, particularly for youth and marginalized groups. This multi-site exposure to victimization can have marked effects on both mental and physical well-being (Butcher et al., 2016; Turner et al., 2016). Trans individuals, particularly trans women, are notably vulnerable to polyvictimization across multiple contexts. This population faces significant victimization in both public and private spheres, including family rejection, school bullying, workplace discrimination, and inadequate support from anti-violence services (Hereth, 2021; Martín-Castillo et al., 2020; Messinger et al., 2022a). In sum, polyvictimization research suggests that trans people face compounded vulnerabilities across various domains of their lives and that victimization in one institutional setting can heighten vulnerability in others—presenting challenges for resilience and recovery efforts.
Current Study
Despite a growing body of evidence that trans individuals face higher rates of DV and other interpersonal victimization, research on harm experienced while seeking help from DV services is rather limited (see Seelman, 2015 for an exception). The current study seeks to build on this area of scholarship using a national sample of Americans who identify as trans or gender nonconforming (TGNC). Gender nonconforming people may not identify as trans but present differently than societal gender expectations—and are thus subjected to transphobia (Grant et al., 2011). I explore whether there is a differential risk of anti-trans sanctuary harm in DV services based on biographical factors and sanctuary harm in other institutional settings. The following research questions guide this study: (a) Among TGNC individuals who sought DV services, do biographical factors (e.g., race/ethnicity, gender identity, visible nonconformity, income, disability, sex work history) predict experiencing anti-trans sanctuary harm? (b) Are these anti-trans experiences in DV services associated with sanctuary harm in other institutional settings (e.g., emergency medical, mental health, or legal services)?
Methods
Data and Participants
Secondary data were drawn from the 2011 NTDS (N = 6,456), a National Center for Transgender Equality study conducted to explore discrimination experiences, other life events, and sociodemographic factors among TGNC Americans (Grant et al., 2011). Participants self-identified as trans and/or gender nonconforming, defined in the survey as, “people whose gender identity or expression is different, at least part-time, than the sex assigned to them at birth” (Grant et al., 2011). They included residents of all 50 states, Washington DC, Puerto Rico, the U.S. Virgin Islands, and Guam who were at least 18 years old at recruitment. The 70-item NTDS was administered electronically and on paper, in both English and Spanish. Data collection occurred from September 2008 to March 2009 through community-based, venue-based, and snowball sampling by over 800 trans-serving partner organizations—producing one of the largest samples of TGNC people ever gathered (National Center for Transgender Equality, 2015). The NTDS data continue to be used in examining TGNC victimization across institutional contexts (e.g., Enogieru et al., 2024; Kattari & Begun, 2017; Kattari & Hasche, 2016; Kattari et al., 2015, 2017; Klein et al., 2018; Lee & Santiago, 2023; Seelman, 2015), despite their age and other limitations (presented later in the Discussion section). Thus, this dataset is appropriate for examining anti-trans sanctuary harm in DV services; the current study uses a deidentified, public-use version of the NTDS housed online by the Inter-university Consortium for Political and Social Research (Grant et al., 2020).
Measures
Dependent Variable
The current study outcome captures anti-trans sanctuary harm in DV services. NTDS respondents were asked, “Based on being transgender/gender nonconforming, please check whether you have experienced any of the following in these public spaces.” For each setting, including DV shelters/programs, response options were: (a) denied equal treatment or service, (b) verbally harassed or disrespected, (c) physically attacked or assaulted, (d) not applicable, I have not tried to access this, (e) not applicable, I do not present as transgender here, and (f) not applicable, I did not experience these negative outcomes. Responses to each type were coded as 1 = yes, 0 = no. Then, a dichotomous variable was created to represent any type of harm in a DV services context, capturing a range of victimization experiences. The present study only examined respondents who reported accessing a DV shelter or program (n = 2,438); thus, current descriptive statistics and analyses reflect this subsample.
Biographical Predictors
The framework of intersectionality (Crenshaw, 1989, 1990) allows victimization scholars to study individuals beyond singular categories of identity—that is, to examine connections among their personal characteristics, experiences, and broader structural contexts. In Seelman’s (2015) NTDS analysis, the odds of being denied equal treatment or service in a DV shelter/program were significantly higher for those who were lower in income, non-U.S. citizens, people of color, disabled, and more often perceived as TGNC by others. These results suggest that TGNC people face differential risk for anti-trans victimization in DV services based on the intersections of their gender identity with other factors. As such, this study expands on Seelman’s (2015) impactful work by considering additional biographical characteristics.
NTDS participant race/ethnicity 1 was measured with a series of six dummy variables: white (the reference group in analyses), Black, Indigenous, Latine, Arab/Asian/Pacific Islander, and multiracial. Arab and Asian/Pacific Islander respondents were collapsed into a single group because of low counts separately. Sexual identity was also measured using a series of dummy variables: gay or lesbian, bisexual, queer or pansexual, heterosexual (reference group), and asexual or other (combined due to low response counts). Write-in options for “other” included identities like butch and demisexual, as well as “don’t know” (Grant et al., 2011). Gender identity was assessed by the question, “What is your primary gender identity today?” Responses were coded as three dummy variables: trans man (reference group), trans woman, and part-time or not listed gender. The “part-time” category was inclusive of those with multiple, situational, or more fluid gender identities (Seelman, 2015), while “not listed” included write-in responses like genderqueer (Grant et al., 2011). Respondents were also asked to report their U.S. citizenship status, answered as either (a) U.S. citizen, (b) documented non-citizen, or (c) undocumented non-citizen. Due to low response counts, I created a binary variable where respondents were coded as 1 = non-citizen (documented or not), 0 = citizen. To assess disability status, participants were asked if they have a physical, learning, or mental health disability that substantially affects major life activity; responses were coded dichotomously as 1 = disabled, 0 = not disabled. For self-reported HIV serostatus, respondents were asked the question, “What is your HIV status?” and answered with one of the following options: (a) HIV negative, (b) HIV positive, or (c) don’t know. Then, I created a binary variable, coded as 1 = HIV-positive and 0 = HIV-negative or don’t know (following Klein et al., 2018). Age was measured in years (18–81), and annual household income was coded as 5 ordinal categories, ranging from 1 = <$10,000 to 5 = $100,000 or more.
Being perceived as TGNC by others because of one’s appearance (being “clocked”; Eyre et al., 2004; Sausa et al., 2007; Schiffer, 2022) was measured using one item: “People can tell I’m transgender/gender nonconforming even if I don’t tell them.” Respondents answered on a 5-point ordinal scale, ranging from (1) always to (5) never. Though conforming to societal gender norms is not a universal goal of trans people, individuals may be targeted for anti-trans discrimination due to their perceived nonconformity (Grant et al., 2011). For the present study, this visible nonconformity variable was reverse-coded (i.e., 1 = never, 5 = always) so that nonconformity increases with higher values. Gender may also be indicated legally, through markers on legal identification (ID) such as a passport. These gender markers may not match a trans person’s identity for a variety of reasons, which can alert formal institutions of their transness and put them at risk for victimization (Spade, 2008). As access to healthcare and social services often requires the presentation of valid ID (James et al., 2016), I incorporate an ordinal measure of incorrect legal gender markers. Respondents were asked to indicate whether none, some, or all of their IDs and records show their correct gender. Again, this variable was reverse-coded (1 = all, 2 = some, 3 = none) so that gender nonconformity increases with higher response values. Marginal housing was assessed using the question, “What are your current living arrangements?” Those who were homeless; living in a shelter; living in a group home facility or other foster care situation; or staying with friends or family temporarily were coded as 1 for marginally housed, while the remainder were coded as 0 for not marginally housed (see Klein et al., 2018). Respondents also reported whether they ever did sex work, coded as 1 = yes, 0 = no.
Sanctuary Harm Predictors
Lastly, measures of any sanctuary harm were included for seven other public contexts: doctor’s office or hospital; ambulance or emergency medical technician (EMT); emergency room; mental health clinic; drug treatment program; legal services clinic; and judge or court official. As with the dependent variable, these indicators include being denied equal treatment or service; verbally harassed or disrespected; and physically attacked or assaulted. For each setting, respondents were coded as 1 if they endorsed any victimization type, and 0 if they endorsed none. These measures consider the breadth of sanctuary harm experienced by TGNC people (Peterson et al., 2016) and allow exploration of sanctuary harm in other settings as predictors of the same in DV services.
Analytic Strategy
Statistical analyses were conducted using RStudio version 1.4.1103. First, all study variables were examined for missingness; none were more than 10% missing. However, when examining complete cases, only 1,728 respondents of the total 2,438 were retained following listwise deletion. As such, multiple imputation was conducted using the R package mice (van Buuren & Groothuis-Oudshoorn, 2011). Five datasets were imputed, which Rubin (1996) posits is sufficient. In addition, tests for multicollinearity showed no VIF higher than 2.508, and no correlation higher than r = .60 among study variables (Kumari, 2008). Lastly, G*Power version 3.1 was used to conduct a power analysis. A binomial logistic regression model with an assumed DV services sanctuary harm rate of 9–19% (see Grant et al., 2011; Kattari et al., 2017; Seelman, 2015), 28 covariates, an alpha level of .05, and a power level of 0.80 (Cohen, 1988; Faul et al., 2007) would require a minimum of 1,093 respondents, meaning the study sample of 2,438 is more than adequate. Considering the binary outcome, a two-step logistic regression was run to examine predictors of anti-trans sanctuary harm in DV services. In model 1, only the biographical variables were included as independent variables. Then, in model 2, measures of sanctuary harm in different settings were added as predictors. Odds ratios (OR) with 95% confidence intervals are reported for all predictors, in addition to model goodness-of-fit statistics.
Results
Participant descriptive statistics (N = 2,438) are displayed in Table 1. Sample members were predominantly white (78.4%) and nearly half were trans women (49.8%). Additionally, there were similar proportions of gay/lesbian (26.1%), bisexual (23.8%), and heterosexual participants, but fewer identified as queer/pansexual (18.8%) or asexual/other (7.0%). Almost one-third of participants were disabled (41.4%), while far fewer were non-U.S. citizens (4.3%) or living with HIV (4.5%). Most sample members also reported an annual household income below $50,000 (60.5%), and that “none” or “some” of their legal gender markers were correct (73.6%). Some participants reported a history of sex work (15.5%) and being marginally housed (7.3%). On average, sample members reported a medium-to-high degree of visible nonconformity (M = 2.54, SD = 1.20) and were nearly 38 years old (M = 37.88, SD = 13.54). Regarding victimization experiences, 9.2% of participants reported any sanctuary harm in DV services (the outcome of interest). Sanctuary harm prevalence varied across the other settings; drug treatment programs were reported the least (6.2%) while doctor’s offices/hospitals were reported the most (29.8%).
Sample Descriptives (N = 2,438).
Table 2 presents the results of a two-step logistic regression predicting anti-trans sanctuary harm in DV services. Model 1 results indicate several significant biographical predictors: Indigenous (OR = 1.68, p < .05), non-U.S. citizen (OR = 2.06, p < .05), and disabled participants (OR = 2.23, p < .001) each had higher odds of experiencing anti-trans sanctuary harm in DV services. Visible nonconformity also emerged as a significant factor, with more nonconforming respondents having significantly greater odds of this sanctuary harm outcome (OR = 1.21, p < .01). Furthermore, respondents who were marginally housed (OR = 3.07, p < .001) and ever did sex work (OR = 3.83, p < .001) each reported significantly higher odds of the dependent variable—the two strongest predictors of model 1. Meanwhile, age and annual household income were inversely related to the outcome (age: OR = 0.97, p < .01; income: OR = 0.85, p < .05), meaning that participants of older age and higher income categories had significantly lower odds of experiencing transphobia in DV services. Lastly, several biographical factors were non-significant in model 1: being Black, Latine, Arab/Asian/Pacific Islander, or multiracial; having incorrect ID gender markers; being HIV-positive; all sexualities; and all gender identity groups.
Logistic Regressions Predicting Transphobic Mistreatment in Domestic Violence Services (N = 2,438, Pooled After Multiple Imputation).
Note. OR = odds ratio; CI = confidence interval; HL = Hosmer-Lemeshow test.
p < .05. **p < .01. ***p < .001.
The second regression step (model 2) added measures of anti-trans sanctuary harm in seven other institutional settings; results indicate that about half were significantly related to the outcome variable. Notably, being mistreated in an ambulance/by an EMT (OR = 2.67, p < .05), in a mental health clinic (OR = 3.67, p < .001), in a drug treatment program (OR = 5.47, p < .01), and in a legal services clinic (OR = 6.73, p < .001) were all significant correlates of anti-trans sanctuary harm in a DV shelter/program context. Model 2 also controlled for the same biographical factors present in model 1, although fewer were significant predictors. Again, being more visibly nonconforming was associated with greater odds of experiencing anti-trans sanctuary harm in DV services (OR = 1.28, p < .05). Lastly, being marginally housed (OR = 3.23, p < .001) and ever doing sex work (OR = 2.71, p < .001) were also significant covariates in model 2, consistent with model 1. However, while model 2 controlled for the same biographical factors present in model 1, fewer of them were significant. For instance, the odds associated with being Indigenous and a non-U.S. citizen were again elevated (OR = 1.63 and OR = 2.73, respectively), but they did not reach statistical significance as in model 1. Similarly, the citizenship status (OR = 2.73), age (OR = 0.98), and income (OR = 0.89) measures were not significant in model 2, contrasting model 1.
Discussion
To pursue more inclusive violence research (Bent-Goodley, 2021) and counter the growing anti-trans political climate, scholars increasingly use their work to challenge outdated, transphobic ideas—whether espoused by online hate commenters (Walker, 2023) or colleagues in their own field (Valcore et al., 2021). Following their lead, the current study investigated biographical and victimization correlates of anti-trans mistreatment in DV services. Among an analytic sample of TGNC Americans, three major findings were identified. First, of the biographical factors considered in both models, marginal housing, sex work history, and visible gender nonconformity were strong, significant predictors of this anti-trans sanctuary harm in DV services. This is consistent with much of the existing trans victimization research. For instance, studies of trans people’s socioeconomic vulnerability find that many—especially trans youth and women—are mistreated in family, employment, and education contexts (James et al., 2016); as such, they are increasingly prone to homelessness (Durso & Gates, 2012; McCann & Brown, 2021) and doing survival sex work (Kattari & Begun, 2017; Sausa et al., 2007). Further, TGNC people simply perceived as sex workers—regardless of the reality—are subject to more interpersonal violence (Steele et al., 2020) and discrimination from institutions like law enforcement (Carpenter & Marshall, 2017; Grant et al., 2011). Similarly, individuals who are “clocked” as trans by others (Eyre et al., 2004; Sausa et al., 2007; Schiffer, 2022) face increased victimization in public spaces (Grant et al., 2011; Reisner et al., 2014).
The second finding is that other biographical variables appear partially related to the study outcome. With race, Indigenous respondents had significantly higher odds of sanctuary harm in DV services than their white counterparts, an observation resembling current research on race differences in IPV. For instance, nearly 48% of Indigenous women and 41% of Indigenous men have ever experienced intimate partner-perpetrated sexual assault, physical violence, or stalking (Smith et al., 2017)—more victimization than any other U.S. ethnoracial group. While there is some evidence that Indigenous trans survivors are more likely to seek out DV or rape services (Messinger et al., 2022b), settler institutions like the criminal legal system may present this population with linguistic and cultural barriers, jurisdictional issues when on tribal lands, and overt racism (Finfgeld-Connett, 2015; Miller et al., 2022). Citizenship status, age, annual household income, and disability status were also significant in model 1, and their directionality echoed prior work on transphobia across institutions (see Kattari et al., 2021; Kurdyla, 2022; Mottet & Ohle, 2006). Specifically, being a non-U.S. citizen, younger, in a lower income category, or disabled all had higher odds of anti-trans sanctuary harm in DV services. However, when the independent variables for anti-trans sanctuary harm in several non-DV services were included for model 2, Indigeneity, citizenship status, age, income, and disability status were not significantly related to the DV services outcome. In a statistical sense, it may be that the significant sanctuary harm independent variables—ambulance/EMT, mental health clinic, drug treatment program, and legal services clinic—are mediating the relationship between biographical risk factors and anti-trans sanctuary harm in DV services. Polyvictimization, then, can help us understand these dynamics. For one, the biographical correlates of anti-trans sanctuary harm in DV services might also be associated with anti-trans sanctuary harm in these other settings; in other words, different institutional services have transphobia risk factors in common. Additionally, it may be the case that TGNC people harmed by one setting are better able to perceive (and then report) this harm in others. For instance, if a trans woman is denied equal service in a drug treatment program, she may be more attuned to recognize this mistreatment in a legal services clinic. Lastly, this study focused on TGNC individuals who sought help from a DV shelter or program. Considering the general reluctance of trans victim-survivors to approach formal institutions for help (Scheer et al., 2020), it is likely that this sample approached more social and healthcare services due to high frequency or severity of victimization and its associated problems (e.g., psychological trauma, drug use). In other words, findings may reflect a particular group of TGNC people subjected to intense victimization and associated problems, with an elevated need to seek resources from multiple institutions (that can be harmful, in turn).
Third, it is notable that several biographical factors were not significant in either model. For example, Black, Latine, Arab/Asian/Pacific Islander, and multiracial people did not significantly differ in their odds of anti-trans sanctuary harm in DV services. Similarly, being gay or lesbian, bisexual, queer or pansexual, or asexual/other sexual orientation were not significantly related to the sanctuary harm outcome when compared to straight respondents. For gender, being a trans woman, or a part-time or not-listed identity were also non-significant factors; the same was true for having incorrect gender markers on legal ID and being HIV positive. These results are counterintuitive when considering the known victimization susceptibility of trans people with other marginalized attributes (Dinno, 2017; Westbrook, 2023). However, some work on trans survivors’ help-seeking has similarly surprising findings—for example, that assigned-female-at-birth nonbinary participants in the 2015 USTS are more likely to seek help from a DV shelter/program or rape crisis center (Messinger et al., 2022b). The current study’s null findings have two potential explanations. First, because the survey questions only asked participants about mistreatment based on being transgender or gender nonconforming, they were not able to assess mistreatment based on other characteristics (e.g., race/ethnicity, sexual orientation, HIV status); thus, these variables have no significant effects in the current models. Second, similar to Seelman’s (2015) findings, the factors significant in both models—marginal housing, sex work history, and visible gender nonconformity—might capture risk of anti-trans sanctuary harm beyond that captured by race/ethnicity, HIV status, and other variables. For instance, given that respondents of color in the total NTDS sample had higher rates of unemployment and workplace discrimination than white respondents (Grant et al., 2011), sex work history and marginal housing may mediate effects from the race/ethnicity indicators. Similarly, given the established role of visible nonconformity in polyvictimization risk (Messinger et al., 2022a; Sterzing et al., 2017), it may matter more for transphobia in DV services than one’s legal gender markers do.
Limitations and Implications for Research
Of course, the current study has limitations. First, the NTDS data collected in 2008 to 2009 precede significant political and social shifts, including recent anti-trans state legislation (Trans Legislation Tracker, 2024) and executive orders (Trump, 2025). While DV services have likely improved inclusivity in states that have trans legal protections, providers in states with trans-exclusionary policies may now be even less receptive, suggesting the sanctuary harm identified here may underestimate current conditions. While the 2015 USTS (James et al., 2016) collected data more recently and from a larger sample, some NTDS sanctuary harm indicators were no longer present or measured in the same way. For example, the NTDS measures anti-trans mistreatment in several specific health care settings but the USTS asks about these harms “in a health care setting” broadly, erasing nuance (Grant et al., 2011; James et al., 2016). The USTS also uses a single sanctuary harm measure for “Domestic violence shelter/DV program/Rape crisis center,” unlike the NTDS; while similar, services for DV and sexual assault survivors are often distinct in specific needs served, staff size, and resources available (see DeDomenico-Payne, 2006). Ultimately, the NTDS data remain suitable for exploring DV sanctuary harm.
Second, NTDS participants were not randomly sampled and, therefore, selection bias may limit interpretation of findings; however, the large national scope is undoubtedly a strength. Third, as a cross-sectional survey, causal claims cannot be established. Furthermore, the timeframe of discrimination relative to other experiences or identities is unclear. It is not possible to determine if mistreatment in other settings (e.g., in a legal services clinic) took place before, concurrently with, or after DV shelter mistreatment—warranting future research that explicitly measures risk factors at the time of service use. Fourth, the dependent variable examined here asks about experiences of mistreatment in a domestic violence shelter or program. While DV shelters are designed to serve survivors, certain DV programs (e.g., batterer interventions) are intended to aid perpetrators in desisting from violence. Since this distinction was not outlined for participants in the NTDS questionnaire, differences in specific DV services accessed could be obscured here. Future surveys should separately measure DV victimization and perpetration experiences and differentiate between partner and family violence, given their distinction under the larger umbrella of DV (Barocas et al., 2016; Berkshire District Attorney’s Office, n.d.; United Nations, n.d.). Additionally, identifying sanctuary harm perpetrators (staff or clients) would provide more precise insights. Fifth, the NTDS asked about mistreatment solely based on being transgender or gender nonconforming; thus, future studies should explore victimization driven by other intersecting factors. Lastly, while participants reported if they have “physical, learning, [or] mental health” disabilities, they were not asked how visible or hidden they are—excluding a key dimension of disability (Olney & Brockelman, 2005).
Implications for Practice and Conclusion
This study also has implications for social services at a time when formal resources and protections are increasingly denied to trans people and, in some cases, their allies (Trans Legislation Tracker, 2024). First, given that polyvictimization can occur when serving violence survivors, providers should be further attuned to clients’ experiences in other settings. Specifically, it may benefit TGNC survivors of DV if social services screen for experiences of sanctuary harm in emergency medicine, legal services, mental healthcare, and other institutional contexts. DV service providers being attuned to this victimization in other settings may improve their own treatment of TGNC clients. Second, there is a clear need for DV shelters and programs to implement comprehensive policies recognizing the nuances of gender identity and expression. Considering the notable impacts of visible nonconformity here and in other TGNC victimization work (Messinger et al., 2022a; Sterzing et al., 2017), DV service providers should train staff (and encourage clients) to be aware and affirming of not only different gender identities but also varied physical appearance as a facet of gender expression. Finally, results indicate that trans people who are additionally marginalized face greater risks of sanctuary harm; this is especially true for the marginally housed, and current or former sex workers. As such, social services should tailor their practices to trans individuals’ unique needs.
This kind of inclusive work is already being done by and for Black trans communities. In May 2024, activist, actress, and media personality TS Madison Hinton announced her own housing initiative for women leaving survival sex work: “a supportive and thriving community where black trans women can live freely, embracing their identity without fear” (NAESM Inc. [@naesminc], 2024). Beginning March 31, 2025, the initiative provides comprehensive counseling, healthcare, and financial education services alongside safety from violence. Hinton’s efforts will primarily be funded by non-profit organizations and individual donors (including herself), but this type of trans-affirming program should also be created through direct government resources. Ultimately, the study reported here indicates that TGNC survivors of DV have varied experiences in help-seeking, contingent on both other identities and polyvictimization across settings. Making practical changes in DV services that are tailored to these differences can foster a protective, healing environment—a sanctuary without harm.
Footnotes
Acknowledgements
I thank my colleagues Chase Childress, Carlos Cuevas, Megan Denver, and Brandon Welsh for their feedback on earlier versions of this manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interests with respect to the authorship and/or publication of this article.
Funding
The author(s) received no financial support for the research and/or authorship of this article.
