Cultural Adaptation and Evaluation of Evidence for Validity of a Gender Minority Stress Measure Among Transgender Women and Transfeminine People in China

Abstract

Purpose:

Gender minority stress is a culturally situated, complex construct. Despite high levels of gender minority stress reported by transgender and gender-diverse (TGD) people in mainland China, measurement of this construct in this setting is not well understood. We conducted cultural adaptation of a gender minority stress measure (discrimination, rejection, victimization, nonaffirmation of gender identity, internalized transphobia, negative expectations for the future, and nondisclosure of gender identity subscales) and evaluated evidence of validity and reliability in a sample of Chinese transgender women, transfeminine people, and other gender-diverse people assigned male at birth (n = 277).

Methods:

Community-engaged cultural adaptation included needs assessment, cognitive interviewing, translation, and community and academic expert consultation. Participants (n = 277) completed an online questionnaire assessing seven stress subscales adapted from the Gender Minority Stress and Resilience (GMSR) measure (November 2019–June 2020). We conducted confirmatory factor analyses to confirm the underlying structure and examined divergent and criterion validity.

Results:

A confirmatory factor analysis found that a first-order seven-factor model had adequate fit and was more appropriate than a second-order model. The seven subscales had positive correlations with each other of under 0.60, which supported the divergent validity hypothesis. However, while all subscales were positively correlated with depressive symptoms, only nonaffirmation was correlated with cigarette use, suggesting partial support for the criterion validity hypothesis.

Conclusion:

Results suggest good evidence for validity of the GMSR stress subscales and support its future use among other Chinese TGD populations. Community-engaged methods may be beneficial for adapting complex measures across settings.

Introduction

Transgender and gender-diverse (TGD) people report disproportionately high levels of health burden across several outcomes and behaviors, including depression and tobacco use, due to stigma at the structural, interpersonal, and individual levels.^1–3 These forms of stigma and resulting gender minority stress—excess stress arising from one’s identity and experiences as a member of a minoritized group—are hypothesized to contribute to disproportionately high levels of poor health outcomes and risk behaviors.^4,5 Studies on transgender women worldwide have found positive associations between multiple forms of gender minority stress and depression ⁶ and cigarette use. ⁷ In China, transgender women report high levels of depression ¹ and cigarette use, ⁸ which may similarly be attributable to gender minority stress.

High proportions of TGD people in China report societal and institutional discrimination (27.3−80.1%)^9–11 and lack of access to medical gender affirmation; ¹² violence (13.9−80.8%);^11,13–15 bullying (75.2–76.2%);^9,16 and interpersonal rejection (43.1−98.9%).^11,15 While there is a long history of diverse transgender identities and practices in China,^17,18 exclusionary policies and cultural and historical context may contribute to these heightened minority stressors. For example, a public opinion study found that 36.7% of Chinese respondents felt that transgender people violated cultural traditions. ¹⁹ These views are shaped by Confucian values such as continuing the family line through marriage and having biological children, which exclude many Chinese transgender women. ²⁰

Political and economic forces since 1949 have also shaped views on TGD people in contemporary China. Maoist socialism suppressed most public discussion and behaviors related to transgender identities and practices, while Opening Up and Reform policies starting in 1978 introduced both medical pathologization of TGD people and gender affirmation technology. ¹⁷ Since the 2000s, the state expanded the regulation of medical and legal gender affirmation, ²¹ in line with public ²² and political ¹⁷ tolerance of lesbian, gay, bisexual, and transgender (LGBT) people. These complex dynamics may underlie TGD people’s experiences of gender minority stress in contemporary China.

Accurately measuring gender minority stress in the Chinese context is critical for advancing research on the health and well-being of TGD populations. The widely used Gender Minority Stress and Resilience (GMSR) measure was developed in the United States of America and Canada using psychometric best practices. ⁴ The GMSR includes seven stress subscales, covering distal (gender minority identity-related discrimination, rejection, victimization, and nonaffirmation of gender identity) and proximal stressors (internalized transphobia, negative expectations for the future, and nondisclosure of gender identity). ⁴

However, there is limited research assessing the psychometric properties of the GMSR in Chinese contexts. One of the few studies on the general TGD population in China found evidence for divergent and criterion validity and some support for convergent and structural validity for the stress measures. ²³ Other adapted versions of the victimization and internalized transphobia subscales demonstrated evidence of test-retest validity ¹⁵ and convergent validity, ²⁴ respectively. These studies may have limited generalizability as participants were primarily sex workers, younger individuals, or students.^15,23,24 Additionally, studies have not described critical scale development steps, such as back translation and cognitive interviewing to ensure scale items are understood as intended by respondents. ²³

Given the complexity of this measure and cultural specificity in experiences of gender minority stress, community-engaged cultural adaptation is important for developing a meaningful measure, especially as the original measure was developed among North American populations. When research teams are primarily composed of those outside of the community these approaches also provide important insights from lived experience and ensure that measures are responsive to community priorities.

This study addressed these gaps by culturally adapting the stress subscales of the GMSR measure using standard and community-engaged measurement methods and evaluating the validity and reliability of the adapted measure in a sample of transgender women, transfeminine people, and gender-diverse people assigned male at birth in China. This study did not include the resilience subscales (pride and community connectedness) as resilience is a distinct construct from stress ⁵ and community experts and members prioritized the stress subscales during the formative development process. Additionally, to optimize our data, this study focused on the experiences of transgender women, transfeminine people, and gender-diverse people assigned male at birth. The study team prioritized this group given the growing research on this population in China and resource constraints that limited the feasibility of adapting the measure for additional populations.

Materials and Methods

TGD community members and experts were engaged throughout this study. The first stage (July 2017–August 2017, June 2018–August 2018, July 2019–January 2020) involved original formative ²⁵ and qualitative fieldwork to ensure that adapting this measure aligned with community priorities and to build collaborations. Informed by this fieldwork, the second stage (September 2019–June 2020) adapted the measure and assessed evidence for validity.

Cultural adaptation of the GMSR measure

Community and academic expert review

The GMSR measure was translated from English to Chinese and then back-translated into English by bilingual graduate-level researchers. Academic and community experts reviewed scale items to improve content validity and wording. Academic experts not involved in the study reviewed the items for face validity and comprehensiveness. Community experts, including one transgender woman and one nonbinary person, with extensive experience conducting survey research on transgender health in China provided additional feedback on psychometric performance of items and cultural tailoring.

Cognitive testing

We conducted cognitive testing to understand how study population members perceived questionnaire directions and items. ²⁶ Two study team members conducted cognitive interviews (n = 10, a typical sample size for cognitive testing), ²⁶ in Guangzhou and Shenyang (September 2019–January 2020). An interviewer read the question aloud and followed up with probes. The participant described their thought process as they responded. The second interviewer took detailed field notes, which were coded using a scheme informed by the Lee Cross-National Appraisal system. ²⁷ The first author listened to recordings immediately after data collection to confirm field note accuracy and wrote memos summarizing key findings.

Psychometric testing

Participants and Setting. Inclusion criteria were: assigned male sex at birth, age 18 or older, not identifying as male, and could read simplified Chinese. Online questionnaire data collection was open to those typically residing in Guangzhou, Guiyang, Jinan, Qingdao, Kunming, Nanjing, Nanning, Shenyang, and Wuhan.

Recruitment

Participants were recruited online and in-person through community-based organizations (CBOs) (November 2019–June 2020). Partner organizations, which included HIV services and TGD community organizations, posted announcements in WeChat (a smartphone chat application) groups and social media accounts. Starting in February 2020, due to disruptions from the COVID-19 pandemic, participants completed the questionnaire remotely if preferred.

Data collection

We pretested and revised the questionnaire and administered the final questionnaire online through Wenjuanxing (Zhongdian, China, www.wjx.cn). Participants accessed the questionnaire through a link and completed a screener and informed consent. Those who completed the questionnaire were compensated with 100 renminbi (∼15 US dollars). Methods to reduce fraud and increase response quality included only being able to access the questionnaire through direct contact with a partnering CBO, attention checks, reviewing internet protocol (IP) addresses, and excluding questionnaires completed in under 5 minutes based on pretesting.

Measures

Demographics. Demographics included age, education, income, rural or urban residency, sexual attraction, gender identity, housing, marital status, and ethnicity.

GMSR subscales. See Table 1 for all items. For discrimination and rejection, response options included “Yes” (scored 1) or “No” (scored 0). Based on community expert and participant feedback, the victimization response options were modified to include partner; family member; teacher, classmate, or other student; friend; and someone else (scored 1 for any “Yes” response), “No” (scored 0), or “Refuse to answer” (missing).

Table 1.

Latent Factors, Scale Items, Descriptive Statistics, and Standardized Factor Loadings for the First-Order Gender Minority Stress Model

Subscale and items (complete responses)	Responses a
	No n (%)	Yes n (%)	Cronbach’s Alpha	Factor loadingb	Standard error
Discrimination (n = 274)			0.72
1. I have had difficulty getting medical or mental health treatment (transition-related or other) because of my gender identity or expression.	133 (48.0)	144 (52.0)		0.683	0.067
2. Because of my gender identity or expression, I have had difficulty finding a bathroom to use when I am out in public.	112 (40.4)	163 (58.8)		0.705	0.059
3. I have experienced difficulty getting identity documents that match my gender identity.	89 (32.1)	186 (67.2)		0.701	0.059
4. I have had difficulty finding housing or staying in housing because of my gender identity or expression.	199 (71.8)	77 (27.8)		0.786	0.053
5. I have had difficulty finding employment or keeping employment, or have been denied promotion because of my gender identity or expression.	135 (48.7)	141 (50.9)		0.903	0.038
Rejection (n = 271)			0.80
1. I have had difficulty finding a partner or have had a relationship end because of my gender identity or expression.	106 (38.3)	170 (61.4)		0.650	0.068
2. I have been rejected or made to feel unwelcome by a religious community because of my gender identity or expression.	168 (60.6)	105 (37.9)		0.748	0.048
3. I have been rejected by or made to feel unwelcome in my ethnic/racial community because of my gender identity or expression.	177 (63.9)	99 (35.7)		0.846	0.038
4. I have been rejected or distanced from friends because of my gender identity or expression.	158 (57.0)	117 (42.2)		0.812	0.055
5. I have been rejected at school or work because of my gender identity or expression.	147 (53.1)	128 (46.2)		0.958	0.037
6. I have been rejected or distanced from family because of my gender identity or expression.	159 (57.4)	117 (42.2)		0.813	0.056
Victimization (n = 246)			0.84
1. I have been verbally harassed or teased because of my gender identity or expression.	43 (15.5)	223 (80.5)		0.892	0.101
2. I have been threatened with being outed or blackmailed because of my gender identity or expression.	166 (59.9)	99 (35.7)		0.854	0.045
3. I have had my personal property damaged because of my gender identity or expression.	155 (56.0)	113 (40.8)		0.904	0.038
4. I have been threatened with physical harm because of my gender identity or expression.	150 (54.2)	112 (40.4)		0.933	0.031
5. I have been pushed, shoved, hit, or had something thrown at me because of my gender identity or expression.	144 (52.0)	118 (42.6)		0.924	0.037
6. I have had sexual contact with someone against my will because of my gender identity or expression.	196 (70.8)	66 (23.8)		0.702	0.069

Cronbach’s Alpha refers to Internal Consistency Reliability.

^aFrequencies may not add up to 100% due to missing responses.

^bAll factor loadings were statistically significant (p < 0.05).

Nonaffirmation, internalized transphobia, negative expectations, and nondisclosure were measured using a four-point Likert scale with responses ranging from “Strongly Disagree” (scored 0) to “Strongly Agree” (scored 3). This decision to use a four-point rather than the five-point Likert scale in the original measure was based on past studies indicating that Chinese respondents may be more likely to choose midpoint responses compared with North American respondents^28,29 and concerns about respondents’ level of engagement given instrument length.

Depressive symptoms. The Center for Epidemiological Studies Depression Scale, 10-item measure (CES-D 10) adapted for Chinese populations ³⁰ was used to assess depressive symptoms with four-point Likert scale response options: “Rarely or none of the time (<1 day)” (scored 0), “Some or a Little of the Time (1–2 days)” (scored 1), “Occasionally or a Moderate Amount of Time (3–4 days)” or “Most or All of the Time (5–7 days)” (scored 3). A score of 12 or higher indicated risk for clinical depression. ³¹

Current combusted cigarette use. Current combusted cigarette use (referred to as cigarette use) was defined as nonexperimental past 30-day use and measured by adapted 2018 Behavioral Risk Factor Surveillance Survey items for experimentation (“Have you smoked at least 100 cigarettes in your life?”) and point prevalence abstinence (“Have you smoked a cigarette, even a puff, in the past: week, month, 3 months?”). ³²

Data analysis

Descriptive statistics and internal consistency. Internal consistency reliability of the subscales was assessed by calculating Cronbach’s alpha. Descriptive statistics were generated and inspected. ³³ Models with missing data were estimated using pairwise deletion in Mplus. ³⁴

Confirmatory factor analyses. We conducted confirmatory factor analyses (CFA) of the subscales to confirm the underlying item structure. ³⁴ We used the weighted least square mean and variance-adjusted estimator as items were ordinal. We modeled each subscale as a latent construct correlated with all other subscales and items were loaded onto these latent constructs. ⁴ Model fit was compared to that of an alternate, second-order model (assessed when there is substantial correlation among constructs). ³⁵ Here, the latent constructs were loaded onto a higher-order latent construct representing an underlying stress construct.

Model fit was assessed using the chi-square value (nonsignificant), comparative fit index (CFI, >0.95), Tucker–Lewis index (TLI, >0.95), root mean square error of approximation (RMSEA, <0.06), ³⁶ and the standardized root mean square residual (SRMR) (<0.05). ³⁷ We hypothesized that gender minority stress will consist of seven minority stressors (discrimination, rejection, victimization, nonaffirmation, internalized transphobia, negative expectations, and nondisclosure) (Hypothesis 1).

Criterion validity. As there are correlations between gender minority stress and depression ⁴ and tobacco use ⁷ among TGD samples, we assessed criterion validity by examining the significance and direction of correlations between the subscales with depressive symptoms and cigarette use. We hypothesized that the seven subscales will be positively correlated with depressive symptoms (Hypothesis 2) and that the seven subscales will be positively correlated with current cigarette use (Hypothesis 3).

Divergent validity. Divergent validity was assessed through examining the correlations of the seven subscales with each other. A positive and significant correlation less than or equal to 0.6 indicates that the subscales are related but conceptually distinct. ³⁸ We hypothesized that the seven subscales will be positively correlated with each other, with correlations less than or equal to 0.6 (Hypothesis 4).

Ethics statement

This study was approved by the Institutional Review Boards of the University of North Carolina at Chapel Hill (IRB# 18-3215) and the Ethics Committee at Southern Medical University Dermatology Hospital (IRB# GDDHLS-20190505I).

Results

Cognitive interview results

Participants noted that the instructions providing definitions on gender expression, history, and identity were confusing and unnecessary and these instructions were deleted. Second, some participants did not find discrimination items relevant. For example, discrimination subscale items on difficulty obtaining documents with the correct gender identity marker and finding employment were irrelevant as some participants were not interested in changing their identity documents, had not attempted to do so, or felt that they did not experience employment discrimination as they worked in sex work. However, taken together with prior research^9,10,20 and fieldwork, we retained the discrimination items.

Participants identified issues with tailoring items to the Chinese context. Some did not know how to respond to items on ethnic or religious group rejection as they felt there was no consensus within the Han majority ethnic group on gender identity and most were not religious. Similarly, an item with the word “sinful” (negative expectations) was considered strange as this religious concept was not meaningful. For nonaffirmation, participants found an item on pronouns confusing since the words “he” and “she” are phonetically identical in Mandarin Chinese and the word “pronoun” is rarely used in everyday contexts. Finally, two participants expressed emotional distress during victimization items, including tearing up or refusing to answer.

The research team modified the measure based on academic, community, and participant feedback. Response options in the adapted discrimination, rejection, and victimization subscales omitted recency found in the original items to reduce participant burden. The victimization subscale added response options for perpetrator, as CBOs were interested in this information. Given emotional distress during cognitive interviewing, these items also included a “Refuse to answer” option. We modified the wording of other items that participants found difficult to understand. For example, we changed a nonaffirmation item to “I have to repeatedly explain my gender identity to people or correct how people address me” and a negative expectations item on expecting others to find one “sinful” to a culturally relevant term.

Questionnaire participants

A total of 324 participants completed the questionnaire (November 2019–June 2020). We excluded 47 questionnaires from data analysis: 18 submitted from the same IP address, 28 completed in under 5 minutes, and 1 individual under 18 years, resulting in an analytic sample of 277 participants. The mean age was 29.0 years (standard deviation = 7.7, n = 253). The majority of participants (60.6%, 168/277) had an urban residence permit and identified specifically as a woman (58.8%, 163/277). Around 45.5% (126/277) had a high school or below education. Over one-quarter (26.4%, 73/277) reported cigarette use and over half (53.4%, 138/258) scored at risk of clinical depression, respectively (Table 2).

Table 2.

Demographics for a Nine-City Online Sample of Transgender Women, Transfeminine People, and Gender-Diverse People Assigned Male at Birth in Mainland China (n = 277)

Variable	Mean (standard deviation)	Frequency (%)
Age (years, n = 253, range: 18–67 years)	29.0 (7.7)
Gender Identity (n = 277)
Transgender woman		104 (37.5)
CD		37 (13.4)
TS or Yao		39 (14.1)
Woman		59 (21.3)
Gender nonconforming, nonbinary, fluid		37 (13.4)
Other		1 (0.3)
Sexual Attraction (n = 277)
Men only		224 (80.9)
Men and women		23 (8.3)
Women only		15 (5.4)
No one		3 (1.1)
Unsure or questioning		8 (2.9)
Other		4 (1.4)
Residency (n = 277)
Rural		109 (39.4)
Urban		168 (60.6)
Income (Renminbi/month, n = 277)
3000 or below		75 (27.1)
3001–6000		129 (46.6)
6001–10,000		56 (20.2)
More than 10,000		17 (6.1)
Education (highest completed, n = 277)
High school or below		126 (45.5)
Some college		68 (24.5)
College/Bachelor’s		77 (27.8)
Master’s or above		6 (2.2)
Ethnicity (n = 270)
Han		217 (80.4)
Other		53 (19.6)
Sex work involvement, lifetime (n = 275)		116 (42.4)
Cigarette use, past 30 days (n = 277)		73 (26.4)
Depressive symptoms (CES-D 10), 12 or more (n = 258)		138 (53.4)

CD, TS, and Yao refer to local transfeminine identities. CES-D 10 refers to the Center for Epidemiological Studies Depression Scale, 10-item version.

CES-D, Center for Epidemiological Studies Depression.

Descriptive statistics

Frequencies, percentages, and factor loadings for the first-order CFA, and the most frequently endorsed items are reported in Table 1. Missing data for GMSR measure items ranged from 0% to 5.2%. While the distributions of some items were mildly leptokurtic, kurtosis and skewness were within the recommended limits of below 7 and 2, respectively. ³⁹

Internal consistency reliability

Alpha reliabilities of all subscales exceeded the criterion of adequate internal consistency and reliability of 0.70, indicating that the items were likely measuring the same underlying construct ⁴⁰ (Table 1).

Confirmatory factor analyses

Standardized factor loadings ranged from 0.620 to 0.958. All factor loadings were significant (p < 0.001), which indicated that all items were strongly associated with the constructs. Covariance coverage ranged from 0.921 to 1.000. The fit of the first-order measurement model for the seven subscales was good and met criteria for the RMSEA 0.059 (0.055–0.063), the CFI (0.956), and the TLI (0.953). No modifications were made to the first-order model in the interest of parsimony. Following best practices, an alternate second-order model was also examined, but model fit worsened. These findings indicated that the data fit the first-order model better than the second-order model and support Hypothesis 1 (Table 3).

Table 3.

Goodness-of-Fit Statistics for First- and Second-Order Models of Gender Minority Stress (n = 277)

	Model Fit Index
Model	Chi-Square Test of Model Fit (degrees of freedom)	RMSEA (90% Confidence Interval)	CFI	TLI	SRMR
Acceptable values	Nonsignificant	≤0.08	≥0.950	≥0.950	≤0.050
First-order model	1825.61 (924)*	0.059 (0.055–0.063)	0.956	0.953	0.075
Second-order model	2101.033 (938)*	0.067 (0.063–0.071)	0.943	0.940	0.094

^*p < 0.05.

CFI, comparative fit index; RMSEA, root mean square error of approximation; SRMR, standardized root mean square residual; TLI, Tucker–Lewis index.

Criterion validity

All subscales were positively correlated with CES-D 10 scores, with correlations ranging from 0.195 to 0.471, which supports Hypothesis 2 (subscales would be associated with higher depressive symptom scores) (Table 4). In contrast to Hypothesis 3, only nonaffirmation was positively correlated with cigarette use (r = 0.134) (Table 4).

Table 4.

Correlations of Depressive Symptoms and Current Combusted Cigarette Use with the First-Order Model of Gender Minority Stress (n = 277)

	DSC	REJ	VIC	NA	INT	NE	ND	DEP	TOB
DEP	0.208*	0.363*	0.258*	0.226*	0.471*	0.291*	0.195*	1.000
TOB	0.028	−0.020	0.025	0.134*	−0.040	0.028	0.063	−0.012	1.000

^*p < 0.05.

DEP, depressive symptoms; DSC, Discrimination; INT, Internalized Transphobia; NA, Nonaffirmation; ND, Nondisclosure; NE, Negative Expectations; REJ, Rejection; VIC, Victimization; TOB, current combusted cigarette use (past 30 days, nonexperimental use).

Divergent validity

In the first-order model, all subscales were positively and significantly correlated with each other (r = 0.154–0.507) (Table 5). As all correlations were <0.600, these results support Hypothesis 4.

Table 5.

Interfactor Correlations for the First-Order Model of Gender Minority Stress (n = 277)

	DSC	REJ	VIC	NA	INT	NE	ND
DSC	1.000
REJ	0.385*	1.000
VIC	0.206*	0.306*	1.000
NA	0.308*	0.259*	0.215*	1.000
INT	0.262*	0.247*	0.212*	0.367*	1.000
NE	0.340*	0.274*	0.274*	0.465*	0.470*	1.000
ND	0.256*	0.186*	0.154*	0.407*	0.392*	0.507*	1.000

^*p < 0.05.

Discussion

This study provides good evidence for the validity and reliability of our adapted version of the GMSR stress subscales in a sample of primarily transgender women and transfeminine people in mainland China; extends the existing literature by providing insights on culturally specific dimensions of gender minority stress; and strengthens prior evidence for how a gender minority stress measure performs in an understudied population. ²³ Overall, this study’s findings affirm the benefits of community-engaged measurement methods for adapting complex constructs across diverse cultural contexts. ⁴¹

We found that the measure was comprehensive and had face validity after modifications to account for cultural relevance and participant distress. To address difficulties in comprehension arising from lack of cultural relevance, it would be beneficial for researchers to collaborate with community members and experts and integrate the insights of disciplines that examine local conceptions of gender identity. Notably, three out of five initial model fit indices for our scale were adequate, which may reflect the strengths of a community-engaged approach in promoting the cultural relevance of our adapted subscales.

There was mixed evidence for criterion validity in this sample. While correlations with depressive symptoms were consistent with Hypothesis 2 and a prior psychometric study among TGD people in China, ²³ mostly insignificant correlations with cigarette use indicate that gender minority stress may not necessarily contribute to the likely high proportion of cigarette use relative to cisgender women.^8,42 An alternate explanation may be social norms facilitating cigarette use, such as in bars serving LGBT populations, ⁴³ among other LGBT people in one’s social networks, ⁴⁴ or as part of sex work, ⁴⁵ which 42.4% of participants reported ever performing. Further research could assess correlations between tobacco use and community connectedness in informing criterion validity for gender minority resilience, as these findings on social norms in other settings and populations often reflect community connectedness.^43,46

Psychometric findings support the applicability of the subscales to transgender women and transfeminine people in China, but future adaptations could revise items to better reflect the context, improve model fit, and understand the performance of this measure in other Chinese TGD populations. Similar to other studies,^4,23,47,48 discrimination had a low Cronbach’s alpha. During cognitive interviews, some participants may not have framed situations such as accessing medical care as ones where they encountered discrimination, as they were accustomed to presenting as men in these situations. These findings are in line with the high levels of nondisclosure reported by online study participants. Discrimination items may need to be further refined to be more sensitive to local context.

Finally, research on additional TGD populations such as transgender men and transmasculine people, nonbinary people, ethnic minority TGD people, and emigrant Chinese TGD populations would be beneficial to understand how well this measure performs for additional populations and to inform studies on gender minority stress with high numbers of Chinese people outside of mainland China.

Limitations

Due to resource limitations, the study sample was limited to primarily participants across diverse transfeminine identities and results may not be generalizable to other TGD populations. However, the relatively robust sample size provides good evidence for validity of this measure for these participants. Missing data ranged from 0% to 5.2%. Little’s test indicated that data were not missing at random and parameter and standard error estimates may have been biased. ⁴⁹ Finally, we used a four-point Likert scale response option ⁵⁰ for severity items and information on the applicability of five-point Likert scales used in the original measure was unavailable.

Strengths

Strengths of our study arose from community engagement across phases. Substantial feedback and participation from community experts and members provided rich information for cultural adaptation of complex topics and was reflected in the relatively low proportions of refusal for sensitive questions. These relationships also allowed data collection to continue during the initial phase of the COVID-19 pandemic and reach diverse participants.

Conclusion

There is good evidence for validity of the GMSR stress subscales among transgender women and transfeminine people in China. These subscales may allow researchers and public health practitioners to capture robust data on minority stress, understand how specific stressors impact health, and inform interventions.

Authors’ Contributions

W.D.: Conceptualization, methodology, software, formal analysis, investigation, data curation, writing-original draft, supervision, project administration, funding acquisition. Y.S.: Data curation, writing—original draft, supervision, project administration. W.T.: Conceptualization, resources, writing—original draft, supervision, project administration, funding acquisition. J.D.T.: Resources, writing—original draft, supervision, project administration, funding acquisition. W.J.H.: Methodology, writing—original draft, supervision. L.B.H.-W.: Methodology, writing—original draft, supervision. E.F.: Methodology, writing—original draft, supervision. A.F.L.: Methodology, writing—original draft, supervision. X.H.: Methodology, writing—original draft, investigation. K.E.M.: methodology, writing—original draft, supervision.

Abbreviations

CBO

community-based organization

CES-D 10

Center for Epidemiological Studies Depression Scale, 10-item measure

CFA

confirmatory factor analyses

CFI

comparative fit index

COVID-19

coronavirus disease 2019

DEP

depressive symptoms

DSC

discrimination

GMSR

Gender Minority Stress and Resilience

HIV

human immunodeficiency virus

INT

internalized transphobia

IP

internet protocol

IRB

Institutional Review Board

LGBT

lesbian, gay, bisexual, and transgender

NA

nonaffirmation

ND

nondisclosure

NE

negative expectations

REJ

rejection

RMSEA

root mean square error of approximation

SRMR

standardized root mean square residual

TGD

transgender and gender-diverse

TLI

Tucker-Lewis index

TOB

current combusted cigarette use (past 30 days, nonexperimental use)

VIC

victimization

Disclaimer

The content is solely the responsibility of the authors and does not necessarily represent the official views of the funders.

Disclosure Statement

The authors report there are no competing interests to declare.

 

Cite this article as: Dong W, Sha Y, Tang W, Tucker JD, Hall WJ, Hightow-Weidman LB, Fisher E, Lightfoot AF, Huang X, Muessig KE (2026) Cultural adaptation and evaluation of evidence for validity of a gender minority stress measure among transgender women and transfeminine people in China, Transgender Health 11:2, 126–136, DOI: 10.1089/trgh.2024.0028.