Development and Validation of the Gender Identity and Expression Microaggressions in Therapy Scale (GIEMTS)

Abstract

Developing affirming interventions for transgender and nonbinary (TNB) therapy clients requires understanding their experiences with microaggressions in psychotherapy, yet no self-report measure of anti-TNB microaggressions in this context exists. Moreover, few studies have tested the associations between anti-TNB microaggressions and therapy processes. To better address the burden of unmet mental healthcare needs among TNB people, this three-study investigation designed and tested the psychometric properties of the Gender Identity and Expression Microaggressions in Therapy Scale (GIEMTS), a measure of TNB individuals’ encounters with microaggressions in psychotherapy. Study 1 (N = 225) identified a four-factor model, comprising the themes of Educational Burdening, Lack of Affirmation, Inflation, and Invalidation. These subscales exhibited strong internal consistency reliabilities and demonstrated convergent and discriminant validity. The results of Study 2 (N = 435) replicated the four-factor structure through confirmatory factor analysis. However, bifactor analysis revealed that the Educational Burdening, Inflation, and Invalidation subscale scores were mostly accounted for by a General Anti-TNB Microaggressions scale score–though Lack of Affirmation showed evidence of its independence. Also in Study 2, both scales were uniquely negatively associated with the working alliance. Study 3 (N = 151) found evidence for the test-retest reliability of GIEMTS scores over a two-to-three-week period. Overall, the GIEMTS emerged as a robust and psychometrically sound instrument that captures the experiences of TNB individuals in therapy settings. The study concludes with valuable recommendations for training and clinical practice to bolster TNB mental health.

Microaggressions–or subtle verbal or non-verbal insults or humiliations that are often perpetrated unintentionally–are pervasive in the lives of marginalized populations (Sue et al., 2007). These everyday exchanges, which communicate identity-related hostility, are associated across minoritized groups with negative mental health outcomes, such as depression, anxiety, hypervigilance, internalized stigma, and even suicidality (Nadal, 2019). Transgender and/or nonbinary (TNB) people encounter considerable structural and interpersonal discrimination (Breslow et al., 2015; Testa et al., 2015). TNB people report a range microaggressions, including encountering transphobic language or being directly asked invasive questions about their body (Chang & Chung, 2015). Though microaggressions have primarily been studied in the contexts of work and school, they also occur in the therapy room (Owen et al., 2014). A microaggressions framework resonates well with examinations of the therapeutic interaction: because therapists are invested in clients’ well-being—and thus not likely to intentionally seek to offend clients—manifestations of bias in psychotherapy are likely subtle and/or unintentional.

Across studies with diverse populations, experiences of microaggressions in psychotherapy are associated with poorer ratings of the working alliance (Owen et al., 2014; Owen et al., 2010) and higher likelihood of early termination (Dowshen et al., 2017). They may also cause erode clients’ trust in healthcare systems (Shelton & Delgado-Romero, 2011). Given the significant mental health needs of TNB populations (e.g., Reisner et al., 2016), it is vital for therapists to be aware of the ways that anti-TNB microaggressions manifest in their work. To this end, the present study developed and tested the psychometric properties of a scale of anti-TNB microaggressions in psychotherapy.

The Mental Healthcare Needs of Transgender and Nonbinary People

Despite strides in visibility and representation, TNB people continue to be at elevated risk for adverse mental health outcomes, including depression, anxiety, and suicidality (e.g., Reisner et al., 2016). For example, in a national study of 27,715 TNB adults, 40% of participants reported having attempted suicide at least once during their lifetime (James et al., 2016)—roughly nine times the estimate for the general population. From the perspective of minority stress theory (Brooks, 1981; Meyer, 2003), these mental health concerns are partially due to the noxious influence of anti-TNB stressors, such as structural and interpersonal discrimination, internalized transphobia, and expectations of rejection. Indeed, empirical research has documented the association of such stressors with poorer mental health among TNB populations (e.g., Breslow et al., 2015; Testa et al., 2015). In turn, pronounced levels of distress may contribute to high healthcare service utilization among TNB people (Kachen & Pharr, 2020)—despite persistent structural barriers to accessing healthcare, such as prohibitive healthcare costs, suboptimal insurance coverage rates, compartmentalization of gender-related care in siloed departments, and a lack of affirmative training for therapists (Safer et al., 2016).

TNB people may also pursue mental healthcare as part of their pursuit of gender-affirming procedures. Version 8 of the Standards of Care for the Health of Transsexual, Transgender, and Gender Nonconforming People (SOC-8; Coleman et al., 2022) recommends that clients attain a letter of support from healthcare providers–who are often mental health professionals–in order to access gender-affirming medical or surgical interventions. Notably, the SOC-8 do not require such letters. Furthermore—though the SOC-8 notes that psychotherapy may be beneficial for TNB people pursuing gender affirming interventions, it also indicates that it should not be required. Nonetheless, clinicians and insurance companies who are unfamiliar with the SOC-8 may continue to require that TNB people attain letters of support or undergo psychotherapy before approving their access to gender affirming interventions. Within the context of this gatekeeping, it is particularly important to ensure that mental healthcare affirms TNB clients.

The Experiences of Transgender and Nonbinary Clients in Psychotherapy

Unfortunately, compared to cisgender people, TNB clients in psychotherapy are disproportionately likely to report experiencing and anticipating stigma from a healthcare provider (White Hughto et al., 2015). Furthermore, many TNB clients who rely on mental healthcare for access to medical or surgical gender affirmation report interpersonal discrimination from therapists, which may contribute to high attrition and dropout (Dowshen et al., 2017). Over the last decade, researchers have employed qualitative methods to describe TNB people’s experiences in counseling or psychotherapy (Anzani et al., 2019; Mizock & Lundquist, 2016; Morris et al., 2020). Much of this work has emphasized clients’ perceptions of negative experiences that may be construed as microaggressions (e.g., Mizock & Lundquist, 2016; Morris et al., 2020). Across studies, prominent themes include therapists invalidating or disrespecting clients’ gender identity (e.g., being repeatedly “deadnamed,” or called by one’s name given at birth rather than one’s chosen name); therapists depending on clients to educate them on TNB-related issues (educational burdening); therapists exaggerating the importance of clients’ gender identities relative to other, more salient topics (gender inflation); therapists expressing preconceived notions about TNB people or gender identity, such as belief in the gender binary (gender narrowing); therapists framing TNB identities as expressions or consequences of psychopathology (gender pathologizing) in need of repair (gender repairing); or gatekeeping, instances in which therapists limit, delay, or prevent access to gender-affirming healthcare (e.g., Mizock & Lundquist, 2016; Morris et al., 2020).

At times, anti-TNB sentiment was expressed by what was unsaid. For example, therapists may not make blatantly discriminatory statements yet may withhold support for access to care (e.g., by encouraging a client to ‘wait and see’ how their identity may develop before writing a letter of support for gender affirming interventions) or avoid the topic of gender altogether (gender avoidance) (Mizock & Lundquist, 2016; Morris et al., 2020). Thus, the absence of positive therapist behaviors may also have a negative impact on TNB clients. This idea converges with research on TNB microaffirmations, or subtle gestures that validate and celebrate TNB identities (e.g., Anzani et al., 2019). Anzani and colleagues (2019) found that TNB microaffirmations in psychotherapy may foster a stronger therapeutic alliance, cultivate trust, and enhance perceived treatment efficacy. Examples of TNB-specific microaffirmations include disrupting cisnormativity (celebrating clients’ pursuit of gender affirmation) and seeing authentic gender identity (acknowledging the legitimacy of clients’ gender identities). The absence of microaffirmations in therapy may be construed by clients as a form of microaggression.

The need to measure and mitigate anti-TNB microaggressions is especially pressing in the contemporary political landscape. TNB communities, who already face well-documented, crisis-level rates of violence and mental health concerns (James et al.., 2016), must also contend with the enactment of anti-TNB legislation that serves as structural manifestations of anti-TNB stigma (Human Rights Campaign [HRC], 2023). Such legislation often restricts access to gender-affirming care, thereby exacerbating existing disparities in mental health services. Anti-TNB microaggressions in psychotherapy may compound the negative impact of this structural stigma on the mental health of TNB clients.

Present Study

The present investigation had two overarching aims. First, we sought to develop the Gender Identity and Expression Microaggressions in Therapy Scale (GIEMTS), a self-report measure of TNB individuals’ experiences of anti-TNB microaggressions in therapy. In order to maximize content validity, we generated items that reflected themes observed in prior qualitative research (e.g., Mizock & Lundquist, 2016; Morris et al., 2020). We then solicited feedback from expert reviewers to enhance confidence that GIEMTS items were relevant to the clinical experiences of TNB clients.

Our second aim was to test the psychometric properties of GIEMTS. To this end, we collected and analyzed survey data collected online from three samples of TNB adults in the United States (U.S.) who had received psychotherapy in the past 12 months. Data were used to test the GIEMTS’s structural validity (Studies 1 and 2); convergent and discriminant validity (Study 1); concurrent and incremental validity (Study 2); internal consistency reliability (Studies 1, 2, and 3); and test-retest reliability (Study 3).

Study 1: Instrument Development and Initial Psychometric Validation

Study 1 consisted of instrument development and initial psychometric validation. To maximize the content validity of the GIEMTS, scale items were based on themes reported in extant research on TNB people’s experiences in psychotherapy. Furthermore, GIEMTS items were evaluated by expert reviewers to ensure that they captured the construct of interest. Next, online survey data was collected from a sample of TNB people who had undergone psychotherapy within the last 12 months to a) explore the underlying factor structure of the GIEMTS; b) collect evidence of the GIEMTS’s internal consistency reliability; and c) to provide initial tests of the GIEMTS’s convergent and discriminant validity.

It was hypothesized that the GIEMTS factor structure–identified through exploratory factor analysis (EFA)–would be multidimensional and reflect the themes observed in TNB people’s experiences of microaggressions in psychotherapy as reported in prior qualitative research (Anzani et al., 2019; Mizock & Lundquist, 2016; Morris et al., 2020). It was also hypothesized that the GIEMTS would demonstrate adequate internal consistency reliability. Regarding convergent validity, it was expected that GIEMTS scores, conceptualized as a manifestation of anti-TNB discrimination in therapy, would be positively correlated with participants’ experiences of general anti-TNB discrimination–based on the idea that experiences of anti-TNB bias in a specific context (i.e., psychotherapy) partially inform judgments of the anti-TNB stigma encountered across contexts. Because therapists’ propensity to enact microaggressions in psychotherapy may reflect, in part, a lack of general competence or a lack of cultural humility, it was anticipated that GIEMTS scores would be negatively correlated with measures of these two constructs. Finally, given the reality of widespread anti-TNB sentiment–as reflected in the ubiquity of anti-TNB legislation (HRC, 2023)–we did not expect that participants would be motivated to exaggerate the extent to which they have experienced anti-TNB microaggressions in psychotherapy to receive social validation. Thus, we hypothesized that GIEMTS scores would be unrelated to measures of socially desirable responding.

Method

Participants

Data were analyzed from 225 TNB adults who ranged in age from 18 to 63 years old (M = 27.53, SD = 8.25, Mdn = 26). One-hundred eighty-eight (83.6%) participants were currently receiving psychotherapy, whereas 37 (16.4%) had received psychotherapy within the last 12 months. Additional demographic characteristics for Sample 1 are presented in Table 1.

Table 1.

Demographics of Samples in Studies 1, 2, and 3

	Study 1 (N = 225)		Study 2 (N = 435)		Study 3 (N = 151)
	n	%	n	%	n	%
Gender Identity
Agender	16	7.1	13	3.0	10	6.6
Genderqueer	22	9.8	133	30.6	30	19.9
Nonbinary	65	28.9	113	26.0	38	25.2
Transgender men	69	30.7	73	16.8	44	29.1
Transgender women	29	12.9	71	16.3	17	11.3
Two Spirit	1	0.4	12	2.8	0	0
Another gender minority identity	23	10.2	20	4.6	12	7.9
Race/ethnicity
African American or Black	6	2.7	19	4.4	6	4.0
American Indian or Native American	4	1.8	12	2.8	7	4.6
Arab, Middle Eastern, or North African	1	0.4	4	0.9	4	2.6
Asian American or Pacific Islander	5	2.2	21	4.8	7	4.6
European American or white	182	80.9	250	57.5	107	70.9
Hispanic or Latinx	6	2.7	102	23.4	9	6.0
Multiracial	9	4.0	21	4.8	7	4.6
Another racial/ethnic identity	7	3.1	4	0.9	3	2.0
Sexual identity
Asexual	23	10.2	20	4.6	10	6.6
Bisexual	40	17.8	82	18.9	37	24.5
Heterosexual	7	3.1	40	9.2	16	10.6
Lesbian or gay	38	16.9	59	13.6	17	11.3
Pansexual	32	14.2	81	18.6	19	12.6
Queer	64	28.4	117	26.9	37	24.5
Questioning	4	1.8	12	2.8	4	2.6
Same-gender loving	1	0.4	11	2.5	2	1.3
Another sexual identity	11	4.9	11	2.5	8	5.3
Highest level of education
Some high school	8	3.6	6	1.4	2	1.3
High school or GED	22	9.8	40	9.2	14	9.3
Some college	54	24.0	124	28.5	32	21.2
Associate’s degree	10	4.4	55	12.6	24	15.9
Bachelor’s degree	63	28.0	135	31.0	36	23.8
Some postgraduate education	19	8.4	30	6.9	17	11.1
Postgraduate degree (e.g., M.S., M.D.)	44	19.6	42	9.7	25	16.6

Note. Column ns may not sum to the full sample size because of small levels of missing data.

Measures

Gender Identity and Expression Microaggressions in Therapy.

An initial 54-item pool was generated that reflected the content of anti-TNB microaggressions TNB clients reported experiencing in psychotherapy (Mizock & Lundquist, 2016; Morris et al., 2020). Themes represented in the initial pool of items included educational burdening, gender inflation, gender narrowing, gender avoidance, gender pathologizing, gender repairing, and gatekeeping.

We next solicited feedback from a panel of 13 psychologists (some of whom identify as TNB people) with research and clinical expertise working with TNB communities. Eight of these individuals responded to our survey and provided anonymous feedback. Expert reviewers were asked to rate the scale instructions, each of the 54 preliminary GIEMTS items, and the item response scale regarding clarity, cultural appropriateness, and overall quality using a five-point Likert-type scale ranging from 1 (terrible) to 5 (excellent). Additionally, reviewers had the option of providing qualitative feedback to supplement their ratings.

GIEMTS items that yielded an average rating of less than 4.5 in any of the three rating dimensions (n = 31) were flagged for revision or removal. Of these 31 items, 17 were revised using the open-ended feedback as a guide. Revisions often focused on using more culturally salient terminology or enhancing specificity of item content. A good example was item 8, which was revised from “Needed me to inform them about the process of medically transitioning” to “Needed me to educate them about the process of getting gender-affirming medical procedures (such as taking hormones or hormone blockers, getting top or bottom surgery, etc.).” An additional 14 items were completely removed from the item pool. We also added 14 items to capture areas that were not as strongly represented in the initial item pool. Specifically, we generated items that reflected TNB microaffirmations (e.g., seeing clients’ authentic gender).

The procedures above resulted in a total of 54 GIEMTS items, which are presented in Supplemental Materials A. Participants indicated their agreement with each item using a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Participants also had the option to respond “not applicable” if an item was not relevant to their lived experience. Not applicable responses were coded as missing in subsequent analyses. Items intended to reflect microaffirmations were reverse-scored so that higher scores reflected less affirmation.

General Anti-TNB Discrimination.

Mistreatment, derogation, harassment, or rejection due to one’s TNB identity (outside of the therapeutic context) was assessed with the 21-item Transgender Discrimination Scale-21 (TDS-21; Watson et al., 2019). Participants indicated how frequently each TDS-21 item (e.g., “Experienced harassment from family members”) occurred to them in their lifetime using a 6-point rating scale (1 = Never to 6 = Almost all of the time). Item responses were averaged to derive an overall scale score, with higher scale scores indicating more frequent experiences of discrimination. In support of construct validity, TDS-21 scores were positively associated with scores on another measure of anti-TNB discrimination and with expectations of rejection (Watson et al., 2019). This same study found that Cronbach’s alpha for the TDS-21 was .89. Cronbach’s alpha for the present sample was .92.

Therapist General Competence.

The 12-item Counselor Rating Form—Short Form (CRF-SF; Corrigan & Schmidt, 1983) was used to assess participants’ perceptions of their therapists’ overall competence. Participants rated the extent to which they believed 12 adjectives (e.g., “Sincere,” “Trustworthy,” “Warm”) characterized their therapists using a 7-point rating scale (1 = Not very to 7 = Very). Item responses were averaged to derive an overall scale score, with higher scores indicating greater perceived therapist competence. In a study in which TNB people evaluated videos of mock therapy sessions between a TNB client and a therapist, therapists who engaged in non-affirming behavior (e.g., pressuring the client to identify with their sex-assigned-at-birth) received lower CRS-SF scores than therapists who affirmed clients’ identities (Bettergarcia & Israel, 2018). This same study found that Cronbach’s alpha for CRS-SF items was .98. Cronbach’s alpha in the present sample was .95.

Therapist Cultural Humility.

The 12-item Cultural Humility Scale (CHS; Hook et al., 2013) was used to assess participants’ perceptions that their therapists were other-oriented and respectful of their cultural backgrounds. Participants indicated the extent to which their therapist demonstrated cultural humility (e.g., “Is genuinely interested in learning more”) toward participants’ core cultural background(s) using a 5-point rating scale (1= Strongly disagree to 5 = Strongly agree). Appropriate items were reverse-scored and item responses were averaged to derive a full-scale score, with higher scores indicating greater therapist cultural humility. In four samples of adults in psychotherapy, CHS scores were positively associated with perceptions of therapists’ multicultural competence and the working alliance (Hook et al., 2013). Across samples in the same study, Cronbach’s alpha for CHS items ranged from .86 to .93. CHS items yielded a Cronbach’s alpha of .93 in the present sample.

Socially Desirable Responding.

The 24-item Balanced Inventory of Desirable Responding-Short Form (BIDR-SF; Asgeirsdottir et al., 2016) was used to assess the extent to which participants demonstrated two socially desirable response styles: self-deceptive enhancement (SDE; providing honest but overly positive responses) and impression management (IM; intentional presentation of the self to an imagined audience). Participants indicated the extent to which they agreed with each SDE (e.g., “I never regret my decisions”) and IM (e.g., “I don’t gossip about other people’s business”) item using a 7-point rating scale (1 = Not true to 7 = Very True). Appropriate items were reverse-scored and SDE and IM items were averaged to derive overall scale scores, with higher SDE and IM scores indicating greater socially desirable responding. In a sample of Icelandic undergraduate students, instructions encouraging participants to fake their responses resulted in higher scores on both the SDE and IM subscales of the BIDR-SF (Asgeirsdottir et al., 2016). In the same study, SDE and IM items yielded Cronbach’s alphas of .71 and .73, respectively. In the present sample, SDE and IM items yielded Cronbach’s alphas of .76 and .67, respectively.

Procedures

Institutional review board approval was attained for all phases of this study. A purposeful snowball sampling method was used to recruit TNB adults who had received mental health therapy within the past 12 months. Participants were recruited via messages sent to TNB-focused listservs and online communities (e.g., subReddits) that included a description of the study and a link to the Qualtrics hosted survey. To complete the survey, participants needed to provide informed consent and confirm that they met all inclusion criteria. At the start of the survey, participants were informed that they would encounter validity check items in the survey (e.g., “Please select ‘Disagree Strongly’”) that assessed attentive responding. Participants who completed the survey could enter a raffle with a 1/25 chance to win a $50 Amazon.com gift card. In addition, participants were informed that for each completed survey, $1 would be donated to the National Center for Transgender Equality (https://transequality.org/), a non-profit organization that advocates for TNB people. Participants encountered the GIEMTS first and the demographics questionnaire last; all other instruments were presented in random order.

A total of 566 individuals provided consent to participate in the study. Of these individuals, 310 (approximately 55%) were removed from the dataset because they stopped the survey after providing consent (likely because they did not meet inclusion criteria). An additional 16 cases (3%) were removed because demographic questionnaire responses indicated that they did not meet inclusion criteria. One case (< 1%) was removed for having a duplicate IP address and three cases (1%) were removed because they missed more than two of the four validity check items inserted in the survey. Finally, 11 cases were removed for missing more than 20% of the data analyzed in the study (Parent, 2013), which resulted in a final sample of 225 cases (40%). Comparable retention rates have been documented in other online surveys with TNB populations (e.g., Cascalheira & Choi, 2023). Only 12 cases (5%) were missing any data (including “not applicable” GIEMTS responses that were recoded as missing). Little’s Missing Completely at Random (MCAR) test was nonsignificant, χ²(13695) = 13,820.78, p = .207, which suggested that the data were MCAR. Because of the low levels of missingness and the data, available item analysis was used in SPSS and maximum likelihood estimation was used in Mplus v. 8.7 (Muthén & Muthén, 2012–21).

Results

Exploratory Factor Analysis

We first conducted an exploratory factor analysis (EFA) to examine the underlying structure of associations between items. An initial principal components analysis was run in IBM SPSS Statistics (Version 27). The GIEMTS items were suited for factor analysis and approximated multivariate normality, as indicated by a Kaiser-Meyer-Olkin value of .88 (Tabachnick & Fidell, 2007) and significant Bartlett’s test of sphericity (p < .001). Nine components yielded eigenvalues greater than 1. Scree plots suggested three or five factors should be extracted. A parallel analysis was performed in SPSS using syntax developed by O’Connor (2000). The results of the parallel analysis suggested no more than five factors should be extracted. Subsequent analyses examined the three-, four-, and five-factor solutions.

Next, a series of EFAs were conducted in Mplus v. 8.6 (Muthén & Muthén, 2012–21). These analyses were conducted with maximum likelihood estimation with standard errors and chi-square test statistics that are robust to nonnormality (MLR). The analyses used principal axis factoring with oblique promax rotation because we expected factors to be correlated. Model fit was assessed with the root mean square error of approximation (RMSEA) and the standardized root mean residual (SRMR), with good model fit indicated by RMSEA and SRMR < .08 (Bentler & Bonett, 1980). The fit for the three-factor model was S-B χ² (1272) = 3120.39, p < .001, RMSEA = .080, 90% CI (.077, .084), SRMR = 0.05, which approached—but did not yield—acceptable fit. The fit for both the four-factor model, S-B χ² (1221) = 2794.57, p < .001, RMSEA = .076, 90% CI (.072, .079), SRMR = 0.05, and the five-factor model, S-B χ² (1171) = 2510.15, p < .001, RMSEA = .071, 90% CI (.067, .075), SRMR = 0.04, were acceptable.

In deciding between the three-, four- and five- factor solutions, we used the following criteria (Tabachnick & Fidell, 2007; Worthington & Whittaker, 2006): there had to be at least three items assigned to a factor; evidence of a clear simple structure, with a preference for solutions that minimized the number of cross-loadings (operationalized as a difference between the two highest pattern coefficients of at least .20); and clear conceptual interpretations, meaning that items assigned to the same factor reflected a clear, common theme. The four-factor solution met all criteria, so it was adopted.

Pattern coefficients were used to interpret the factors in the four-factor solution. The first factor comprised items in which clients believed they needed to educate therapists about TNB issues (Educational Burdening). The second factor contained items indicating that therapists did not engage in affirming behaviors toward TNB clients (Lack of Affirmation). The third factor included perceptions that therapists focused on the clients’ gender identity rather than other, more clinically relevant topics (Inflation). Items in the fourth factor reflected clients’ belief that therapists denied the legitimacy of TNB identities or avoided discussing gender (Invalidation).

To both reduce scale length and enhance construct specificity distinctions between factors, we removed items whose highest factor loadings were less than .50, meaning that less than 25% of the variance in the item was accounted for by the factor (Tabachnick & Fidell, 2007). If factors contained more than ten items, we kept only those ten items with the highest factor loadings to avoid excessive scale length (Kase & Mohr, 2022). To maximize the distinction between factors, we also removed items that had cross-loadings smaller than .20. This process resulted in the removal of 24 items, resulting in a final item pool of 30. The 30-item version of the GIEMTS is presented in Supplemental Materials B. We then reran the EFA with the reduced item pool to determine if the factor structure changed. The four-factor solution with the reduced item pool yielded acceptable fit, S-B χ² (321) = 549.76, p < .001, RMSEA = .056, 90% CI (.048, .064), SRMR = 0.03.

Descriptive statistics, factor loadings, and communalities for the final 30-item GIEMTS are presented in Table 2. Cronbach’s alphas for the subscales and the total scale were computed in SPSS. The eight-item Educational Burdening subscale (e.g., “Placed the burden on me to educate them on the experiences of TNB people”) yielded a Cronbach’s alpha of .95. The ten-item Lack of Affirmation subscale (e.g., “Affirmed my gender identity or expression during sessions” [reverse-scored]) yielded a Cronbach’s alpha of .93. The three-item Inflation subscale (e.g., “Brought up gender identity or expression even when it was unrelated to what I wanted to talk about”) yielded a Cronbach’s alpha of .84. The nine-item Invalidation subscale (e.g., “Made me feel like I needed to change my gender identity or expression”) yielded a Cronbach’s alpha of .94. The Cronbach’s alpha for the 30-item total GIEMTS scale was .96. Each of the subscales of the GIEMTS and the total scale yielded strong internal consistency reliabilities.

Table 2.

EFA Factor loadings and Communalities for 30-item GIEMTS in Study

	Descriptive statistics		Factor analysis structure coefficients
Item	M	SD	1	2	3	4	h 2
Factor 1: Educational Burdening
9. Needed me to educate them about the process of legally transitioning (e.g., changing my name and gender marker on my ID, changing the sex on my birth certificate, etc.).	2.89	2.03	.91	−.08	.06	−.08	.84
8. Needed me to educate them about the process of getting gender-affirming medical procedures (such as taking hormones or hormone blockers, getting top or bottom surgery, etc.).	3.24	2.26	.88	−.09	.06	−.01	.79
3. Needed me to educate them about TNB issues.	3.50	2.28	.86	.06	−.06	−.02	.75
7. Needed me to educate them about the process of socially transitioning (such as using a new name or pronoun).	2.67	2.16	.80	.10	−.02	.06	.65
5. Placed the burden on me to educate them on the experiences of TNB people.	2.52	1.99	.75	.12	.05	.04	.59
6. Demonstrated that they did not have good enough training in issues affecting TNB people.	3.00	2.21	.72	.18	−.08	.13	.58
4. Demonstrated that they did not have basic knowledge about TNB identities and experiences.	2.76	2.12	.68	.21	−.03	.10	.51
48. Did not know how to help me access gender-affirming medical treatment or services.	3.21	2.11	.59	.15	.04	.02	.37
Factor 2: Lack of Affirmation
20. Encouraged me to explore my gender identity or expression (r).	2.77	1.92	−.08	.84	.07	−.03	.72
24. Encouraged me to talk about my gender identity or expression if I wanted to (r).	2.01	1.59	−.05	.79	.05	−.01	.62
51. Expressed pride or joy when I reached a milestone in my gender exploration (r).	2.48	1.77	−.05	.78	−.00	.12	.63
50. Stated that exploring my gender identity or expression was normal (r).	2.57	1.80	.06	.76	−.03	.04	.59
45. Helped me access gender-affirming care or resources (r).	2.92	2.00	.14	.73	−.02	−.02	.56
34. Affirmed my gender identity or expression during sessions (r).	2.15	1.64	.06	.73	−.07	.09	.55
47. Actively sought out more information about TNB identities in order to work more effectively with me (r).	3.57	2.05	.10	.69	.14	−.17	.54
37. Indicated that my gender identity or expression was healthy (r).	2.55	1.79	−.07	.65	.01	.19	.46
13. Discussed gender identity or expression at relevant times (r).	2.35	1.59	.06	.59	−.06	.09	.36
53. Voiced frustration that society makes things so difficult for TNB people like me (r).	2.90	1.90	.16	.54	−.07	−.04	.32
Factor 3: Inflation
12. Brought up gender identity or expression even when it was unrelated to what I wanted to talk about.	2.73	1.90	−.03	.03	.81	−.02	.66
41. Focused on gender dysphoria when it wasn’t relevant in session.	2.09	1.65	.03	−.06	.81	.10	.67
14. Overlooked other important aspects of my life beyond my gender identity or expression.	2.28	1.79	.01	.18	.54	.16	.35
Factor 4: Invalidation
36. Implied my gender identity or expression was just a phase.	1.81	1.67	.02	−.09	.03	.92	.85
33. Implied my gender identity or expression was invalid.	1.88	1.70	−.02	.12	−.11	.86	.76
35. Made me feel like I needed to change my gender identity or expression.	1.83	1.51	−.01	−.01	.07	.83	.69
19. Pressured me to adopt a particular gender identity.	1.65	1.43	−.04	.16	−.11	.79	.66
52. Told me that I was lesbian, gay, or bisexual rather than TNB.	1.54	1.35	−.00	−.03	.04	.78	.61
40. Implied that my gender identity or expression needed to conform to society’s norms.	1.89	1.54	.04	.09	.05	.69	.48
21. Suggested gender identity or expression and sex assigned at birth were the same thing.	1.73	1.39	.07	.06	.24	.55	.37
27. Changed the subject whenever I brought up gender.	1.75	1.45	.08	.19	.06	.53	.32
16. Called me by the wrong name (for example, my dead name, a different version of my name, etc.)	1.78	1.61	.12	.07	.09	.52	.30

Note. N = 225. TNB = Transgender and/or nonbinary. GIEMTS = Gender Identity and Expression Microaggressions in Therapy Scale. Boldface type indicates that item was assigned to that factor. r = reverse-scored item. h² = communality. Factor 1 = Educational Burdening, Factor 2 = Lack of Affirmation, Factor 3 = Inflation, Factor 4 = Invalidation.

Associations Among GIEMTS Scales

Bivariate correlations among and descriptive statistics for the GIEMTS scales in Study 1 are presented in Table 3. Cohen’s (1992) benchmarks were used to characterize the magnitude of correlations as small (r = .10), medium (r = .30), or large (r = .50). Correlations among the Educational Burdening, Lack of Affirmation, Inflation, and Invalidation subscales were all significant, positive, and medium-to-large in magnitude. The correlations of the Educational Burdening, Lack of Affirmation, Inflation, and Invalidation subscales with the GIEMTS total scale were all significant, positive, and large.

Table 3.

Bivariate Correlations Among and Descriptive Statistics and Cronbach’s Alphas for Measures of Convergent and Discriminant Validity in Study 1

Measure	1	2	3	4	5	6	7	8	9	10
1. Educational Burdening	__
2. Lack of Affirmation	.68***	__
3. Inflation	.46***	.47***	__
4. Invalidation	.59***	.65***	.61***	__
5. GIEMTS Total	.87***	.89***	.66***	.84***	__
6. Anti-TNB Discrimination	.34***	.33***	.36***	.45***	.43***	__
7. General Competence	−.56***	−.64***	−.63***	−.67***	−.73***	−.34***	__
8. Cultural Humility	−.57***	−.71***	−.66***	−.76***	−.79***	−.42***	.83***	__
9. Self-Deceptive Enhancement	.07	.04	−.10	.05	.03	.04	−.07	−.05	__
10. Impression Management	.04	.07	−.08	.02	.03	.00	−.03	−.02	.35***	__
Possible Range	1–7	1–7	1–7	1–7	1–7	1–6	1–7	1–5	1–7	1–7
M	3.00	2.62	2.38	1.77	2.43	2.24	5.85	4.13	3.65	3.83
SD	1.88	1.41	1.55	1.26	1.28	0.72	1.15	0.81	0.84	0.85
⍺	.95	.93	.84	.94	.96	.92	.95	.93	.76	.67

Note. N = 225. GIEMS-T = Gender Identity and Expression Microaggressions in Therapy Scale. TNB = Transgender and/or nonbinary.

^***p < .001.

Convergent and Discriminant Validity

Second, we tested the convergent and discriminant validity of the GIEMTS. Bivariate correlations of the GIEMTS scales with the other variables of interest are presented in Table 3. In support of convergent validity, the four subscales and the GIEMTS total scale each yielded medium positive correlations with general anti-TNB discrimination (rs ranging from .34 to .43).

In further support of convergent validity, the four GIEMTS subscales and the total scale yielded large significant negative correlations with therapist general competence and therapist cultural humility. Though the correlations of the GIEMTS scales with general competence and cultural humility were clearly large–with rs ranging from −.56 to −.73 for general competence and from −.57 to −.79 for cultural humility–they all fell below the rule of thumb cut-off (r = |.85|) generally used in scale development research for triggering concern about discriminant validity (Rönkö & Cho, 2020). These results suggest that the GIEMTS subscales assess constructs that are similar to–but likely distinct from–general anti-TNB discrimination, general competence, and cultural humility. As further support for discriminant validity, the correlations of GIEMTS subscale and total scale scores with self-deceptive enhancement and impression management were all trivial to small in magnitude (rs ranging from .02 to |−.10|) and nonsignificant.

Study 2: Confirmation of Factor Structure and Additional Psychometric Evaluation

The first goal of Study 2 was to determine if the four-factor structure for the GIEMTS observed in Study 1’s sample would be replicated in an independent sample. To this end, we conducted a series of confirmatory factor analyses (CFAs) that compared four possible factor structures. The second goal of the study was to collect further evidence for the internal consistency reliability of the GIEMTS. A final goal of Study 2 was to test the concurrent and incremental validity of the GIEMTS. Prior research with TNB samples has documented positive correlations between discrimination or microaggressions and both expectations of rejection (belief that one’s group is devalued and that one will encounter discrimination in the future) and internalized oppression (prejudiced attitudes towards one’s oneself as a member of a marginalized group) (e.g., Breslow et al., 2015; Cascalheira & Choi, 2023). Thus, it was hypothesized that GIEMTS scores would be positively associated with these two constructs. Lastly, following prior research focused on sexist and racist microaggressions, (Owen et al., 2014; Owen et al., 2010), it was hypothesized that GIEMTS scores would be negatively associated with participants’ perception of the working alliance–even after controlling for potential confounds like participants’ pretherapy functioning, their perceived improvement in functioning due to psychotherapy, and their current psychological distress.

Method

Participants

Data were analyzed from 435 TNB adults who ranged in age from 18 to 74 years old (M = 29.06, SD = 6.65, Mdn = 28). Four-hundred sixteen (95.6%) participants were currently receiving psychotherapy, whereas 19 (4.4%) had received psychotherapy within the last 12 months. Additional demographic characteristics for Sample 2 are presented in Table 1.

Measures

Gender Identity and Expression Microaggressions in Therapy.

The 54-item GIEMTS item pool was used to assess gender identity and expression-related psychotherapy microaggressions participants had experienced within the prior 12 months.

Expectations of Rejection.

The nine-item Negative Expectations for the Future subscale (NEFS) of the Gender Minority Stress and Resilience Measure (GRSRM; Testa et al., 2015) was used to assess participants’ belief that others will reject them in the future due to their TNB identities. Participants responded to items (e.g., “If I express my gender identity/history, most people would look down on me”) using a 5-point rating scale (1 = Strongly disagree to 5 = Strongly agree). Responses were averaged into a total scale score, with higher scores indicating stronger expectations of rejection. In support of validity, NEFS scores were positively associated with anti-TNB discrimination in a sample of TNB adults (Testa et al., 2015). In the same sample, Cronbach’s alpha for NEFS items was .89. Cronbach’s alpha was .86 in the current sample.

Internalized Transphobia.

The eight-item Internalized Transphobia subscale (ITS) of the GRSRM (Testa et al., 2015) was used to assess participants’ negative attitudes and beliefs toward their own TNB identities. Participants responded to ITS items (e.g., “I resent my gender identity or expression”) using a 5-point rating scale (1 = Strongly disagree to 5 = Strongly agree). Item responses were averaged, with higher scores indicating greater internalized transphobia. In support of validity, ITS scores were significantly positively associated with nondisclosure of TNB identity and negatively associated with pride in one’s TNB identity in a sample of TNB adults (Testa et al., 2015). In the same sample, ITS items yielded a Cronbach’s alpha of .91. Cronbach’s alpha in the present sample was .87.

Working Alliance.

The 12-item short form of the Working Alliance Inventory (WAI-SF; Tracey & Kokotovic, 1989) was used to assess participants’ evaluation of the quality of their relationship with their therapist as well as the extent to which they and their therapist agreed on the tasks and goals of treatment. Participants responded to WAI-SF items (e.g., “My therapist and I were working towards mutually agreed upon goals”) using a 7-point scale (1 = Strongly disagree to 7 = Strongly agree). Item responses were averaged into a total scale score, with higher scores indicating a stronger working alliance. In a sample of Black adults receiving psychotherapy, WAI-SF scores were positively associated with perceived improvement in psychotherapy (Hook et al., 2013). In four samples of individuals who were currently undergoing psychotherapy, WAI-SF scores ranged from .92 to .96 (Hook et al., 2013). WAI-SF items in the current study yielded a Cronbach’s alpha of .89.

Improvement in Psychotherapy.

The 16-item Patient’s Estimate of Improvement (PEI; Hatcher & Barends, 1996) was used to assess participants’ perceptions that psychotherapy enhanced their functioning. Participants responded to items (e.g., “To what extent have your original complaints or symptoms improved?”) using a 5-point scale (1 = Not very much at all to 5 = Very much). Item responses were averaged into a total scale score, with higher scores indicating greater perceived improvement. In support of the validity, PEI scores were positively correlated with clients’ perception of the working alliance in a sample of adults undergoing psychotherapy (Hook et al., 2013). PEI items yielded a Cronbach’s alpha of .95 in the same study. Cronbach’s alpha for PEI items was .92 in the current study.

Perceived Pretherapy Functioning.

A single item (“How were you feeling when you started counseling?”) (Nielsen et al., 2004) was used to assess pretherapy emotional functioning. Participants responded to the item using a 5-point rating scale (1 = Very good [life was much the way I liked it to be] to 5 = Very poor [I barely managed to deal with things]). Responses were reverse-scored so that higher scores indicated better pretherapy emotional functioning. Nielsen and colleagues (2004) found that responses to this item (completed, on average, 55 weeks after intake) yielded a strong positive correlation (r = .55) with university counseling center clients’ actual intake scores on a measure of psychological distress.

Psychological Distress.

The 10-item Kessler Psychological Distress Scale (K-10; Kessler et al., 2002) was administered to measure general psychological distress. Participants indicated how frequently in the last 30 days they felt a manifestation of distress (e.g., “Nervous,” “Hopeless”) using a 5-point rating scale (1 = None of the time to 5 = Almost all of the time). Item responses were averaged into a total scale score, with higher scores reflecting greater psychological distress. In support of validity, K-10 scores were strongly positively correlated with a measure of perceived stress in a sample of transgender adults (Watson et al., 2019). In the same sample, Cronbach’s alpha was .90. Cronbach’s alpha was .89 in the present sample.

Procedures

Recruitment procedures and inclusion criteria for Study 2 were the same as those described for Study 1. At the conclusion of the survey in Study 2, participants were asked if they were interested in participating in a shorter survey consisting solely of the GIEMTS and demographic questionnaire two-to-three weeks in the future. A total of 1,422 individuals consented to participate in Study 2. Of these individuals, 27 (2%) were removed from the dataset because they did not provide data after providing consent or confirming the inclusion criteria. An additional 263 cases (18%) were removed because they did not meet inclusion criteria. Next, 534 cases (38%) were removed because they missed one or more of the validity check items. Seven cases (< 1%) were removed because of duplicate IP addresses. We also examined additional data fidelity indices (recaptcha, Fraud score) provided by Qualtrics because of concerns that our survey link was picked up by Internet bots.¹ These indices flagged an additional 155 (11%) cases that were removed from the dataset. One more case was removed for missing more than 20% of the data. These procedures resulted in a sample of 435 cases (31%).

Only 18 of these 435 cases (4%) had any missing data, with no case missing more than three items. Little’s MCAR test was significant, χ²(1,957) = 2,214.92, p < .001, which indicated that the data were not MCAR. We computed a continuous missingness variable that counted the total number of items missing and examined its correlations with the variables of interest. Psychological distress yielded a significant small negative correlation with missingness, r = −.13, p = .007; no other variable of interest was associated with missingness. The subsequent CFAs handled missingness using Mplus Version 8.7’s MLR. Tests of concurrent and incremental validity used available item analysis in SPSS.

Results

Confirmatory Factor Analyses

To confirm the factor structure of the 30-item GIEMTS, a series of CFAs using MLR in Mplus (Version 8.7) was conducted using data from the second sample (N = 435). Four competing measurement models (i.e., unidimensional, oblique four-factor model, second-order, bifactor) were examined. The unidimensional model assigns each item to a single factor. The oblique four-factor model assigned the items to be predicted by the four factors observed in Study 1 (Educational Burdening, Lack of Affirmation, Inflation, and Invalidation). The second-order model builds upon the oblique four-factor model by estimating a higher-order latent variable that models the shared variance among the four first-order latent variables. The bifactor model divides the variance of each item among (a) a general factor and (b) its respective specific subfactor. Model fit was assessed by the comparative fit index (CFI), the Tucker-Lewis index (TLI), the RMSEA, and the SRMR, with good model fit indicated by a CFI and TLI > .90 and an RMSEA and SRMR < .08 (Bentler & Bonett, 1980). Akaike’s information criterion (AIC), and Bayesian information criterion (BIC) were used to compare models that achieved adequate fit.

The unidimensional model yielded poor fit to the data, S-B χ² (405) = 2203.75, p < .001, CFI = .73, TLI = .71, RMSEA = .101, 90% CI (.097, .105), SRMR = 0.14—and thus was not considered further. The oblique four-factor model demonstrated adequate fit, S-B χ² (399) = 860.55, p < .001, CFI = .93; TLI = .92; RMSEA = .052, 90% CI (.047, .056), SRMR = 0.07. Standardized factor loadings for this model are presented in Table 4. Factor loadings for each item were significant (ps < 0.001). All latent factor intercorrelations were significant (ps < .001) and ranged from r = .28 (Educational Burdening with Lack of Affirmation) to r = .96 (Inflation with Invalidation).

Table 4.

Standardized Factor Loadings for the Four-Factor Oblique CFA Model in Study 2

Item	β
Factor 1: Educational Burdening
9. Needed me to educate them about the process of legally transitioning (e.g., changing my name and gender marker on my ID, changing the sex on my birth certificate, etc.).	.80***
8. Needed me to educate them about the process of getting gender-affirming medical procedures (such as taking hormones or hormone blockers, getting top or bottom surgery, etc.).	.82***
3. Needed me to educate them about TNB issues.	.80***
7. Needed me to educate them about the process of socially transitioning (such as using a new name or pronoun).	.89***
5. Placed the burden on me to educate them on the experiences of TNB people.	.87***
6. Demonstrated that they did not have good enough training in issues affecting TNB people.	.86***
4. Demonstrated that they did not have basic knowledge about TNB identities and experiences.	.88***
48. Did not know how to help me access gender-affirming medical treatment or services.	.70***
Factor 2: Lack of Affirmation
20. Encouraged me to explore my gender identity or expression.	.64***
24. Encouraged me to talk about my gender identity or expression if I wanted to.	.71***
51. Expressed pride or joy when I reached a milestone in my gender exploration.	.73***
50. Stated that exploring my gender identity or expression was normal.	.73***
45. Helped me access gender-affirming care or resources.	.70***
34. Affirmed my gender identity or expression during sessions.	.72***
47. Actively sought out more information about TNB identities in order to work more effectively with me.	.60***
37. Indicated that my gender identity or expression was healthy.	.75***
13. Discussed gender identity or expression at relevant times.	.59***
53. Voiced frustration that society makes things so difficult for TNB people like me.	.53***
Factor 3: Inflation
12. Brought up gender identity or expression even when it was unrelated to what I wanted to talk about.	.81***
41. Focused on gender dysphoria when it wasn’t relevant in session.	.83***
14. Overlooked other important aspects of my life beyond my gender identity or expression.	.81***
Factor 4: Invalidation
36. Implied my gender identity or expression was just a phase.	.87***
33. Implied my gender identity or expression was invalid.	.82***
35. Made me feel like I needed to change my gender identity or expression.	.84***
19. Pressured me to adopt a particular gender identity.	.81***
52. Told me that I was lesbian, gay, or bisexual rather than TNB.	.83***
40. Implied that my gender identity or expression needed to conform to society’s norms.	.84***
21. Suggested gender identity or expression and sex assigned at birth were the same thing.	.84***
27. Changed the subject whenever I brought up gender.	.81***
16. Called me by the wrong name (for example, my dead name, a different version of my name, etc.)	.74***

Note. N = 435. TNB = Transgender and/or nonbinary.

^***p < 0.001.

The second-order model also demonstrated adequate fit, S-B χ² (401) = 861.69, p < .001, CFI = .93; TLI = .93; RMSEA = .051, 90% CI (.047, .056), SRMR = 0.07. Loadings of first-order factors on the second-order factor were .87 for Educational Burdening, .29 for Lack of Affirmation, .97 for Inflation, and .99 for Invalidation (ps < .001). The factor loading for Lack of Affirmation appeared notably lower than the loadings for the other three factors.

Next, the bifactor model was examined. Factors were set as orthogonal (i.e., covariances were fixed to zero). The bifactor model demonstrated excellent fit, S-B χ² (375) = 677.03, p < .001, CFI = .95, TLI = .95, RMSEA = .043, 90% CI (.038, .048), SRMR = .04. Standardized factor loadings for the general factor and subfactors in the bifactor model are presented in Supplemental Materials C. For the general factor, all but two factor loadings were significant; significant factor loadings ranged from .12 to .88. For the Educational Burdening subfactor, all factor loadings were significant and ranged in size from .23 to .53. All factor loadings for items assigned to Lack of Affirmation were also significant and ranged in size from .52 to .72. None of the three factor loadings were significant for the Inflation subfactor. Lastly, six of the nine factor loadings for the Invalidation subfactor were significant, with values for significant loadings ranging from |.16 to .30.| However, two of these significant factor loadings were negative.

Because the oblique four-factor, second-order, and bifactor models all yielded acceptable fit, their AIC and BIC values were compared to determine which yielded relatively better fit. Lower AIC and BIC values suggest better fit, and AIC and BIC value differences exceeding 10 indicate substantive differences between models in fit (Kass & Raftery, 1995). The difference in AIC between the second-order and oblique four-factor model was modest (−3.14), but the difference in BIC (−11.29) provided evidence that the second-order model had superior fit. The AIC and BIC values for the bifactor model were clearly lower than those of the oblique four-factor model (−225.26, −127.45), suggesting superior fit. The AIC and BIC values for the bifactor model were also clearly lower than those of the second-order model (−222.12, −116.16). Thus, the bifactor model was considered the best model and was examined further.

Exploring Dimensionality

Coefficient omegas for the Educational Burdening, Lack of Affirmation, Inflation, Invalidation subfactors were .95, .89, .89, and .96, respectively. The coefficient omega for the general GIEMTS factor was .97. Coefficient omega (and its derivatives) can be used to evaluate the dimensionality of scales, which is important in deciding if it would be best to calculate and interpret total or subscale scores for the GIEMTS. The hierarchical omega coefficient (ωH) measures the proportion of raw total scale score variance that can be attributed to the general factor after accounting for the contributions of the subfactors. In turn, the subscale coefficient omega hierarchical (ωHS) measures the proportion of raw subscale score variance that is uniquely due to that specific subfactor after controlling for the general factor. According to Rodriguez et al. (2016), a ωH greater than .80 and a ωHS less than .50 indicate that a scale’s score largely reflects a single general factor (i.e., the unique variance explained by a specific subscale factor is negligible). In the current study, the ωH was .82 for the GIEMTS total score. This means that 85% (.82/.97 = .85) of the reliable variance in the GIEMTS total score is attributable to the general factor, whereas 15% is attributable to the subscale factors.

The ωHS for the Educational Burdening, Lack of Affirmation, Inflation, and Invalidation subfactors were .23, .84, .11, and .01, respectively. Thus, variance in scores on the Educational Burdening, Inflation, and Invalidation subscales was largely accounted for by the general GIEMTS factor. However, the high ωHS (> .50) for the Lack of Affirmation subscale suggests variance in its scores is largely independent of the general GIEMTS factor. In light of these results, Educational Burdening, Inflation, and Invalidation items were reassigned to a general factor labeled “General Anti-TNB Microaggressions” in the subsequent analyses, but Lack of Affirmation items were allowed to remain assigned to their own independent scale.

Concurrent and Incremental Validity

SPSS was used to estimate bivariate correlations among and descriptive statistics for variables used to evaluate the concurrent and incremental validity of the General Anti-TNB Microaggressions and Lack of Affirmation scales of the GIEMTS. Results are presented in Table 5. In support of concurrent validity, the General Anti-TNB Microaggressions and Lack of Affirmation scales yielded significant medium and small positive correlations with internalized transphobia and medium and large negative correlations with the working alliance. However, the correlations of the General Anti-TNB Microaggressions and Lack of Affirmation scales with expectations of rejection were both nonsignificant. The correlation between the General Anti-TNB Microaggressions and Lack of Affirmation scales was positive and significant–but, in support for the relative independence of the scales–the magnitude of the correlation was small.

Table 5.

Bivariate Correlations Among and Descriptive Statistics and Cronbach’s Alphas for Measures of Concurrent and Incremental Validity in Study 2

Variable	1	2	3	4	5	6	7	8
1. General Anti-TNB Microaggressions	--
2. Lack of Affirmation	.26***	--
6. Expectations of Rejection	.00	−.06	--
7. Internalized Transphobia	.32***	.21***	.45***	--
8. Working Alliance	−.48***	−.67***	−.03	−.28***	--
9. Perceived Improvement	−.10*	−.42***	−.14**	−.26***	.56***	--
10. Pretherapy Functioning	.55***	.18***	−.18***	.06	−.34***	.09	--
11. Psychological Distress	.05	.09	.31***	.30***	−.16***	−.27***	−.20***	--
Possible Range	1–7	1–7	1–5	1–5	1–7	1–5	1–5	1–5
M	3.71	2.82	3.03	2.77	5.01	3.84	3.10	2.51
SD	1.60	1.06	0.75	0.83	0.93	0.56	1.20	0.72
⍺	.97	.88	.86	.87	.89	.92	--	.89

Note. N = 435. TNB = Transgender and/or nonbinary.

^*p < .05.

^**p < .01.

^***p < 0.001.

We next sought to determine if the General Anti-TNB Microaggressions and Lack of Affirmation scales explained unique variance in the working alliance–even after controlling for the potential confounding roles of perceived improvement in psychotherapy, pre-therapy functioning, and current psychological distress. To this end, a multiple linear regression was performed in SPSS in which the General Anti-TNB Microaggressions and Lack of Affirmation scales of the GIEMTS, perceived improvement in psychotherapy, pre-therapy functioning, and current psychological distress were simultaneous predictors of the working alliance (Table 6). The model was significant, F(5, 429) =175.02, p < .001, and explained 67% of the variance in the working alliance. Regarding covariates, perceived improvement and pretherapy functioning yielded significant unique positive and unique negative associations, respectively, with the working alliance. Psychological distress’ unique association with the working alliance was nonsignificant. Both the General Anti-TNB Microaggressions and Lack of Affirmation yielded significant negative unique associations with the working alliance. General Anti-TNB Microaggressions accounted for 3% of the unique variance in the working alliance, whereas Lack of Affirmation accounted for 13%.

Table 6.

Tests of Incremental Validity of GIEMTS Scales in Predicting Working Alliance

	B	SE	95% CI	β	t	sr2	R 2
Constant	4.79	.28	(4.24, 5.35)		16.94***		.67
Perceived Improvement	.62	.05	(.51, .72)	.37	11.67***	.10
Pretherapy Functioning	−.15	.03	(−.20, −.09)	−.19	−5.35***	.02
Psychological Distress	−.07	.04	(−.14, .01)	−.05	−1.79	.00
General Anti-TNB Microaggressions	−.13	.02	(−.17, −.09)	−.23	−6.57***	.03
Lack of Affirmation	−.37	.03	(−.42, −.31)	−.42	−13.27***	.13

Note. N = 435. GIEMTS = Gender Identity and Expression Microaggressions in Therapy Scale. TNB = Transgender and/or nonbinary.

^*p < .05.

^**p < .01.

^***p < 0.001.

Study3: Test-Retest Reliability

In Study 3, we tested the short-term (two- to three-week) temporal stability of scores on the GIEMTS scales and retested internal consistency reliability.

Method

Participants

The final sample consisted of 151 participants TNB adults who had received psychotherapy within the last 12 months and who ranged in age from 18 to 61 years old (M = 29.22, SD = 7.01, Mdn = 28). One-hundred thirty-six (90.1%) participants were currently receiving psychotherapy, whereas 15 (9.9%) had received psychotherapy within the last 12 months. Additional demographic characteristics for Sample 3 are presented in Table 1.

Measures

Gender Identity and Expression Microaggressions in Therapy.

Participants were administered the 54-item GIEMTS twice, two-to-three weeks apart. Participants’ responses to GIEMTS items in Study 3 were linked to their responses in Study 2. Following the results of Study 2, the General Anti-TNB Microaggressions and Lack of Affirmation scales were calculated in the present study.

Procedures

As previously mentioned, participants who completed Study 2 were provided the opportunity to sign-up for a follow-up survey that would be administered two-to-three weeks later by providing an email address. Those participants who provided their email addresses were then automatically sent Study 3’s survey two weeks later, with a reminder sent a week thereafter. Individuals who completed this second, shorter survey could enter an additional raffle with a 1/25 chance of winning a $25 Amazon.com gift card.

A total of 406 individuals consented to be contacted for Study 3 and 174 responded to the follow-up survey. Data cleaning (e.g., removal of duplicate cases and cases flagged as likely fraudulent) resulted in a final analytic sample of 151 participants (86.8%). Of these 151 participants, only seven cases (5%) were missing any data. Little’s MCAR test was nonsignificant, χ²(762) = 760.85, p = .505, which supports the possibility of the data being MCAR. Available item analysis was used in subsequent analysis.

Results

Cronbach’s alphas for the General Anti-TNB Microaggressions and Lack of Affirmation scales of the GIEMTS at Time 2 were .99 and .93, respectively. The two- to three-week test-retest correlation for the General Anti-TNB Microaggressions scale was .90. The two- to three-week test-retest correlation for Lack of Affirmation was .66. Both test-retest correlations were significant (ps < .001) and large.

General Discussion

These studies detail the development and evaluation of the GIEMTS, a tool for TNB clients to report their experiences of anti-TNB microaggressions in psychotherapy. Study 1 results supported a four-factor model, reflecting the dimensions Educational Burdening, Lack of Affirmation, Inflation, and Invalidation. These factors are consistent with themes described in qualitative studies of TNB people’s experiences in psychotherapy (Anzani et al., 2019; Mizock & Lundquist, 2016; Morris et al., 2020). Study 2 data also supported this four-factor structure, but a bifactor analysis conducted on the same data showed that three of these factors (Educational Burdening, Inflation, and Invalidation) were more parsimoniously accounted for by a General Anti-TNB Microaggressions factor. In contrast, the Lack of Affirmation persisted as a distinct subfactor. Across studies, the GIEMTS demonstrated robust psychometric properties and thus may be a helpful tool to advance research, training, and clinical practice with TNB people.

Structural Validity and Reliability

The EFAs performed in Study 1 identified four facets of anti-TNB microaggressions: Educational Burdening, Inflation, Invalidation, and Lack of Affirmation. Educational Burdening comprised therapist behaviors that communicated a lack of knowledge about TNB-related concerns—a phenomenon documented in other social contexts, such as interactions with family members educational settings (Matsuno et al., 2022). In psychotherapy, Educational Burdening may pressure clients to step into the roles of educator and content expert, which “takes them out” of the client role (e.g., Mizock & Lundquist, 2016). Educational burdening may be especially salient in the current political climate of the U.S. in which anti-TNB legislation is being proposed and enacted at an alarming rate (HRC, 2023). We recommend clinicians stay apprised of anti-TNB stigma at local, state, and national levels to account for such stressors in their case conceptualizations of TNB clients.

A second manifestation of anti-TNB microaggressions that emerged in Study 1 was Inflation, or therapist behaviors that overly focused on a client’s TNB identity. Though acknowledging clients’ TNB identities is key to affirmative care (APA, 2015), clients may interpret therapists’ excessive focus as stemming from the pernicious stereotype that TNB identities themselves are pathological (Morris et al., 2020). Exclusively focusing on TNB identities may also preclude discussions that clients believe are more salient, thus impeding the development of therapeutic goal consensus (Tracey & Kokotovic, 1989). Determining when a topic or aspect of clients’ identities is relevant to psychotherapy requires rigorous clinical training, supervision, and open dialogue with TNB clients.

The third manifestation of anti-TNB microaggressions that emerged in Study 1 was Invalidation, which comprised microaggressions that denied the authenticity of a client’s TNB identity or expression. Invalidation is commonly reported in previous studies of TNB people’s experiences inside and outside of psychotherapy (e.g., Morris et al., 2020; Nadal, 2019). Invalidation in therapy may be experienced by clients as condescending or as conveying insufficient understanding of TNB issues. TNB clients often enter therapy with a history of invalidation from various sources, which may make therapists’ explicit, intentional validation of TNB identity even more crucial (Johnson et al., 2023).

The last dimension of the GIEMTS that emerged in Study 1 was Lack of Affirmation. These items were included in the GIEMTS item pool in light of qualitative research on TNB microaffirmations in psychotherapy, or therapist behaviors that affirm TNB clients (Anzani et al., 2019). We conceptualized the absence of microaffirmations as conveying the same anti-TNB bias that underlies anti-TNB microaggressions. An example Lack of Affirmation item is “Expressed pride or joy when I reached a milestone in my gender exploration” (reverse-scored). Recent studies suggest that achieving transition milestones such as initiating hormone replacement therapy or changing one’s name signify steps toward psychological wellness for some TNB people (e.g., Restar et al., 2019). Lack of Affirmation underscores the importance of actively celebrating TNB clients’ identity development milestones, which may both foster identity pride as well as model the support clients should receive outside therapy.

In support of the structural validity of these four GIEMTS dimensions, the four-factor structure that was identified in Study 1 was replicated in the independent sample of Study 2. Moreover, Cronbach’s alphas (Study 1; see Table 3) and coefficient omegas (Study 2) for Educational Burdening, Inflation, Invalidation, and Lack of Affirmation were high, which supports the internal consistency of these constructs. However, a bifactor model that estimated a general Anti-TNB Microaggressions factor in addition to the four subdimensions yielded relatively better fit than the four-factor model. Moreover, the general Anti-TNB Microaggressions factor explained more variance in Educational Burdening, Inflation, and Invalidation items than their respective subfactors. Although these dimensions were described as separate themes in qualitative studies (Mizock & Lundquist, 2016; Morris et al., 2020), it may be more parsimonious to utilize the General Anti-TNB Microaggressions scale rather than separate subscales. In contrast, the Lack of Affirmation subfactor accounted for most of the variance in its items. The relative independence of the Lack of Affirmation subscale from the General Anti-TNB scale is also supported by their significant but small positive correlation in Study 2 (Table 5). Though we conceptualized the absence of microaffirmations as reflecting microaggressions, it may be experienced by TNB clients as a more independent phenomenon.

In light of these results, we recommend that clinicians, educators, and researchers employing the GIEMTS utilize the General Anti-TNB Microaggressions and Lack of Affirmations scales. It may be tempting to administer the shorter subscales of the GIEMTS that reflect specific themes that emerged in the qualitative literature. Doing so may be permissible in some circumstances, but overall we recommend administering and calculating the General Anti-TNB Microaggressions scale to accumulate evidence for its psychometric properties and utility in understanding TNB clients’ experiences in psychotherapy.

Construct Validity

The GIEMTS also demonstrated content, convergent, discriminant, concurrent, and incremental validity. In Study 1, we sought to maximize content validity by grounding item generation in extant literature and soliciting feedback on the initial item pool from expert reviewers (Boateng et al., 2018). Study 1 also provided evidence for the convergent and discriminant validity of the GIEMTS. Regarding convergent validity, the GIEMTS subscales and total scale yielded moderate positive correlations with general anti-TNB discrimination. TNB clients who reported experiencing microaggressions in psychotherapy were more likely to report experiencing discrimination in other contexts. The GIEMTS also yielded large negative correlations with therapists’ general competency and cultural humility. Thus, therapists perceived as less competent or less culturally humble were more likely to perpetrate microaggressions in psychotherapy—or vice versa. These findings align with previous research (Owen et al., 2014; Shelton & Delgado-Romero, 2011). Future research may further explore the overlap and distinctness of these constructs by examining their unique contributions to therapy process and outcome variables.

In support of the discriminant validity of the GIEMTS, its correlations with two forms of socially desirable responding–self-deceptive enhancement and impression management–were nonsignificant. Given the current sociopolitical climate of the U.S.—in which the rights of TNB people are openly debated and assailed (HRC, 2023)—it is possible that TNB people are contextually motivated to inflate or exaggerate the extent or impact of perceived anti-TNB microaggressions to garner support for equitable treatment. Nevertheless, null associations between GIEMTS and socially desirable responding assure that participants’ responses likely reflect accurate perceptions of their therapists’ behavior.

Study 2 also provided partial support for the concurrent validity of the GIEMTS. Minority stress theory (Hatzenbuehler, 2009; Meyer, 2003) posits that exposure to others’ prejudice may lead individuals to adopt those same views, enhancing internalized oppression. Consistent with this perspective, GIEMTS scores were positively associated with internalized transphobia. Thus, encountering anti-TNB views from therapists may forestall the development of TNB identity pride and enhance prejudice towards one’s own TNB identity. Contrary to prior research (Testa et al., 2015; Watson et al., 2019), GIEMTS scores were not associated with expectations of rejection—another consequence of discrimination (Hatzenbuehler, 2009). Anti-TNB microaggressions in psychotherapy may be experienced so differently than the discrimination clients anticipate encountering in other contexts that their experiences of the former do not enhance the latter. Future research might directly assess how experiences of anti-TNB microaggressions in psychotherapy shape expectations of rejection in other contexts.

Additional evidence for the concurrent validity of the GIEMTS was garnered via significant negative associations of GIEMTS scores with the working alliance, a psychotherapy process variable with a robust relation with positive psychotherapy outcomes (Flückiger et al., 2018). Indeed, the General Anti-TNB Microaggressions and Lack of Affirmation subscales were both uniquely associated with poorer working alliance scores even after accounting for potential confounds. These findings converge with research on racist and sexist microaggressions in psychotherapy (e.g., Owen et al., 2014; Owen et al., 2010). The cross-sectional nature of the data in the current studies precludes causal interpretation of the findings. However, they raise the possibility that microaggressions in psychotherapy harm the working alliance, which, in turn, may hinder treatment success. Conversely, when the working alliance is strong, TNB may be less likely to interpret therapists’ ambiguous behaviors as indicating bias. Longitudinal research examining microaggressions in psychotherapy, the working alliance, and client functioning over time would better elucidate the directionality of associations.

Implications for Practice

Therapists should avoid engaging in anti-TNB microaggressions in psychotherapy, which may replicate the same hurtful interactions that precipitated TNB clients to seeking validating mental health services in the first place. Training programs may look to GIEMTS items as examples of therapeutic behaviors to avoid, or—in the case of the Lack of Affirmation subscale items—of behaviors to engage in more frequently. Intervention science may consider using pre-post designs to determine if receiving TNB-affirmative training leads to decreases in therapists’ engagement in anti-TNB microaggressions.

Because microaggressions are often unconscious and unintentional (Sue et al., 2007), we recommend that therapists monitor changes in session quality and dynamic (e.g., shifts in clients’ body language, response length and tone) because they may be subtle signs that anti-TNB microaggressions have occurred. Clinicians may also consider administering the GIEMTS to their own TNB clients. If anti-TNB microaggressions have occurred, therapists need not assume that they have irrevocably harmed the therapeutic relationship: In a study of racial or ethnic minority clients’ experiences of racial microaggressions, Owen and colleagues (2014) found that participants who discussed the microaggressions they experienced with their therapists reported similar working alliance levels as participants who had not experienced microaggressions. As is the case with therapeutic ruptures more generally (Safran et al., 2006), therapists may use the occurrence of microaggressions as opportunities to deepen the therapeutic bond by taking responsibility for their mistakes; validating clients’ experiences; praising clients’ courage for broaching a difficult subject; reinforcing clients’ self-advocacy in and outside of therapy; and committing to change their behavior in the future.

Items in the Lack of Affirmation factor highlight the importance of therapists advocating for TNB clients rather than gatekeeping. Given the stressors TNB people face in their everyday lives (James et al., 2016), it is critical that therapists actively support clients who are pursuing gender affirmation. This is consistent with informed consent models for TNB-affirmative practice with adults 18 years of age and older (Ducheny et al., 2017). Referring clients to other professionals providing needed services and educating clients about the logistics of accessing care are ways that therapists may both provide clients with instrumental support as well as expressing investment in clients’ holistic well-being.

Limitations and Future Directions

These findings must be considered in the context of several limitations that offer opportunities for future research. Participants were recruited from Internet communities focused on TNB identities, which may have led to an overrepresentation of individuals with stronger, more positive views toward their TNB identity. Second, the generalizability of results is limited by the demographic characteristics of each sample. Across studies, samples were predominantly white, transmasculine, genderqueer or nonbinary, young, and highly educated (see Table 1). Future research should enhance representation of TNB individuals who are racial or ethnic minoritized people, transfeminine, older, and who have fewer years of formal education.

In the context of prior qualitative research and the results of the current studies, we contend that GIEMTS scores largely reflect actual variation in the occurrence of anti-TNB microaggressions. Nonetheless, perceptions of microaggressions are likely also informed by factors such as awareness of oppression, current affective state, and the salience of an identity in a given relationship or interaction (Williams, 2020). Although GIEMTS responses were not associated with socially desirable responding, they may nonetheless be shaped by the limitations of retrospective recall or lack of knowledge of anti-TNB oppression. Research may build upon the findings of the current investigation—which focused exclusively on clients’ perspectives—by exploring therapists’ perceptions of their interactions with TNB clients. Furthermore, research may determine if clients’ and therapists’ reports of anti-TNB microaggressions converge with outsider observations of psychotherapy (e.g., Bettergarcia & Israel, 2018).

Given the scope of the current studies, we did not explore within-group differences in GIEMTS scores or test group invariance by gender, race or ethnicity, or sexual orientation. These are important goals for future research. It is also crucial to acknowledge that the GIEMTS focuses exclusively on microaggressions related to TNB identities. Future studies may adopt an intersectional lens by exploring the convergence of anti-TNB microaggressions with racism, sexism, heterosexism, and classism. Scholars have developed scales to capture intersectional microaggressions among groups with multiple marginalized identities (e.g., Huyhn et al., 2022); this work can be extended to microaggressions experienced in psychotherapy.

Future research may garner further support for the construct validity of the GIEMTS. Because knowledge of systems of oppression may be needed to identify when one has personally experienced bias, another test of convergent validity would be to explore the association of GIEMTS scores with facets of critical consciousness. Furthermore, given the multidimensional nature of the GIEMTS, it may be possible that the General Anti-TNB Microaggressions and Lack of Affirmation subscales may be differentially related to important aspects of identity and psychotherapy process and outcome. For example, because Lack of Affirmation taps into the absence of positive behaviors–rather than the presence of negative behaviors–perhaps the association of Lack of Affirmation with positive psychological constructs, like identity pride (Testa et al., 2015), would be stronger than identity pride’s association with General Anti-TNB Microaggressions. Regarding concurrent validity, studies may examine the associations of GIEMTS scores with other psychotherapy process variables, such as client attendance, session duration/frequency, and the resolution of transference or ruptures.

All three samples in this investigation encountered the 54-item version of the GIEMTS rather than the final 30-item version. It is important for future research to administer the 30-item version to enhance confidence that the findings of the three studies were not shaped by methodological artifacts such as scale length or the inclusion of superfluous items. Another methodological decision to consider is the recoding of “N/A” item responses as missing data. Though missingness in the datasets appeared to be largely at random, it is possible for this procedure to systematically inflate missingness. Thus, future research may explore alternative response formats to determine if they meaningfully impact GIEMTS scores.

Despite the study’s limitations, the GIEMTS demonstrates robust psychometric properties. The GIEMTS has the potential to be used in training to enhance trainees’ multicultural competence to work with TNB clients. In addition, therapists may use the GIEMTS with their clients to ensure that they are establishing a safe, healing therapeutic environment. We encourage future research to further test and refine the GIEMTS to maximize its applicability to diverse TNB populations.

Clinical Impact Statement:

• Question: How do microaggressions against transgender and nonbinary (TNB) clients manifest in therapy, and is there evidence for their association with important psychotherapy process variables, such as clients’ perception of the working alliance?

• Findings: The Gender Identity and Identity and Expression Microaggressions in Therapy Scale (GIEMTS) yielded two factors—General Anti-TNB Microaggressions and Lack of Affirmation—that showed evidence of construct validity, including negative associations with clients’ perceptions of the working alliance

• Meaning: The GIEMTS is a psychometrically sound self-report instrument that may inform TNB-affirmative training and may be used by therapists to assess and reduce the potential harm done to TNB clients.

• Next Steps: Future longitudinal clinical research may explore if changes to GIEMTS scores over time are associated with changes in the working alliance and client functioning.