Abstract
Transgender and/or nonbinary (TNB) people often highlight romantic partners as a key source of social support related to their TNB identity. Accordingly, there is a need for validated measures to quantify TNB identity support in romantic relationships. We used a multi-phase process to develop and evaluate a multidimensional measure of TNB identity support from romantic partners. First, we generated items based on prior qualitative research, then used cognitive interviews with six TNB young adults (ages 18–30 years) to refine the items. Second, we piloted 59 items in a survey of 321 TNB people (16 years and older) reporting on current or recent romantic relationships. Using those data, we performed exploratory factor analysis to identify subscales, then selected 19 items based on factor loadings, item-subscale correlations, α-if-item-deleted, discrimination parameters, and item information curves. Third, we used weekly diary survey data from 299 romantically partnered TNB young adults (ages 18–30 years) to perform confirmatory factor analysis and assess reliability at the within- and between-relationship levels. Results indicate that the 19-item Transgender/Nonbinary Identity Support from Partners Scale (TISPS) is a reliable and valid measure of overall TNB identity support and five subtypes of support: Facilitation [of identity exploration], Identity Validation, Celebration, Allyship, and Understanding. The TISPS demonstrated strong psychometric properties at the between-relationships level, including a consistent factor structure, good full-scale and subscale reliability, and measurement invariance across key characteristics. The TISPS can be used in clinical and community settings and future research on the role of social support in TNB mental health.
Keywords: transgender, nonbinary, couples, romantic relationships, social support
For transgender and/or nonbinary (TNB) people, romantic partners can be crucial sources of support on issues related to TNB identity, such as exploring one’s gender identity and coping with cisgenderism (Galupo et al., 2019; Gorman et al., 2022; Murchison et al., 2022). However, quantitative research on this construct, which we term TNB identity support, has been hampered by a lack of validated measures. Consequently, key questions remain unaddressed, including how TNB identity support varies across subgroups of TNB people and whether it encompasses quantitatively distinguishable subtypes of support.
Positive Romantic Relationship Experiences and TNB Mental Health
There is growing evidence that positive romantic relationship experiences—broadly defined—shape TNB people’s mental health. In one study with transgender women partnered with cisgender men, the association between interpersonal stigma and greater psychological distress was attenuated in women reporting greater commitment to their partners and relationships (Gamarel et al., 2019). Similarly, two studies found that support from “significant others” was associated with fewer depression/anxiety symptoms in transgender adults (Fuller & Riggs, 2021; Trujillo et al., 2017), and one found that the association between discrimination and suicidal ideation was attenuated at higher levels of “significant other” support (Trujillo et al., 2017). However, these studies considered general relational experiences, not experiences directly related to TNB identity.
Alongside the quantitative literature on general relationship experiences among TNB people, there is a growing qualitative literature on positive romantic relationship processes directly related to TNB identity. TNB people and their partners have described how collaboratively navigating challenges related to being TNB—particularly others’ prejudice and assumptions—improved their communication, intimacy, security, and commitment (Coppola et al., 2021; Motter & Softas-Nall, 2021; Platt & Bolland, 2018; Siboni, Rucco, et al., 2022). Partners of TNB people have also described personal growth related to their partner’s TNB identity, including a richer understanding of their own identities (Platt & Bolland, 2018; Siboni, Prunas, et al., 2022). In turn, TNB people have described numerous ways in which partners provide TNB identity support, such as supporting emotional processing related to gender, using gender-affirming language, correcting misgendering, and respecting gender-related social and physical boundaries (Coppola et al., 2021; Galupo et al., 2019; Motter & Softas-Nall, 2021; Pulice-Farrow et al., 2019).
In our prior qualitative work on TNB people’s romantic relationship experiences, which focused on young adulthood, we posited that TNB identity support can fulfill six identity-related psychological needs (“identity needs”) often salient for TNB people: (1) Supported Exploration, impartial encouragement and dialogue while exploring gender possibilities; (2) Identity Validation, interpersonal feedback confirming one’s gender as legitimate; (3) Positive Regard, feeling loved and appreciated regarding one’s gender and TNB identity; (4) Stigma Buffering, support in dealing with cisgenderism; (5) Shared Understanding, feeling understood in one’s experiences as a TNB person; and (6) Trigger Management, mitigating gender-related emotional triggers (Murchison et al., 2022). The concept of identity needs could explain why TNB identity support from romantic partners is particularly beneficial in the context of interpersonal and structural cisgenderism, which can directly undermine a TNB person’s identity needs while limiting opportunities for fulfilling those needs (Murchison et al., 2022). Moreover, most examples of TNB identity support in the literature correspond to at least one identity need (e.g., gender-affirming language and Identity Validation), suggesting that each identity need may have a parallel subtype of TNB identity support.
Measures of TNB-Specific Relationship Dynamics
At least three scales address TNB-specific dynamics in romantic or sexual relationships, with two focusing on harmful behaviors. The Transgender-Specific Intimate Partner Violence (T-IPV) Scale (Peitzmeier et al., 2019; Peitzmeier et al., 2021) assesses controlling behaviors and psychological abuse that exploits TNB identity and/or cisgenderism, such as threatening to disclose a partner’s TNB identity. The Gender Non-Affirmation from Cisgender Male Partners Measure assesses gender-invalidating behaviors in transgender men’s sexual relationships with cisgender men (Reisner et al., 2020). Higher scores on these scales have been linked to depression/anxiety, posttraumatic stress, and substance use disorders (Peitzmeier et al., 2019; Peitzmeier et al., 2021; Reisner et al., 2020), providing evidence that TNB-specific relationship dynamics are consequential for health. However, these measures address only harmful behaviors, not supportive ones. A third, recently-published measure, the Partner Affirmation of Transgender Identity Scale (PATIS; Scott et al., 2024), encompasses both supportive and harmful behaviors related to TNB identity. The PATIS represents a major advance in assessing TNB identity support, particularly because it includes a companion scale on which partners self-assess their behaviors. However, the PATIS does not assess specific subtypes of TNB identity support.
Multilevel Measure Validation in Relationships Research
In research on interpersonal relationships, there is considerable interest in multilevel analyses that distinguish within-participant associations (e.g., do participants have fewer health symptoms during weeks with better relationship functioning?) from between-participant associations (e.g., do participants with better relationship functioning, overall, have fewer symptoms?; Stadler et al., 2012). However, psychometric analyses are typically carried out at only the between-participant level (Trull & Ebner-Priemer, 2020). This is problematic because a scale’s reliability and factor structure can differ markedly for the between- and within-participant levels (Limberger et al., 2023; Trull & Ebner-Priemer, 2020; Wilhelm & Schoebi, 2007). To promote rigorous multilevel research on TNB people’s romantic relationships, there is a need to establish measures’ psychometric properties both between and within participants/relationships.
Present Study
We set out to develop and validate a quantitative self-report measure of perceived TNB identity support from a romantic partner: the Transgender/Nonbinary Identity Support from Partners Scale (TISPS). We anticipated that exploratory factor analysis of the TISPS would identify subtypes of TNB identity support aligned with the six identity needs discussed above that have emerged from qualitative research. If so, we planned to define subscales for each subtype.
We developed the TISPS in three phases, following steps recommended by DeVellis (2017). In Phase 1, we generated items based on our previous qualitative research (Murchison et al., 2022) and refined them through expert feedback and cognitive interviews with TNB young adults. In Phase 2, we piloted candidate items with a development sample of TNB adults and older adolescents, using these data to select final items. We then confirmed construct validity based on correlations with other constructs, hypothesizing that the TISPS would be highly positively correlated with other measures of romantic relationship functioning, would be moderately correlated with general (i.e., not relationship-specific) measures of psychological needs fulfillment (positive) and gender non-affirmation (negative), and would have no more than small correlations with measures of perceived social support in other social relationships. Finally, we conducted initial analyses of reliability and factor structure. In Phase 3, we used weekly survey data from TNB young adults to confirm the TISPS’ factor structure, assess reliability, and test for measurement invariance.
To inform future research with the TISPS, we estimated mean scores and group differences by participant age, relationship duration, and partner gender. We also tested hypotheses about group differences based on our prior qualitative work (Murchison et al., 2022). Specifically, because multiple marginalization may constrain options and strategies for achieving holistically fulfilling romantic relationships, we hypothesized that being nonbinary, a trans woman/transfeminine person, a person of color, and/or less financially secure would be associated with lower mean TISPS scores. Furthermore, since prior romantic experience as an openly TNB person may contribute to more fulfilling relationships, we hypothesized that relationships that began after participants realized they were TNB would have higher mean TISPS scores than those formed beforehand. Finally, because TNB-identified partners may draw on personal experience in providing TNB identity support, we hypothesized that TNB/TNB relationships (often termed t4t relationships) would have higher mean TISPS scores than TNB/cisgender relationships.
Data for Phases 1 and 3 were collected as part of broader studies addressing TNB people’s romantic relationships during young adulthood. Given the eligibility criteria for those parent studies, all Phase 1 and 3 participants were young adults, 18 to 30 years of age. However, in Phase 2, we validated the TISPS with a sample including older adolescents (≥ 16 years) and adults of any age, in order to begin the process of ensuring broader applicability of the TISPS across age groups.
Phase 1: Item Pool Development
Method
Item Generation
The initial item pool for the TISPS was based on our prior qualitative research on TNB young adults’ romantic relationships, which identified six domains in which romantic partners fulfilled or undermined psychological needs related to TNB identity (Murchison et al., 2022). The first author reviewed preexisting interview excerpts that had been coded to each domain, using language and examples from those excerpts to draft seven to eight preliminary items per domain. Items were drafted to be broadly relevant. For instance, our prior qualitative research highlighted sexual interactions as a key context for TNB identity support (Murchison et al., 2022), but not all participants were sexually active with their partners. Accordingly, we incorporated sexuality-related dynamics into more general items (e.g., “My partner would be glad if I told them what language I prefer for my body or my gender identity”).
The first author presented the items to an interdisciplinary, multi-institution works-in-progress group of faculty and trainees working in LGBTQ health, who provided focus group-style feedback on clarity, relevance, and additional items. Items were revised accordingly, and the process was repeated with a second group of faculty and trainees from an LGBTQ mental health research lab at another institution. We refined the resulting items through cognitive interviews with a new sample of TNB young adults, as described below.
Participants
Participants were recruited via LGBTQ social media groups (e.g., on Reddit and Facebook). Young adults were eligible if they identified as transgender and/or nonbinary, were 18 to 30 years of age, lived in the United States, could communicate in English, and had a past-year romantic relationship of at least six weeks. The six-week cutoff was intended to exclude very “casual” relationships for which it might be difficult to provide meaningful TISPS responses. Prior research with young adults has defined casual relationships with cutoffs of two to four months (Meier & Allen, 2009); we chose a less restrictive criterion with the goal of applying the TISPS to the upper end of the “casual” range.
Based on an online screening form, we selected interviewees, N = 6, to maximize diversity in gender and ethnic-racial identity. We concluded data collection upon reaching saturation, i.e., when additional interviews generated little novel feedback. Table 1 presents sample characteristics (see Sample 1).
Table 1.
Characteristics of transgender/nonbinary people who participated in development and validation studies for the Transgender/Nonbinary Identity Support from Partners Scale.
| Participant characteristics | Sample 1 (n = 6) | Sample 2 (n = 321) | Sample 3 (n = 293) |
|---|---|---|---|
| Age in years, mean (SD) | 23.00 (4.05) | 27.37 (6.27) | 24.80 (3.10) |
| Age group, n (%) | |||
| Under 18 years | – | 3 (0.9) | – |
| 18 to 24 years | 4 (66.7) | 107 (33.3) | 137 (46.8) |
| 25 to 30 years | 2 (33.3) | 158 (49.2) | 156 (53.2) |
| Over 30 years | – | 53 (16.5) | – |
| Self-described gender modality, n (%) | |||
| Transgender and nonbinary | 0 (0.0) | 203 (63.2) | 183 (62.5) |
| Transgender (but not nonbinary) | 5 (83.3) | 69 (21.5) | 59 (20.1) |
| Nonbinary (but not transgender) | 1 (16.7) | 35 (10.9) | 46 (15.7) |
| Another non-cisgender responsea | 0 (0.0) | 14 (4.4) | 5 (1.7) |
| Gender identity, n (%) b | |||
| Agender | 0 (0.0) | 27 (8.3) | 19 (6.3) |
| Demigirl or demiboy | 0 (0.0) | 4 (1.2) | 6 (2.0) |
| Gender fluid | 0 (0.0) | 29 (9.0) | 25 (8.4) |
| Genderqueer | 0 (0.0) | 35 (10.8) | 45 (15.1) |
| Gender nonconforming | 0 (0.0) | 7 (2.1) | 3 (1.0) |
| Man (incl. “trans man”) | 3 (50.0) | 57 (17.6) | 29 (9.7) |
| Nonbinary | 1 (16.7) | 167 (51.5) | 156 (52.2) |
| Transfeminine | 0 (0.0) | 19 (5.9) | 34 (11.4) |
| Transmasculine | 0 (0.0) | 91 (28.1) | 74 (24.7) |
| Woman (incl. “trans woman”) | 2 (33.3) | 29 (9.0) | 43 (14.4) |
| Additional responses | 0 (0.0) | 17 (5.2) | 21 (7.0) |
| Romantic orientation, n (%) b | |||
| Aromantic | 0 (0.0) | 9 (2.8) | 11 (3.8) |
| Biromantic | 1 (16.7) | 87 (27.1) | 86 (29.4) |
| Gay | 0 (0.0) | 36 (11.2) | 36 (12.3) |
| Lesbian | 0 (0.0) | 37 (11.5) | 34 (11.6) |
| Panromantic | 1 (16.7) | 42 (13.1) | 32 (10.9) |
| Queer | 0 (0.0) | 103 (32.1) | 89 (30.4) |
| Straight/heteroromantic | 0 (0.0) | 7 (2.2) | 6 (2.1) |
| Unlabeled | 2 (33.3) | 42 (13.1) | 30 (10.2) |
| Additional responses | 2 (33.3) | 21 (6.5) | 35 (11.9) |
| Country, n (%) | |||
| United States | 6 (100.0) | 304 (94.7) | 274 (93.5) |
| Canada | 0 (0.0) | 10 (3.1) | 6 (2.0) |
| Another country | 0 (0.0) | 7 (2.2) | 13 (4.4) |
| Personal financial situation, n (%) | |||
| Don’t meet basic expenses | – | 16 (5.0) | 31 (10.6) |
| Just meet basic expenses | – | 66 (20.6) | 82 (28.0) |
| Meet needs with a little left | – | 120 (37.4) | 107 (36.5) |
| Live comfortably | – | 119 (37.1) | 73 (24.9) |
| Ethnic-racial identity, n (%) b | |||
| Black African | 0 (0.0) | 1 (0.3) | 6 (2.0) |
| Black American | 1 (16.7) | 5 (1.6) | 31 (10.4) |
| Black Caribbean | 0 (0.0) | 4 (1.3) | 11 (3.7) |
| Indo-Caribbean | 0 (0.0) | 0 (0.0) | 1 (0.3) |
| East Asian | 1 (16.7) | 15 (4.7) | 27 (9.3) |
| South Asian | 0 (0.0) | 7 (2.2) | 15 (5.0) |
| Southeast Asian | 1 (16.7) | 6 (1.9) | 14 (4.7) |
| Pacific Islander | 0 (0.0) | 5 (1.6) | 0 (0.0) |
| Middle Eastern | 0 (0.0) | 8 (2.5) | 16 (5.3) |
| Indigenous – North America | 0 (0.0) | 8 (2.5) | 15 (5.0) |
| Latin American | 0 (0.0) | 22 (6.9) | 56 (18.7) |
| White American or Canadian | 2 (33.3) | 262 (81.6) | 160 (53.5) |
| White European | 0 (0.0) | 44 (13.7) | 51 (17.1) |
| Another background | 1 (16.7) | 8 (2.5) | 10 (3.3) |
| Partner/relationship characteristics | Sample 1 (n = 6) | Sample 2 (n = 321) | Sample 3 (n = 352)c |
| Relationship duration in years, mean (SD) | 3.99 (3.58) | 3.64 (6.57) | 2.43 (2.46) |
| Participant’s awareness of their own gender modality when relationship began, n (%) | |||
| Not aware of being trans/nonbinary | 3 (50.0) | 55 (18.3) | 49 (15.0) |
| Questioning being trans/nonbinary | 0 (0.0) | 33 (11.0) | 38 (11.7) |
| Aware of being trans/nonbinary | 3 (50.0) | 195 (64.8) | 218 (66.9) |
| Another response | 0 (0.0) | 18 (6.0) | 16 (4.9) |
| Relationship status changes, n (%) | |||
| Partnered throughout study | 5 (83.3) | 261 (81.3) | 308 (87.5) |
| Broke up before enrollment | 1 (16.7) | 45 (14.0) | – |
| Broke up during study | – | – | 44 (12.5)d |
| Another response | 0 (0.0) | 15 (4.7) | – |
| Partner gender modality, n (%) | |||
| Transgender and/or nonbinary | 1 (16.7) | 162 (50.6) | 234 (64.8) |
| Cisgender | 5 (83.3) | 126 (39.4) | 102 (28.3) |
| Another response | 0 (0.0) | 32 (10.0) | 17 (4.7) |
| Partner gender identity, n (%) b | |||
| Agender | 0 (0.0) | 12 (4.3) | 12 (3.4) |
| Demigirl or demiboy | 0 (0.0) | 0 (0.0) | 5 (1.4) |
| Gender fluid | 0 (0.0) | 12 (4.3) | 12 (3.4) |
| Genderqueer | 0 (0.0) | 11 (4.0) | 9 (2.6) |
| Gender nonconforming | 0 (0.0) | 3 (1.1) | 5 (1.4) |
| Man (incl. “cis man” and “trans man”) | 2 (33.3) | 47 (17.0) | 69 (19.6) |
| Nonbinary | 0 (0.0) | 93 (33.6) | 142 (40.3) |
| Transfeminine | 0 (0.0) | 9 (3.3) | 19 (5.4) |
| Transmasculine | 0 (0.0) | 18 (6.5) | 34 (9.7) |
| Woman (incl. “cis woman” and “trans woman”) | 4 (66.7) | 100 (36.1) | 95 (27.0) |
| Additional responses | 0 (0.0) | 14 (5.1) | 25 (7.1) |
| Legal relationship recognition, n (%) | |||
| Currently or formerly married | – | 36 (12.0) | 28 (7.8) |
| Currently or formerly in a registered domestic partnership | – | 7 (2.3) | 12 (3.3) |
| No legal relationship recognition | – | 242 (80.4) | 303 (83.9) |
| Another response | – | 16 (5.3) | 10 (2.7) |
Note. SD=standard deviation.
Examples of write-in responses recoded to this category include “Transgender – not non-binary but certainly less binary” and “Always nonbinary, sometimes transgender.”
Participants could select multiple identities; accordingly, percentages may sum to more than 100%.
In Sample 3, participants with more than one current romantic partner could choose to report separately on multiple relationships.
Includes 20 relationships that subsequently restarted during the study.
Procedures
Interviews were conducted via Zoom in June 2022, by the first author, a White transmasculine person with a doctoral degree in public health. After verbal informed consent, participants completed cognitive interviews lasting 27–73 minutes (mean = 52.7 minutes). Interviews combined a think-aloud procedure with additional prompts (Willis, 2004). Specifically, participants were asked to complete sets of 7–8 conceptually related TISPS items on a shared screen while verbalizing “whatever you’re thinking as you fill out the questions.” Next, the interviewer used prompts to identify items that were salient (“Were any of these questions particularly important, resonant, or valuable?”) or problematic (e.g., “Were any of these questions confusing?”) and relevant experiences not captured by the questions. This process was repeated for seven item sets. Participants received a $30 USD gift card voucher. Procedures were approved by the Yale University Institutional Review Board.
Analysis
Interviews were audio-recorded and transcribed. The first author identified 109 excerpts of feedback, grouped excerpts based on the issue(s) addressed, summarized each issue, and proposed modifications for each issue. Final modifications were determined by consensus among three coauthors. Phase 1 interview guides and process data are available from the corresponding author; interview transcripts are not available to protect participant privacy.
Results
Modifications addressed 17 issues, including seven raised by multiple participants. We deleted one item (“My partner would respect my gender identity more if I were cisgender”), added 15 items, and revised 12 items, as denoted in Supplemental Table 1. We also made minor changes to the instructions (e.g., specifying a time frame) and added an additional response option.
Some issues informed multiple revisions. For example, three participants commented that “If I told my partner about discrimination I experienced as a result of being trans/nonbinary, they would question whether it was really discrimination” (reverse scored) could describe either supportive or unsupportive behavior. We therefore revised the item to convey a lack of emotional support (“My partner thinks I am too sensitive about transphobia”; reverse scored) and added an item addressing supportive conversations about cisgenderism.
We addressed issues of item clarity and relevance regardless of how many participants raised them. To avoid idiosyncratic additions, we added new items only when they were suggested by multiple participants or resonated with preexisting in-depth interview data (see above). For instance, one participant noted that existing items did not capture how a partner felt about others knowing the participant is TNB, and this issue was also salient in our preexisting interview data. Accordingly, we added two related items: “My partner feels most comfortable when other people don’t know I’m trans/nonbinary” (reverse scored) and “My partner encourages me to be open about being trans/nonbinary whenever it’s safe to do so.”
Phase 2: Item Selection and Initial Psychometric Evaluation
Method
Participants
Eligibility and recruitment were as described for Phase 1 (above), but without purposive sampling and with broader age eligibility (≥ 16 years). Participants were asked to respond for their current or most recent romantic relationship of at least six weeks (see Phase 1 for discussion of the six-week restriction). In this phase, if there were multiple such partners, participants were asked to respond for only one. See Table 1, Sample 2 for sample characteristics.
The final sample (N = 321) slightly exceeded our target of 300, which was based on DeVellis’ (2017) review of guidelines for exploratory factor analysis.
Measures
Transgender Identity Support from Partners Scale.
Participants completed 59 preliminary TISPS items in random order, using Likert-type response options (1 = Strongly Disagree, 6 = Strongly Agree). Instructions read “Now please think about each of the following statements and mark how much you agree or disagree. If these experiences have changed since the beginning of the relationship, please answer based on your most recent experiences (e.g., last few weeks).”
Convergent Validity Measures.
We measured relationship satisfaction with the four-item version of the Couple Satisfaction Index (sample: “I have a warm and comfortable relationship with my partner”; 0 = Not at all true, 5 = Completely true; α = .94), previously validated with romantically partnered, but predominantly young and unmarried, adults (Funk & Rogge, 2007). We measured basic psychological need fulfillment with the Basic Need Satisfaction Scale, specifically the nine-item Relationship subscale (sample: “When I am with my partner, I often feel inadequate or incompetent”; 1 = Not at all true, 7 = Very true; α = .90) and 21-item General subscale (sample: “People in my life care about me”; 1 = Not at all true, 7 = Very true; α = .90), both previously validated with U.S. university students (Johnston & Finney, 2010; La Guardia et al., 2000). We measured gender non-affirmation using the six-item Non-Affirmation subscale of the Gender Minority Stress & Resilience Measure (sample: “I have difficulty being perceived as my gender”; 1 = Strongly disagree, 5 = Strongly agree; α = .91), previously validated with U.S. and Canadian TNB adults (Testa et al., 2015).
Discriminant Validity Measures.
To demonstrate that the TISPS does not merely capture a general tendency to perceive others as supportive, we measured social support from other sources using the Friends and Family subscales of the Multidimensional Scale of Perceived Social Support (Zimet et al., 1988; sample: “My family really tries to help me”; 1 = Very strongly disagree, 7 = Very strongly agree; α = .93).
Gender Identity and Modality.
Gender identity was assessed with the question “How do you describe your gender (e.g., woman, transmasculine, agender, nonbinary, genderqueer)?” with a free response box; responses were categorized for descriptive purposes (see Table 1). For self-reported gender modality (e.g., transgender, cisgender; Ashley, 2021), participants could select “Cisgender (not transgender or nonbinary)” [ineligible], “Transgender (but not nonbinary),” “Nonbinary (but not transgender),” “Both transgender and nonbinary,” or “None of these options describe me (feel free to explain).” These options were informed by research on gender modality measurement by Beischel et al. (2022). For instance, while nonbinary is a subset of transgender by our team’s working definitions, the “Nonbinary (but not transgender)” option reflects that some nonbinary people do not personally identify as transgender (e.g., because they believe the term reinforces binary understandings of gender or connotes medical, social, and/or legal gender transition steps that are not relevant to them; Beischel et al., 2022).
Additional Characteristics.
Romantic orientation was assessed using a free response question, recoded into nine categories. Race and/or ethnicity were assessed using the question “Which of the following reflect your background? You can choose more than one,” with 14 options (see Table 1). We assessed personal financial situation with a previously validated single-item measure (Williams et al., 2017). Participants also reported country of residence and age in years. At the relationship level, we assessed relationship duration, partner’s gender modality and gender identity alignment, legal relationship recognition, and whether the participant knew they were TNB when the relationship began.
Procedures
Data were collected between September 2022 and June 2023. Participants indicated informed consent, completed an online survey (median duration = 27.4 minutes), and received a $12 USD gift card voucher. Procedures were approved by the Yale University Institutional Review Board.
To validate responses, we rated each response using a points system based on features of fraudulent responses in our prior studies (e.g., logical inconsistencies, identical data across participants). We flagged 484 of 805 responses based on a prespecified threshold and asked them to confirm eligibility by contacting the study team from phone number from their reported country. (Internet-based numbers were not accepted because they can be obtained from outside the associated country.) All flagged respondents failed to confirm eligibility and were excluded.
A subset of de-identified Phase 2 survey data will be available through the National Institute on Alcohol Abuse and Alcoholism Data Archive as of February 2026. Questionnaires and R syntax are available from the corresponding author.
Results
Missing Data
We excluded 30 participants who abandoned the survey prior to the TISPS, leaving an analytic sample of 321 responses. In this sample, missingness was < 1% for TISPS items and up to 10% for subsequent scales. The probability of completing all measures was associated with measured variables (e.g., relationship satisfaction), indicating that data were not missing completely at random. However, missingness appeared to result from survey drop-off and inadvertent item skipping, suggesting that missingness was independent of the missing values themselves. Under this “missing at random” assumption, we addressed missingness using multiple imputation with fully conditional specification and predictive mean matching (M = 50; Van Buuren & Groothuis-Oudshoorn, 2011). Imputation models for item selection analyses used all candidate items as predictors. The imputation model for validity analyses used TISPS subscale scores and convergent/discriminant validity measures as predictors; we used passive imputation to calculate subscale scores during imputation (Van Buuren & Groothuis-Oudshoorn, 2011).
Item Selection
We used a multi-stage process to select 19 items for the final TISPS. The goal of the first stage was to identify distinct forms of TNB identity support and construct corresponding subscales. We confirmed that data were suitable for factor analysis based on the Kaiser-Meyer-Olkin test (KMO = .97; Kaiser & Rice, 1974), then determined the number of factors (i.e., six) using parallel analysis (Horn, 1965). We performed exploratory factor analysis with six factors, using an oblimin rotation because we expected (and confirmed empirically) that various forms of TNB identity support were intercorrelated. Based on pattern matrix loadings (Supplemental Table 1), we eliminated 19 items that did not load onto any factor at λ ≥ .40 and one that cross-loaded onto multiple factors at λ ≥ .40. We dropped one factor (seven items) that reflected harmful behaviors rather than TNB identity support; this factor’s highest-loading item was “My partner misgenders me or questions my gender to hurt me or get their way” (reverse scored). We named the remaining five factors based on the associated items: Facilitation [of identity exploration], Identity Validation, Celebration, Allyship, and Understanding.
The goal of the second stage was to select a parsimonious set of items for each subscale. We noted considerable skew, with most responses in categories reflecting more support. Accordingly, we used multidimensional graded response models (item response theory models for Likert-type data; Depaoli et al., 2018) to identify items providing more information at higher levels of TNB identity support. We extracted models’ discrimination parameters, which reflect how much information an item provides overall (Depaoli et al., 2018; Supplemental Table 2), and plotted item information curves (Supplemental Figure 1), which reflect how much information an item provides at various levels of the construct. We also considered classical metrics (pattern matrix factor loadings, item-subscale correlations, and subscale α-if-item-deleted; see Supplemental Table 2) and item content. For Facilitation, we retained the three items with the highest discrimination (items 02, 06, and 08) and one that described a prominent concept from interview data (item 11; “When I share new thoughts about my gender, my partner is excited to hear them”). For Identity Validation, we retained the two items with the highest discrimination (items 06 and 08), one that provided the most information at higher levels of the construct (item 04), and one that described a prominent concept from interview data (item 09; “My partner would take my gender identity more seriously if I were more masculine or feminine”). All items were retained for Celebration (only three candidate items) and Allyship (only four candidate items). For Understanding, we eliminated one item that no longer met the λ ≥ .40 cutoff (item 01) and retained the remaining four. Table 2 presents factor loadings, means, and standard deviations for the 19 final items. Box 1 presents the final 19-item TISPS with participant instructions, response options, and scoring instructions.
Table 2.
Pattern matrix factor loadings, means, and standard deviations for 19 final items in the Transgender/Nonbinary Identity Support from Partners Scale, based on cross-sectional survey data from transgender/nonbinary participants reporting on a current or recent romantic relationship, N = 321.
| Factor loadings | Item mean | Item SD | |||||
|---|---|---|---|---|---|---|---|
| Facilitation | Identity Validation | Celebration | Allyship | Understanding | |||
| If I had questions about my gender identity, my partner would help me think through them | 0.67 | 0.06 | 0.04 | 0.05 | 0.18 | 5.30 | 1.18 |
| If I were questioning something about my gender identity, my partner would be glad to discuss it | 0.93 | 0.02 | 0.03 | −0.01 | −0.04 | 5.47 | 1.05 |
| My relationship would be a safe space to try out a new name or pronouns in the future | 0.78 | 0.01 | 0.01 | 0.04 | 0.02 | 5.56 | 1.02 |
| When I share new thoughts about my gender, my partner is excited to hear them | 0.44 | 0.14 | 0.21 | 0.03 | 0.17 | 5.24 | 1.18 |
| My partner sees my gender exactly as I want them to see it | 0.05 | 0.57 | 0.10 | 0.12 | 0.14 | 5.01 | 1.29 |
| My partner sees my gender as legitimate, no matter what other people think | 0.32 | 0.44 | 0.11 | 0.15 | −0.04 | 5.67 | 0.90 |
| Sometimes I question how much my partner believes in my gender identity (R) | 0.11 | 0.58 | 0.07 | 0.05 | 0.15 | 5.20 | 1.38 |
| My partner would take my gender identity more seriously if I were more masculine or feminine (R) | −0.02 | 0.63 | 0.12 | 0.05 | −0.02 | 5.27 | 1.32 |
| My partner is proud of who I am, including my trans/nonbinary identity | 0.10 | −0.05 | 0.78 | 0.09 | 0.02 | 5.56 | 1.02 |
| My partner gives me compliments about my gender expression | 0.05 | −0.04 | 0.74 | 0.02 | 0.04 | 5.30 | 1.22 |
| My partner loves how I express my gender identity | −0.06 | 0.13 | 0.89 | −0.05 | −0.01 | 5.39 | 1.10 |
| If someone referred to my gender incorrectly, my partner would do something about it | −0.10 | 0.11 | 0.03 | 0.80 | 0.03 | 4.95 | 1.30 |
| My partner would speak up if people made prejudiced comments about trans/nonbinary people, as long as it was safe to do so | 0.02 | −0.08 | 0.02 | 0.79 | 0.01 | 5.23 | 1.17 |
| If we had a conflict with my partner’s family about my gender identity, my partner would stand up for me | 0.12 | 0.11 | −0.07 | 0.64 | −0.01 | 5.23 | 1.17 |
| If I had a conflict with my family related to being trans/nonbinary, my partner would help me deal with it | 0.36 | −0.02 | 0.14 | 0.44 | −0.03 | 5.57 | 0.94 |
| When I talk about being trans/nonbinary with my partner, they usually get what I mean | 0.19 | 0.11 | 0.11 | 0.13 | 0.51 | 5.19 | 1.18 |
| My partner doesn’t really understand how prejudice against trans/nonbinary people affects me (R) | 0.05 | −0.07 | 0.24 | 0.13 | 0.57 | 5.06 | 1.40 |
| There are things about my gender identity I don’t discuss with my partner because they probably wouldn’t understand (R) | 0.15 | 0.24 | −0.05 | −0.03 | 0.50 | 4.14 | 1.74 |
| My partner doesn’t fully understand my experiences with being trans/nonbinary (R) | −0.04 | 0.02 | −0.03 | −0.02 | 0.89 | 4.00 | 1.77 |
Note. Based on exploratory factor analysis with oblimin rotation. (R) indicates reverse scored items. SD = standard deviation. Bold text denotes factor loadings ≥ 0.40.
Box 1.
Final version of the Transgender Identity Support from Partners Scale.
| Please think about each of the following statements and mark how much you agree or disagree. If these experiences have changed since the beginning of your relationship, please answer based on your most recent experiences (e.g., last few weeks). | ||||||
|---|---|---|---|---|---|---|
| Strongly Disagree (1) |
Somewhat Disagree (2) |
Slightly Disagree (3) |
Slightly Agree (4) |
Somewhat Agree (5) |
Strongly Agree (6) |
|
| 1. If I had questions about my gender identity, my partner would help me think through them. | O | O | O | O | O | O |
| 2. If I were questioning something about my gender identity, my partner would be glad to discuss it. | O | O | O | O | O | O |
| 3. My relationship would be a safe space to try out a new name or pronouns in the future. | O | O | O | O | O | O |
| 4. When I share new thoughts about my gender, my partner is excited to hear them. | O | O | O | O | O | O |
| 5. My partner sees my gender exactly as I want them to see it. | O | O | O | O | O | O |
| 6. My partner sees my gender as legitimate, no matter what other people think. | O | O | O | O | O | O |
| 7. Sometimes I question how much my partner believes in my gender identity. | O | O | O | O | O | O |
| 8. My partner would take my gender identity more seriously if I were more masculine or feminine. | O | O | O | O | O | O |
| 9. My partner is proud of who I am, including my trans/nonbinary identity. | O | O | O | O | O | O |
| 10. My partner gives me compliments about my gender expression. | O | O | O | O | O | O |
| 11. My partner loves how I express my gender identity. | O | O | O | O | O | O |
| 12. If someone referred to my gender incorrectly, my partner would do something about it. | O | O | O | O | O | O |
| 13. My partner would speak up if people made prejudiced comments about trans/nonbinary people, as long as it was safe to do so. | O | O | O | O | O | O |
| 14. If we had a conflict with my partner’s family about my gender identity, my partner would stand up for me. | O | O | O | O | O | O |
| 15. If I had a conflict with my family related to being trans/nonbinary, my partner would help me deal with it. | O | O | O | O | O | O |
| 16. When I talk about being trans/nonbinary with my partner, they usually get what I mean. | O | O | O | O | O | O |
| 17. My partner doesn’t really understand how prejudice against trans/nonbinary people affects me. | O | O | O | O | O | O |
| 18. There are things about my gender identity I don’t discuss with my partner because they probably wouldn’t understand. | O | O | O | O | O | O |
| 19. My partner doesn’t fully understand my experiences with being trans/nonbinary. | O | O | O | O | O | O |
| Scoring: Items 7, 8, 17, 18, and 19 are reverse scored. Subscales are scored as follows: Facilitation as mean of items 1–4, Identity Validation as mean of items 5–8, Celebration as mean of items 9–11, Allyship as mean of items 12–15, and Understanding as mean of items 16–19. The full-scale score is the mean of the subscale scores. For respondents with multiple partners, the scale should be administered separately for each. When feasible, we recommend incorporating the partner’s initial/nickname and pronouns (e.g., “C sees my gender exactly as I want her to see it”). | ||||||
Reliability and Validity Analyses
Table 3 presents descriptive statistics for the subscales (scored as the mean of the relevant items) and full scale (mean of subscales). Means were high for all subscales, from 4.60 points (Understanding) to 5.24 points (Celebration) in a possible range of 1 to 6 points. Based on guidelines from DeVellis (2017), interitem reliability was very good (i.e., > .80) for the full TISPS (α = .96) and all subscales (α = .84–.91).
Table 3.
Correlations among Transgender/Nonbinary Identity Support from Partners Scale (TISPS) subscales and theoretically related constructs, and descriptive statistics for TISPS subscales, based on cross-sectional survey data from transgender/nonbinary participants reporting on a current or recent romantic relationship, N = 321.
| TISPS full scale | Subscales | |||||
|---|---|---|---|---|---|---|
| Facilitation | Identity validation | Celebration | Allyship | Understanding | ||
| Scale and subscale descriptive statistics | ||||||
| Scale/subscale mean | 5.18 | 5.39 | 5.29 | 5.42 | 5.24 | 4.60 |
| Scale/subscale standard deviation | 1.10 | 0.99 | 1.04 | 1.01 | 0.96 | 1.28 |
| Cronbach’s α | .96 | .91 | .86 | .89 | .84 | .85 |
| Correlations among TISPS subscales | ||||||
| Facilitation | .82*** | — | ||||
| Identity Validation | .80*** | .63*** | — | |||
| Celebration | .76*** | .68*** | .62*** | — | ||
| Allyship | .69*** | .48*** | .51*** | .50*** | — | |
| Understanding | .89*** | .69*** | .67*** | .59*** | .45*** | — |
| Correlations with theoretically related constructs | ||||||
| Relationship satisfaction (romantic partner) | .49*** | .42*** | .42*** | .52*** | .39*** | .38*** |
| Basic psychological needs fulfillment (romantic partner) | .49*** | .40*** | .41*** | .50*** | .42*** | .37*** |
| Basic psychological needs fulfillment (general) | .17** | .14* | .21*** | .17** | .09 | .14* |
| Gender non-affirmation (general) | −.13* | −.03 | −.21*** | −.07 | −.06 | −.12** |
| Perceived social support (friends) | .20*** | .17** | .21*** | .15* | .11 | .21*** |
| Perceived social support (family) | .00 | .04 | .04 | −.02 | −.02 | .01 |
Note.
indicates P < .05;
indicates P < .01;
indicates P < .001.
To assess construct validity, we examined Spearman correlations among the TISPS full scale, TISPS subscales, and theoretically related constructs (see Measures). Based on empirically informed guidelines for psychological research (Gignac & Szodorai, 2016), we treated r = .10 as the minimum for a small correlation, r = .20 as the minimum for a moderate (“typical”) correlation, and r = .30 as the minimum for a large correlation. All TISPS subscales were significantly, positively intercorrelated (r = .45 to r = .69; all p < .001). Regarding convergent validity, as hypothesized, the full scale and all subscales had large positive correlations with two other relationship functioning constructs: relationship satisfaction and basic psychological needs fulfillment by the romantic partner (r = .37 to r = .52; all P < .001). The TISPS full scale score and several subscales also had significant positive correlations with general basic psychological needs fulfillment and negative correlations with gender non-affirmation, though these did not consistently reach the hypothesized moderate level. Regarding discriminant validity, the TISPS full scale and some subscales slightly exceeded the hypothesized small correlations with perceived social support from friends (all r ≤ .21). However, there were no significant correlations with perceived social support from family (all r ≤ .04; all P > .050), indicating that the TISPS is not strongly confounded by general tendencies to perceive interpersonal relationships as supportive.
Phase 3: Multilevel Factor Analysis and Reliability Assessment
Method
Participants
Participants were recruited via LGBTQ social media groups (e.g., on Reddit, Facebook, and Lex), TNB community events in Massachusetts and New York, and word of mouth. Young adults were eligible if they were 18 to 30 years old at baseline, identified as transgender and/or nonbinary, could communicate in English, and were currently in a romantic relationship of at least six weeks (see Phase 1 for discussion of this restriction). We used stratified sampling to ensure diversity in race, ethnicity, and gender. In this phase, participants with multiple concurrent partners (typically reflecting consensual monogamy) reported separately on each partner. Sample characteristics are presented in Table 1 (see Sample 3).
The final sample size (N = 299) exceeded our target of 250, which was based on published power simulations indicating this sample would provide > 80% power for the planned confirmatory factor analysis, assuming 3–4 indicators per factor and loadings of .65 (Wolf et al., 2013). (Wolf and colleagues did not simulate models with more than three factors, but sample size requirements plateaued after two factors.)
Measures
Transgender Identity Support from Partners Scale.
Participants completed the 19-item TISPS (see Phase 2) with two study-specific modifications. First, items incorporated the partner’s initials and pronouns (e.g., “C sees my gender exactly as I want her to see it”). Second, instructions were modified to read, “Thinking about your relationship with [partner initial] over the past week, please think about each of the following statements and mark how much you agree or disagree. Please focus on your feelings and interactions in the past week, even if those experiences weren’t typical for your relationship.”
Additional Measures.
Participant and partner/relationship characteristics were assessed as described for Phase 2. Other measures (not analyzed here) addressed relationship functioning, exposure to cisgenderism, mental health, and substance use.
Procedures
Data were collected between August 2023 and April 2024. Prospective participants completed an online screening form assessing eligibility and demographics. As described in Phase 2, we excluded 73 potentially fraudulent screening responses. (Two additional participants were excluded after enrollment due to apparent inauthentic responding, e.g., marked logical inconsistencies). Otherwise, based on stratified sampling criteria, participants were invited to schedule orientation calls, during which they completed verbal informed consent.
After enrollment, participants completed a baseline survey (not analyzed here). Subsequently, participants completed eight “weekly surveys,” which collected confirmatory validation data for the TISPS, along with data on relationship functioning, depression, anxiety, and alcohol use for a broader study of romantic relationships and mental health in TNB young adults. Weekly survey links were delivered at seven-day intervals, at 11:00 am in each participant’s local time, via email and/or text message. Participants were asked to complete surveys by 11:59 pm the following day. The median completion time of the entire survey was 8.5 minutes. Participants were compensated up to $105. Procedures were approved by the Yale University Institutional Review Board.
De-identified Phase 3 data will be available through the National Institute on Alcohol Abuse and Alcoholism Data Archive as of February 2026. Questionnaires and Mplus syntax are available from the corresponding author.
Results
Missing Data
Among 2,445 weekly surveys with TISPS data (from 293 participants reporting on 352 partners), missingness for TISPS items was < 0.9%. Little’s test indicated that data were not missing completely at random. However, missingness was similar across items, suggesting that items were skipped inadvertently and could be assumed to be missing at random, without regard to the unobserved values. We therefore addressed missingness via multiple imputation with fully conditional specification and predictive mean matching (M = 50; Van Buuren & Groothuis-Oudshoorn, 2011), using the 19 TISPS items as predictors. We imputed in a two-level structure (weeks nested within partners/relationships).
Confirmatory Factor Analysis
Approach.
We used confirmatory factor analysis to test the hypothesized factor structure of the TISPS. We hypothesized that each item would load onto one of five factors and, thus, that a five-factor model would fit better than a model with all items loading onto a single factor. We considered three potential structures for the five-factor model: correlated factors (factors are intercorrelated but do not reflect a shared latent construct), bifactor (factors are uncorrelated, but an additional latent construct influences all items), and higher-order (factors are influenced by a shared latent construct).
The dataset had a multilevel structure, with weekly responses (Level 1) nested within partners/relationships (Level 2). Accordingly, we used a multilevel confirmatory factor analysis approach, testing the same factor structures at Level 1 (within partners/relationships) and Level 2 (between partners/relationships). Because some participants were in consensually nonomongamous relationships with multiple concurrent partners, partners/relationships were further nested within participants (Level 3). However, since most participants had only one partner/relationship, we did not model a factor structure at Level 3.
We observed negative skew (towards higher levels of support) for all items. Accordingly, our primary analyses were based on the mean- and variance-adjusted weighted least squares (WLSMV) estimator, which provides the most accurate factor loading estimates for skewed Likert-type data (Rhemtulla et al., 2012). For analyses not available with WLSMV in current software (estimating McDonald’s ω and testing Level 2 metric and scalar invariance), we used the robust maximum likelihood (MLR) estimator, which is also appropriate for Likert-type data (Rhemtulla et al., 2012). We also used MLR models to confirm that Level 3 clustering, which could not be addressed with WLSMV, did not appreciably affect results.
Models were fit in Mplus (Muthén & Muthén, 2017). In the original model, two items had small, non-significant negative residual variances. Since model fit otherwise appeared good (see below), we fixed those residual variances to zero. We assessed model fit using the comparative fit index (CFI), Tucker-Lewis Index (TLI), root mean square error of approximation (RMSEA), and standardized root mean residual (SRMR). Level-specific CFI, TLI, and RMSEA were calculated via the partially saturated model method (Ryu & West, 2009). Cutoffs for adequate model fit were based on the Padgett and Morgan (2021) recommendations for multilevel confirmatory factor analysis with ordinal data: CFI and TLI > .970, RMSEA < .020, SRMR < .040 at Level 1, and SRMR < .060 at Level 2.
Level 1 (Within Relationships) Results.
The Level 1 analysis addressed how items covaried from week to week within a given partner/relationship. The Level 1 bifactor model did not converge, and fit indexes for all models (Table 4) indicated inadequate fit at Level 1—potentially because responses were highly consistent from week to week, with a full scale intraclass correlation (ICC) of .90. Accordingly, we do not present further analyses of the TISPS’ factor structure or psychometric properties at the within-relationships level.
Table 4.
Fit statistics for confirmatory factor analysis structural model for 19 final items in the Transgender/Nonbinary Identity Support from Partners Scale based on weekly survey data from 293 transgender/nonbinary young adults reporting on 352 current or recent romantic relationships.
| CFI | TLI | RMSEA | SRMR | |
|---|---|---|---|---|
| Level 1: Within relationships | ||||
| Single factor model | .751 | .719 | .060 | .117 |
| Correlated factors model | .920 | .903 | .036 | .091 |
| Bifactor model | a | a | a | a |
| Higher-order factor model | .788 | .508 | .048 | .139 |
| Level 2: Between relationships | ||||
| Single factor model | .656 | .613 | .033 | .074 |
| Correlated factors model | .977b | .973b | .013b | .036b |
| Bifactor model | .731 | .328 | .057 | .288 |
| Higher-order factor model (final) | .982b | .959 | .014b | .040b |
Model did not converge.
Meets Padgett & Morgan (2021) cutoffs for robust maximum likelihood confirmatory factor analysis models with ≥ 100 Level 2 units: > .970 for comparative fit index (CFI) and Tucker-Lewis index (TLI), < .020 for root mean square error of approximation (RMSEA), < .040 for standardized root mean residual (SRMR) at Level 1, and < .060 for SRMR at Level 2.
Level 2 (Between Relationships) Results.
The Level 2 analysis addressed how items covaried at the partner/relationship level. As hypothesized, fit indexes (see Table 4) indicated that the five-factor models fit better than a single-factor model. Fit for the bifactor model was inadequate, but the correlated factors model exceeded all fit cutoffs and the higher-order factor model exceeded all but the TLI cutoff. Given that the correlated factors and higher-order factor model had similar fit metrics, we selected the higher-order factor model because it allowed us to conceptualize the TISPS full-scale score as reflecting a latent construct of overall TNB identity support rather than simply a composite of specific types of support.
Table 5 displays standardized loadings for the higher-order factor model. All items loaded onto their specific factors with magnitude ≥ .79 and P< .001. The five specific factors loaded onto the general factor with magnitude ≥ .74 and P< .001. Figure 1 depicts the hypothesized Level 2 factor structure with standardized loadings.
Table 5.
Factor loadings, reliability metrics, and intraclass correlation coefficients for the Transgender/Nonbinary Identity Support from Partners Scale based on survey data from 293 transgender/nonbinary young adults reporting on 352 current or recent romantic relationships.
| Standardized loading | P | Reliability | ICC | |||
|---|---|---|---|---|---|---|
| ω | α | Partner/relationship level | Participant level | |||
| Facilitation | .98 | .96 | .83 | .49 | ||
| If I had questions about my gender identity, my partner would help me think through them | 1.00 | < .001 | ||||
| If I were questioning something about my gender identity, my partner would be glad to discuss it | 0.81 | < .001 | ||||
| My relationship would be a safe space to try out a new name or pronouns in the future | 0.96 | < .001 | ||||
| When I share new thoughts about my gender, my partner is excited to hear them | 0.99 | < .001 | ||||
| Identity Validation | .85 | .85 | .76 | .57 | ||
| My partner sees my gender exactly as I want them to see it | 0.93 | < .001 | ||||
| My partner sees my gender as legitimate, no matter what other people think | 1.00 | < .001 | ||||
| Sometimes I question how much my partner believes in my gender identity (R) | 0.81 | < .001 | ||||
| My partner would take my gender identity more seriously if I were more masculine or feminine (R) | 0.79 | < .001 | ||||
| Celebration | .97 | .94 | .83 | .36 | ||
| My partner is proud of who I am, including my trans/nonbinary identity | 1.00 | < .001 | ||||
| My partner gives me compliments about my gender expression | 0.87 | < .001 | ||||
| My partner loves how I express my gender identity | 0.98 | < .001 | ||||
| Allyship | .87 | .85 | .75 | .55 | ||
| If someone referred to my gender incorrectly, my partner would do something about it | 0.85 | < .001 | ||||
| My partner would speak up if people made prejudiced comments about trans/nonbinary people, as long as it was safe to do so | 0.82 | < .001 | ||||
| If we had a conflict with my partner’s family about my gender identity, my partner would stand up for me | 0.79 | < .001 | ||||
| If I had a conflict with my family related to being trans/nonbinary, my partner would help me deal with it | 0.91 | < .001 | ||||
| Understanding | .95 | .92 | .83 | .45 | ||
| When I talk about being trans/nonbinary with my partner, they usually get what I mean | 0.99 | < .001 | ||||
| My partner doesn’t really understand how prejudice against trans/nonbinary people affects me (R) | 0.84 | < .001 | ||||
| There are things about my gender identity I don’t discuss with my partner because they probably wouldn’t understand (R) | 0.92 | < .001 | ||||
| My partner doesn’t fully understand my experiences with being trans/nonbinary (R) | 0.90 | < .001 | ||||
| Overall TNB identity support (general factor) | .85 | .95 | .90 | .61 | ||
| Facilitation | 0.89 | < .001 | ||||
| Identity Validation | 0.90 | < .001 | ||||
| Celebration | 0.93 | < .001 | ||||
| Allyship | 0.74 | < .001 | ||||
| Understanding | 0.85 | < .001 | ||||
Note. (R) indicates reverse scored items. ICC = intraclass correlation coefficient.
Figure 1.
Factor structure for the Transgender Identity Support from Partners Scale with Level 2 (between relationships) factor loadings and correlations from a robust maximum likelihood confirmatory factor analysis model, based on weekly survey data from 293 transgender/nonbinary young adults reporting on 352 current or recent romantic relationships.
Interitem Reliability
We assessed interitem reliability at the partner/relationship level (see Table 5) by calculating Cronbach’s α and McDonald’s ω in R 4.3.2 (Flora, 2020). Based on guidelines from DeVellis (2017), reliability for all factors was very good (i.e., > .80), with ranges of .85–.97 for McDonald’s ω and .85–.96 for Cronbach’s α. For the general factor, ω = .85 estimates reliability for measuring overall TNB identity support via the full scale score (Flora, 2020).
Intraclass Correlation
We used R 4.3.2 to calculate ICCs for each subscale and the full scale (see Table 5). ICCs at Level 2 (partners/relationships) ranged from .75 to .83 across subscales, with ICC = .90 for the full scale, indicating that scores varied much more between partners/relationships than from week to week within partners/relationships. Because 45 consensually nonmonogamous participants reported on multiple concurrent partners, resulting in a mean of 1.2 partners per participant, we also calculated ICCs at Level 3 (participants). Level 3 ICCs were .36–.57 for subscales and .61 for the total score, suggesting that there was roughly as much variation in TISPS scores within participants (i.e., between weeks and partners/relationships) as between participants.
Measurement Invariance
To determine whether TISPS scores can be compared across demographic and relationship characteristics, we tested for configural, metric, and scalar invariance at Level 2 within a multigroup confirmatory factor analysis framework (Putnick & Bornstein, 2016). Because unequal group sizes make invariance tests less sensitive (Chen, 2007), we only assessed measurement invariance for variables where each group represented > 20% of the sample, and we aggregated participants into more balanced groups where possible.
For each variable, we established configural invariance (equivalent factor structure across groups) by assessing fit of the hypothesized factor structure for each subgroup separately (Putnick & Bornstein, 2016) using the same fit cutoffs as the primary confirmatory factor analysis (Padgett & Morgan, 2021; see above). We then assessed metric invariance (equivalent loadings across groups) by comparing two multigroup models, one without group equality constraints and one constraining factor loadings to be equal across groups (Putnick & Bornstein, 2016). We judged metric invariance to be present when differences in these models’ fit indexes did not exceed cutoffs established by Chen (2007): Δ CFI > −.010, Δ RMSEA < .015, and Δ SRMR < .030. Finally, we assessed scalar invariance (equivalent item intercepts across groups) by comparing the metric invariance model to a model constraining intercepts to be equal across groups, comparing differences in fit indexes to cutoffs of Δ CFI > −.010, Δ RMSEA < .015, and Δ SRMR < .010 (Chen, 2007).
Measurement invariance results (Table 6) indicated configural, metric, and scalar invariance for age (18–24 years; 25–30 years), Western racial position (White only; any other ethnic-racial identity), personal financial situation (don’t or just meet basic expenses; live comfortably or meet needs with a little left), and relationship duration (less than 18 months; 18 months or longer). For participant gender identity alignment (woman and/or transfeminine; man and/or transmasculine; unaligned nonbinary), one configural model SRMR exceeded the cutoff, but all other results supported invariance. For partner gender modality (cisgender; TNB) and partner gender identity alignment (woman and/or transfeminine; man and/or transmasculine; unaligned nonbinary), full metric invariance was not supported (Δ CFI = −.014 and −.011, respectively). We therefore pursued partial metric invariance by releasing equal loading constraints for items with the highest modification indexes until Δ CFI rose above −.010 (Putnick & Bornstein, 2016). Specifically, we released loadings for Item 6 (for both variables) and Item 13 (for partner gender modality). Based on these partial metric invariance models, we found support for scalar invariance. Because partial metric invariance is sufficient for comparing group means, so long as scalar invariance is established (Steinmetz, 2013), we concluded that TISPS scores can be compared across these variables.
Table 6.
Measurement invariance results for the Transgender/Nonbinary Identity Support from Partners Scale based on weekly survey data from 293 transgender/nonbinary young adults reporting on 352 current or recent romantic relationships.
| Fit statistics | Sample size | |||||
|---|---|---|---|---|---|---|
| CFI (Δ CFI) | TLI | RMSEA (Δ RMSEA) | SRMR (Δ SRMR) | Level 1 units (weeks) | Level 2 units (relationships) | |
| Age group | ||||||
| Configural fit, age less than 25 years | .998a | .995a | .004a | .049a | 1,390 | 206 |
| Configural fit, age 25 to 30 years | 1.000a | 1.013a | .000a | .041a | 1,055 | 146 |
| Test for metric invariance | (−.002)b | – | (−.002)b | (.000)b | – | – |
| Test for scalar invariance | (−.001)b | – | (−.002)b | (−.001)b | – | – |
| Gender identity alignment | ||||||
| Configural fit, man and/or transmasculine | 1.000a | 1.061a | .000a | .040a | 790 | 114 |
| Configural fit, woman and/or transfeminine | 1.000a | 1.116a | .000a | .068 | 694 | 103 |
| Configural fit, unaligned nonbinary | .993a | .984a | .007a | .050a | 961 | 135 |
| Test for metric invariance | (−.009)b | – | (−.001)b | (.001)b | – | – |
| Test for scalar invariance | (.000)b | – | (−.003)b | (−.003)b | – | – |
| Western racial position | ||||||
| Configural fit, White American, Canadian, or European only | 1.000a | 1.018a | .000a | .050a | 1,262 | 183 |
| Configural fit, any other ethnic-racial identity | 1.000a | .999a | .002a | .040a | 1,183 | 169 |
| Test for metric invariance | (−.002)b | – | (−.002)b | (.000)b | – | – |
| Test for scalar invariance | (.000)b | – | (−.002)b | (−.001)b | – | – |
| Personal financial situation | ||||||
| Configural fit, don’t or just meet basic expenses | 1.000a | 1.002a | 0.000a | .041a | 1515 | 213 |
| Configural fit, live comfortably or meet needs with a little left | 1.000a | 1.008a | 0.000a | .048a | 922 | 136 |
| Test for metric invariance | (−.002)b | – | (−.002)b | (.002)b | – | – |
| Test for scalar invariance | (.000)b | – | (−.002)b | (.000)b | – | – |
| Relationship duration | ||||||
| Configural fit, duration less than 18 months | .997a | .994a | .004a | .054a | 1,103 | 164 |
| Configural fit, duration 18 months or longer | 1.000a | 1.019a | .000a | .036a | 1,326 | 184 |
| Test for metric invariance | (−.005)b | – | (−.001)b | (.001)b | – | – |
| Test for scalar invariance | (−.001)b | – | (−.002)b | (−.001)b | – | – |
| Partner gender modality | ||||||
| Configural fit, cisgender partner | 1.000a | 1.006b | .000a | .054a | 780 | 107 |
| Configural fit, transgender and/or nonbinary partner | 1.000a | .999b | .001a | .044a | 1,657 | 242 |
| Test for full metric invariance | (−.014) | – | (.003)b | (.017)b | – | – |
| Test for partial metric invariancec | (−.008)b | – | (.002)b | (.007)b | – | – |
| Test for scalar invariancec | (−.007)b | – | (.001)b | (−.006)b | – | – |
| Partner gender identity alignment | ||||||
| Configural fit, man and/or transmasculine partner | 1.000a | 1.095a | .000a | .050a | 701 | 98 |
| Configural fit, woman and/or transfeminine partner | 1.000a | 1.070a | .000a | .050a | 751 | 108 |
| Configural fit, unaligned nonbinary partner | 1.000a | 1.015a | .000a | .049a | 985 | 143 |
| Test for metric invariance | (−.011) | – | (.001)b | (.032) | – | – |
| Test for partial metric invarianced | (−.005)b | – | (−.002)b | (−.002)b | – | – |
| Test for partial scalar invarianced | (.000)b | – | (−.003)b | (−.003)b | – | – |
Note. As WLSMV has not been implemented for multigroup two-level analyses in Mplus or R, we used MLR for the metric and scalar invariance models. Hirschfeld and von Brachel (2014) found that WLSMV and MLR produce comparable measurement invariance results with ordinal data.
Meets Padgett & Morgan (2021) cutoffs for robust maximum likelihood confirmatory factor analysis models with ≥ 100 Level 2 units: > .970 for comparative fit index (CFI) and Tucker-Lewis index (TLI), < .020 for root mean square error of approximation (RMSEA), and < .060 for SRMR.
Meets Chen (2007) cutoffs for tests of metric and scalar invariance: Δ CFI > −.010, Δ RMSEA < .015, and Δ SRMR < .030 (metric invariance) or Δ SRMR < .010 (scalar invariance).
Equal loading constraints were released for two items based on their modification indexes (Putnick & Bornstein, 2016).
Equal loading constraint was released for one item based on its modification index (Putnick & Bornstein, 2016).
Subgroup Means and Differences
We assessed how the TISPS full-scale score varied by participant and partner/relationship characteristics. Specifically, we used R 4.4.2 to fit unadjusted multilevel models for TISPS scores from the weekly surveys, using random intercepts to model clustering of weekly scores within partners/relationships and participants. We present group mean scores (estimated marginal means) and differences between groups (fixed effect coefficients) in Table 7.
Table 7.
Differences in Transgender/Nonbinary Identity Support from Partners Scale (TISPS) full-scale scores by individual and partner/relationship characteristics, based on weekly survey data from 293 transgender/nonbinary young adults reporting on 352 current or recent romantic relationships.
| Mean TISPS full-scale score (95% CI)a | Difference from reference group meana | P for difference from reference groupa | |
|---|---|---|---|
| Participant characteristics | |||
| Age | |||
| 18 to 21 years (reference) | 5.15 (4.96, 5.33) | – | – |
| 22 to 24 years | 5.23 (5.09, 5.37) | 0.084 | .485 |
| 25 to 27 years | 5.41 (5.26, 5.55) | 0.262 | .030 |
| 28 to 30 years | 5.35 (5.19, 5.50) | 0.201 | .106 |
| Gender modality | |||
| Transgender but not nonbinary (reference) | 5.22 (5.05, 5.39) | – | – |
| Nonbinary | 5.32 (5.23, 5.40) | 0.096 | .330 |
| Gender identity alignment b | |||
| Man and/or transmasculine | 5.28 (5.15, 5.41) | – | – |
| Woman and/or transfeminine | 5.42 (5.27, 5.57) | 0.138 | .175 |
| Unaligned nonbinary | 5.22 (5.10, 5.34) | −0.061 | .505 |
| Personal financial situation | |||
| Don’t meet basic expenses (reference) | 5.27 (5.11, 5.42) | – | – |
| Just meet basic expenses | 5.38 (5.25, 5.50) | 0.107 | .297 |
| Meet needs with a little left | 5.21 (5.06, 5.35) | −0.062 | .567 |
| Live comfortably | 5.32 (5.08, 5.56) | 0.051 | .728 |
| Western racial position | |||
| White American, Canadian, or European only (reference) | 5.34 (5.23, 5.45) | – | – |
| Any other ethnic-racial identity | 5.25 (5.14, 5.36) | −0.088 | .267 |
| Ethnic-racial identity c | |||
| Asian (incl. East Asian, South Asian, Southeast Asian, Indo-Caribbean) | 5.11 (4.93, 5.29) | −0.223 | .031 |
| Black (incl. Black African, Black American, Black Caribbean) | 5.28 (5.07, 5.50) | −0.016 | .895 |
| Indigenous – North America | 5.21 (4.87, 5.55) | −0.090 | .619 |
| Latin American | 5.39 (5.21, 5.57) | 0.114 | .259 |
| Middle Eastern | 5.45 (5.12, 5.79) | 0.168 | .335 |
| White (incl. White American, White Canadian, White European) | 5.34 (5.24, 5.43) | 0.130 | .124 |
| Partner/relationship characteristics | |||
| Relationship duration | |||
| Less than a year (reference) | 5.33 (5.21, 5.45) | – | – |
| Between one and two years | 5.39 (5.25, 5.53) | 0.062 | .493 |
| Between two and four years | 5.23 (5.08, 5.38) | −0.102 | .285 |
| More than four years | 5.19 (5.05, 5.34) | −0.135 | .136 |
| Participant’s awareness of their own gender modality when relationship began | |||
| Not aware of being trans/nonbinary (reference) | 5.07 (4.88, 5.25) | – | – |
| Questioning being trans/nonbinary | 5.19 (4.99, 5.38) | 0.124 | .351 |
| Aware of being trans/nonbinary | 5.38 (5.29, 5.47) | 0.318 | .002 |
| Partner gender identity alignment b | |||
| Man and/or transmasculine (reference) | 5.16 (5.03, 5.29) | – | – |
| Woman and/or transfeminine | 5.37 (5.25, 5.49) | 0.210 | .019 |
| Unaligned nonbinary | 5.33 (5.22, 5.44) | 0.168 | .044 |
| Partner gender modality | |||
| Cisgender (reference) | 4.95 (4.83, 5.07) | – | – |
| Transgender and/or nonbinary | 5.47 (5.38, 5.56) | 0.522 | < .001 |
Estimated marginal means, fixed effect coefficients, and fixed effect P values from bivariate mixed effects models.
Participants and partners were classified as unaligned nonbinary if their self-reported gender identity did not include woman, transfeminine, man, and/or transmasculine.
Because participants could be in multiple ethnic-racial identity groups, the reference group was all participants not in the specified group.
Mean TISPS scores were significantly higher in participants 25 to 27 years of age (mean = 5.41) than participants 18 to 21 years of age (mean = 5.15; P = .030) and significantly lower among Asian participants (mean = 5.11) than non-Asian participants (mean = 5.34; P = .031). There were no other significant differences by age or ethnic-racial identity, nor by participant gender modality, gender identity alignment, or financial situation. Turning to partner/relationship characteristics, scores were significantly higher in relationships that began after the participant realized they were TNB (mean = 5.38) than those that began beforehand (mean = 5.07; P = .002). There were also significant differences by partner gender identity alignment, with the highest mean TISPS scores for women and/or transfeminine partners (mean = 5.37) and the lowest for men and/or transmasculine partners (mean = 5.16; P = .019). Finally, the variable associated with the largest difference in TISPS scores was whether the partner was TNB (mean = 5.47) or cisgender (mean = 4.95; P < .001).
Discussion
This study presented the TISPS as a tool for measuring TNB identity support in romantic relationships: an interpersonal context that may be particularly salient for TNB people, who too often experience rejection in other relationships, including families of origin (James et al., 2016). We established the TISPS’ content and construct validity as a measure of overall TNB identity support (full-scale score) and five subtypes of TNB identity support: Facilitation [of identity exploration], Identity Validation, Celebration, Allyship, and Understanding (subscale scores). The TISPS also demonstrated a consistent factor structure, good interitem reliability, and measurement invariance across key participant and relationship characteristics. Notably, because TISPS responses had limited week-to-week variability, we were unable to establish its factor structure or reliability at the within-participant level. Nonetheless, the TISPS is an important new resource for understanding how interpersonal support shapes TNB people’s health and wellbeing.
Exploratory and confirmatory factor analysis of the TISPS suggested that perceived TNB identity support in romantic relationships consists of five distinct forms of support within a higher-order construct of overall TNB identity support. We conceptualized the specific forms of support as Facilitation (serving as a sounding board for gender exploration), Identity Validation (affirming one’s gender as legitimate), Celebration (providing positive feedback about one’s gender), Allyship (assisting in navigating cisgenderism), and Understanding (demonstrating comprehension of one’s experience as a TNB person). These dimensions roughly map onto “identity needs” from our formative qualitative research on TNB young adults’ romantic relationships (Murchison et al., 2022); for instance, Identity Validation support could fulfill the Positive Regard identity need. However, we did not identify a form of support corresponding to Trigger Management, which concerns how partners address identity-related emotional triggers. Trigger Management items primarily loaded onto the original sixth factor, which we dropped because it reflected negative rather than supportive behaviors.
While the TISPS demonstrated a consistent factor structure and good interitem reliability between partners/relationships, this factor structure fit poorly at the within-relationships level, potentially because there was little week-to-week variation in responses over the nine-week study period. It is not clear whether TNB identity support itself is highly stable or, alternatively, whether the TISPS is not well suited to capture short-term fluctuations. Regardless, based on these results, we do not recommend the TISPS for short-term within-participant analyses (e.g., over days or weeks). Future research should assess how well the TISPS captures longer-term changes within a relationship (e.g., over months or years).
This study provides some of the first quantitative evidence of variation in TNB identity support by individual and partner/relationship characteristics. The largest of these differences was between t4t (TNB/TNB) relationships and TNB/cisgender relationships, suggesting that a partner’s own experience as a TNB person can aid in providing TNB identity support. However, mean TISPS scores were also relatively high for TNB/cisgender relationships, indicating that many cisgender partners also provide considerable TNB identity support.
Constraints on Generality
Due to inclusion criteria and recruitment strategies, this research included few participants outside the age range of 18–30 years (only included in Phase 2); few participants outside the United States; and few heterosexual participants, who constitute around 20% of U.S. TNB adults (Carone et al., 2021). Further, despite considerable variability in personal financial situation, few participants were currently unable to meet basic expenses. As a result, the TISPS items might disproportionately reflect the experiences of TNB people who are young adults, U.S.-based, non-heterosexual, and not in acute poverty. If researchers use the TISPS with broader TNB populations, including adolescents and older adults, we recommend confirming its reliability and factor structure in these populations. In addition, White participants were notably overrepresented in Phase 2 relative to the U.S. population, meaning that the items selected for the final TISPS could disproportionately represent White TNB people’s experiences of TNB identity support. However, cognitive interviewing (Phase 1) and confirmatory analyses (Phase 3) were conducted in samples with good ethnic-racial diversity.
We also note constraints on generality by relationship characteristics. Participants generally reported high levels of TNB identity support, and the TISPS could have different psychometric properties in samples with lower levels of TNB identity support. Similarly, because most participants reported on current relationships, we were unable to assess whether the TISPS performs differently in past relationships than current ones.
Limitations
One limitation of the TISPS is that it only addresses support directly related to being TNB. Multiple forms of identity support (e.g., regarding ethnic-racial identity or disability) may be salient in a given relationship, perhaps more so than TNB identity support (Murchison et al., 2022). Moreover, TNB identity support itself is shaped by multiple social identities and systems of marginalization, as evident in our prior qualitative work (Murchison et al., 2022) and participant comments in the present research (e.g., “…the things that my gf doesnt understand …are directly related to how my gender is connected to my race…”). We encourage researchers using the TISPS to attend to the confluence of cisgenderism with other systems of marginalization, such as racism, ableism, and sexism (Collins & Bilge, 2016; Crenshaw, 1991).
This study did not assess criterion validity, i.e., comparing the TISPS to an external standard. Because data were collected before the PATIS (Scott et al., 2024) was published, there were no published measures of TNB identity support with which to compare the TISPS. Future studies should examine how the TISPS relates to potential criterion variables, including behavioral observations of couples’ interactions (Perry et al., 2021).
Implications
The TISPS can contribute to clinical and scientific efforts to help TNB people and their loved ones support one another more effectively. The scale may be useful to mental health clinicians, who can use it to prompt TNB clients’ reflection on relationship experiences and priorities. The TISPS can also serve as a training tool for clinicians, exemplifying partner behaviors that contribute to TNB identity support. On the research side, the TISPS is a valid and reliable measure of TNB identity support that can enhance future research on romantic relationships and social support in TNB populations.
One major contribution of this research is in identifying five quantitatively distinct subtypes of TNB identity support (Facilitation, Identity Validation, Celebration, Allyship, and Understanding), each with a corresponding TISPS subscale. These subscales differentiate the TISPS from the PATIS, which has two subscales reflecting positive and negative behaviors (Scott et al., 2024). Assessing subtypes of TNB identity support may be valuable in counseling and psychotherapy by highlighting relative strengths and weaknesses within a relationship. The TISPS can serve as a starting point for these conversations: Clients could reflect on their highest and lowest subscale scores, or simply read and reflect on items from each subscale, to identify specific forms of support to address with their partners. This strengths-and-weaknesses framing may reduce defensiveness and build self-efficacy, particularly when clients and/or partners have difficulty acknowledging unsupportive behavior.
The TISPS may likewise be useful for self-guided reflection. Feedback from research participants suggests that completing the TISPS independently can help TNB people identify strengths and areas for improvement in their current relationships—and, in some cases, set higher expectations for future partners. Similarly, partners of TNB people may find that reviewing the TISPS questions helps them identify ways to provide more support. We have shared a self-administered version of the TISPS online (https://sites.bu.edu/t4thealth/transgender-nonbinary-identity-support-from-partners-scale/) and in TNB community workshops, and we welcome others to do the same.
The TISPS may be a valuable process measure for relationship interventions with TNB people, allowing investigators to test subtypes of TNB identity support as mediators and/or baseline moderators of intervention effects. The TISPS can likewise be used study how TNB identity support relates to mental health, general relationship functioning, and structural determinants of health. Our descriptive analyses of TISPS scores suggest additional research avenues. In a predominantly U.S. sample, we found significantly lower TISPS scores in ethnically Asian (vs. non-Asian) participants, indicating a need for more research on Asian American TNB young adults’ romantic experiences. Furthermore, mean TISPS scores were lowest for partners who were men and/or transmasculine compared with partners of other genders, suggesting a need for further research with cisgender male, transgender male, and/or transmasculine partners of TNB people, who face distinctive challenges related to masculine gender norms and homophobia (Muñoz-Laboy et al., 2017).
Several qualitative studies suggest that TNB identity support benefits not only recipients but also supporting partners and relationships as a whole (Coppola et al., 2021; Motter & Softas-Nall, 2021; Siboni, Prunas, et al., 2022). With these benefits in mind, we developed the TISPS to facilitate clinical, community-based, and research efforts to help TNB people and their partners support one another more effectively. These efforts to improve interpersonal support should be coordinated with efforts to dismantle the systems of marginalization that threaten health and life opportunities for TNB people and their significant others.
Supplementary Material
Transparency and openness:
Phase 1 interview data are not publicly available due to the personal nature of the data. Phase 2 and Phase 3 survey data from participants who consented to data sharing will be available through the National Institute on Alcohol Abuse and Alcoholism Data Archive as of February 2026. Interview guides, revision process data (summaries of feedback and TISPS modifications), questionnaires, R syntax, and Mplus syntax are available from the corresponding author upon request.
Public Significance Statement.
Social support regarding one’s transgender/nonbinary identity may help to protect transgender/nonbinary people’s mental health from the negative effects of cisgenderism (i.e., anti-transgender marginalization). In this study, we developed and validated a scale that measures five types of transgender/nonbinary identity support from one’s romantic partner.
Acknowledgments:
The authors are grateful to colleagues at the Yale LGBTQ Mental Health Initiative for valuable feedback during the scale development process, to the anonymous peer reviewers for valuable feedback on the manuscript, and to all participants for sharing their time and experiences.
Funding:
This research was supported by the National Institute on Alcohol Abuse and Alcoholism under Award Number K99AA030601 (PI: G. R. Murchison) and pilot funding from the Yale University Women Faculty Forum (PI: G. R. Murchison). G. R. Murchison was also supported by the National Institute of Mental Health under Award Number T32MH020031 (PI: T. S. Kershaw). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or other funders. The authors have no conflicts of interest to disclose.
Footnotes
A preprint of this manuscript was posted to PsyArXiv (https://doi.org/10.31234/osf.io/z9fde).
References
- Ashley F (2021). ‘Trans’ is my gender modality: a modest terminological proposal. In Erickson-Schroth L (Ed.), Trans Bodies, Trans Selves (2nd ed.). Oxford University Press. [Google Scholar]
- Beischel WJ, Schudson ZC, Hoskin RA, & van Anders SM (2022). The Gender/Sex 3×3: Measuring and categorizing gender/sex beyond binaries. Psychology of Sexual Orientation and Gender Diversity, 10(3), 355–372. 10.1037/sgd0000558 [DOI] [Google Scholar]
- Carone N, Rothblum ED, Bos HMW, Gartrell NK, & Herman JL (2021). Demographics and health outcomes in a U.S. probability sample of transgender parents. Journal of Family Psychology, 35(1), 57–68. 10.1037/fam0000776 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chen FF (2007). Sensitivity of Goodness of Fit Indexes to Lack of Measurement Invariance. Structural Equation Modeling: A Multidisciplinary Journal, 14(3), 464–504. 10.1080/10705510701301834 [DOI] [Google Scholar]
- Collins PH, & Bilge S (2016). Intersectionality. Polity Press. [Google Scholar]
- Coppola J, Gangamma R, & Hartwell E (2021). “We’re just two people in a relationship”: A qualitative exploration of emotional bond and fairness experiences between transgender women and their cisgender partners. Journal of Marital and Family Therapy, 47(3), 648–663. 10.1111/jmft.12467 [DOI] [PubMed] [Google Scholar]
- Crenshaw K (1991). Mapping the margins: Intersectionality, identity politics, and violence against women of color. Stanford Law Review, 43(6), 1241–1299. 10.2307/1229039 [DOI] [Google Scholar]
- Depaoli S, Tiemensma J, & Felt JM (2018). Assessment of health surveys: fitting a multidimensional graded response model. Psychology, Health & Medicine, 23(sup1), 1299–1317. 10.1080/13548506.2018.1447136 [DOI] [Google Scholar]
- DeVellis RF (2017). Scale development: Theory and applications (4th ed.). SAGE. [Google Scholar]
- Flora DB (2020). Your coefficient alpha is probably wrong, but which coefficient omega is right? A tutorial on using R to obtain better reliability estimates. Advances in Methods and Practices in Psychological Science, 3(4), 484–501. 10.1177/2515245920951747 [DOI] [Google Scholar]
- Fuller KA, & Riggs DW (2021). Intimate relationship strengths and challenges amongst a sample of transgender people living in the United States. Sexual and Relationship Therapy, 36(4), 399–412. 10.1080/14681994.2019.1679765 [DOI] [Google Scholar]
- Funk JL, & Rogge RD (2007). Testing the ruler with item response theory: Increasing precision of measurement for relationship satisfaction with the Couples Satisfaction Index. Journal of Family Psychology, 21(4), 572–583. 10.1037/0893-3200.21.4.572 [DOI] [PubMed] [Google Scholar]
- Galupo MP, Pulice-Farrow L, Clements ZA, & Morris ER (2019). “I love you as both and I love you as neither”: Romantic partners’ affirmations of nonbinary trans individuals. International Journal of Transgenderism, 20(2–3), 315–327. 10.1080/15532739.2018.1496867 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gamarel KE, Sevelius JM, Reisner SL, Coats CS, Nemoto T, & Operario D (2019). Commitment, interpersonal stigma, and mental health in romantic relationships between transgender women and cisgender male partners. Journal of Social and Personal Relationships, 36(7), 2180–2201. 10.1177/0265407518785768 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gignac GE, & Szodorai ET (2016). Effect size guidelines for individual differences researchers. Personality and Individual Differences, 102, 74–78. 10.1016/j.paid.2016.06.069 [DOI] [Google Scholar]
- Gorman KR, Shipherd JC, Collins KM, Gunn HA, Rubin RO, Rood BA, & Pantalone DW (2022). Coping, resilience, and social support among transgender and gender diverse individuals experiencing gender-related stress. Psychology of Sexual Orientation and Gender Diversity, 9(1), 37–48. 10.1037/sgd0000455 [DOI] [Google Scholar]
- Hirschfeld G, & von Brachel R (2014). Improving multiple-group confirmatory factor analysis in R – A tutorial in measurement invariance with continuous and ordinal indicators. Practical Assessment, Research, and Evaluation, 19(7). 10.7275/qazy-2946 [DOI] [Google Scholar]
- Horn JL (1965). A rationale and test for the number of factors in factor analysis. Psychometrika, 30, 179–185. 10.1007/BF02289447 [DOI] [PubMed] [Google Scholar]
- James S, Herman J, Rankin S, Keisling M, & Mottet L (2016). Report of the 2015 U.S. Transgender Survey. National Center for Transgender Equality. https://transequality.org/sites/default/files/docs/usts/USTS-Full-Report-Dec17.pdf [Google Scholar]
- Johnston MM, & Finney SJ (2010). Measuring basic needs satisfaction: Evaluating previous research and conducting new psychometric evaluations of the Basic Needs Satisfaction in General Scale. Contemporary Educational Psychology, 35(4), 280–296. 10.1016/j.cedpsych.2010.04.003 [DOI] [Google Scholar]
- Kaiser HF, & Rice J (1974). Little Jiffy, Mark IV. Educational and Psychological Measurement, 34(1), 111–117. 10.1177/001316447403400115 [DOI] [Google Scholar]
- La Guardia JG, Ryan RM, Couchman CE, & Deci EL (2000). Within-person variation in security of attachment: A self-determination theory perspective on attachment, need fulfillment, and well-being. Journal of Personality and Social Psychology, 79(3), 367–384. 10.1037/0022-3514.79.3.367 [DOI] [PubMed] [Google Scholar]
- Limberger MF, Schmiedek F, Santangelo PS, Reichert M, Wieland LM, Berhe O, Meyer-Lindenberg A, Tost H, & Ebner-Priemer UW (2023). Assessing affect in adolescents with e-diaries: Multilevel confirmatory factor analyses of different factor models. Frontiers in Psychology, 14, Article 1061229. 10.3389/fpsyg.2023.1061229 [DOI] [Google Scholar]
- Meier A, & Allen G (2009). Romantic Relationships from Adolescence to Young Adulthood: Evidence from the National Longitudinal Study of Adolescent Health. The Sociological Quarterly, 50(2), 308–335. 10.1111/j.1533-8525.2009.01142.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- Motter BL, & Softas-Nall L (2021). Experiences of transgender couples navigating one partner’s transition: Love is gender blind. The Family Journal, 29(1), 60–71. 10.1177/1066480720978537 [DOI] [Google Scholar]
- Muñoz-Laboy M, Severson N, Levine E, & Martínez O (2017). Latino men who have sex with transgender women: The influence of heteronormativity, homonegativity and transphobia on gender and sexual scripts. Culture, Health & Sexuality, 19(9), 964–978. 10.1080/13691058.2016.1276967 [DOI] [Google Scholar]
- Murchison GR, Eiduson R, Agénor M, & Gordon AR (2022). Tradeoffs, constraints, and strategies in transgender and nonbinary young adults’ romantic relationships: The Identity Needs in Relationships Framework. Journal of Social and Personal Relationships, 40(7), 2149–2180. 10.1177/02654075221142183 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Muthén LK, & Muthén BO (2017). Mplus User’s Guide (8th ed.). Muthén & Muthén. http://www.statmodel.com/download/usersguide/MplusUserGuideVer_8.pdf [Google Scholar]
- Padgett RN, & Morgan GB (2021). Multilevel CFA with ordered categorical data: A simulation study comparing fit indices across robust estimation methods. Structural Equation Modeling: A Multidisciplinary Journal, 28(1), 51–68. 10.1080/10705511.2020.1759426 [DOI] [Google Scholar]
- Peitzmeier SM, Hughto JMW, Potter J, Deutsch MB, & Reisner SL (2019). Development of a novel tool to assess intimate partner violence against transgender individuals. Journal of Interpersonal Violence, 34(11), 2376–2397. 10.1177/0886260519827660 [DOI] [PubMed] [Google Scholar]
- Peitzmeier SM, Wirtz AL, Humes E, Hughto JMW, Cooney E, & Reisner SL (2021). The transgender-specific intimate partner violence scale for research and practice: Validation in a sample of transgender women. Social Science & Medicine, 291, Article 114495. 10.1016/j.socscimed.2021.114495 [DOI] [Google Scholar]
- Perry NS, Sullivan TJ, Leo K, Huebner DM, O’Leary KD, & Baucom BRW (2021). Using web-based technologies to increase reach, inclusion, and generalizability in behavioral observation research. Journal of Family Psychology, 35(7), 983–993. 10.1037/fam0000856 [DOI] [PubMed] [Google Scholar]
- Platt LF, & Bolland KS (2018). Relationship partners of transgender individuals: A qualitative exploration. Journal of Social and Personal Relationships, 35(9), 1251–1272. 10.1177/0265407517709360 [DOI] [Google Scholar]
- Pulice-Farrow L, Bravo A, & Galupo MP (2019). “Your gender is valid”: Microaffirmations in the romantic relationships of transgender individuals. Journal of LGBT Issues in Counseling, 13(1), 45–66. 10.1080/15538605.2019.1565799 [DOI] [Google Scholar]
- Putnick DL, & Bornstein MH (2016). Measurement invariance conventions and reporting: The state of the art and future directions for psychological research. Developmental Review, 41, 71–90. 10.1016/j.dr.2016.06.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Reisner SL, Moore CS, Asquith A, Pardee DJ, & Mayer KH (2020). Gender non-affirmation from cisgender male partners: Development and validation of a brief stigma scale for HIV research with transgender men who have sex with men (trans MSM). AIDS and Behavior, 24(1), 331–343. 10.1007/s10461-019-02749-5 [DOI] [PubMed] [Google Scholar]
- Rhemtulla M, Brosseau-Liard PÉ, & Savalei V (2012). When can categorical variables be treated as continuous? A comparison of robust continuous and categorical SEM estimation methods under suboptimal conditions. Psychological Methods, 17(3), 354–373. 10.1037/a0029315 [DOI] [PubMed] [Google Scholar]
- Ryu E, & West SG (2009). Level-specific evaluation of model fit in multilevel structural equation modeling. Structural Equation Modeling: A Multidisciplinary Journal, 16(4), 583–601. 10.1080/10705510903203466 [DOI] [Google Scholar]
- Scott SB, Pulice-Farrow L, Huynh KD, Brunett KM, & Balsam KF (2024). The Partner Affirmation of Transgender Identity Scale: Psychometric evaluation and applications. Psychology of Sexual Orientation and Gender Diversity. 10.1037/sgd0000778 [DOI] [Google Scholar]
- Siboni L, Prunas A, & Anzani A (2022). “He helped me in discovering myself.” Rethinking and exploring sexual and gender identity in trans-inclusive relationships. Journal of Sex & Marital Therapy, 49(2), 208–228. 10.1080/0092623X.2022.2092568 [DOI] [PubMed] [Google Scholar]
- Siboni L, Rucco D, Prunas A, & Anzani A (2022). “We Faced Every Change Together”. Couple’s Intimacy and Sexuality Experiences from the Perspectives of Transgender and Non-Binary Individuals’ Partners. Journal of Sex & Marital Therapy, 48(1), 23–46. 10.1080/0092623X.2021.1957733 [DOI] [PubMed] [Google Scholar]
- Stadler G, Snyder KA, Horn AB, Shrout PE, & Bolger NP (2012). Close relationships and health in daily life: A review and empirical data on intimacy and somatic symptoms. Psychosomatic Medicine, 74(4), 398–409. 10.1097/PSY.0b013e31825473b8 [DOI] [PubMed] [Google Scholar]
- Steinmetz H (2013). Analyzing Observed Composite Differences across Groups: Is Partial Measurement Invariance Enough? Methodology, 9(1), 1–12. 10.1027/1614-2241/a000049 [DOI] [Google Scholar]
- Testa RJ, Habarth J, Peta J, Balsam K, & Bockting W (2015). Development of the Gender Minority Stress and Resilience Measure. Psychology of Sexual Orientation and Gender Diversity, 2(1), 65–77. 10.1037/sgd0000081 [DOI] [Google Scholar]
- Trujillo MA, Perrin PB, Sutter M, Tabaac A, & Benotsch EG (2017). The buffering role of social support on the associations among discrimination, mental health, and suicidality in a transgender sample. International Journal of Transgenderism, 18(1), 39–52. 10.1080/15532739.2016.1247405 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Trull TJ, & Ebner-Priemer UW (2020). Ambulatory assessment in psychopathology research: A review of recommended reporting guidelines and current practices. Journal of Abnormal Psychology, 129(1), 56–63. 10.1037/abn0000473 [DOI] [PubMed] [Google Scholar]
- Van Buuren S, & Groothuis-Oudshoorn K (2011). mice: Multivariate Imputation by Chained Equations in R. Journal of Statistical Software, 45(3), 1–67. 10.18637/jss.v045.i03 [DOI] [Google Scholar]
- Wilhelm P, & Schoebi D (2007). Assessing mood in daily life - Structural validity, sensitivity to change, and reliability of a short-scale to measure three basic dimensions of mood. European Journal of Psychological Assessment, 23(4), 258–267. 10.1027/1015-5759.23.4.258 [DOI] [Google Scholar]
- Williams VF, Smith AA, Villanti AC, Rath JM, Hair EC, Cantrell J, Teplitskaya L, & Vallone DM (2017). Validity of a subjective financial situation measure to assess socioeconomic status in US young adults. Journal of Public Health Management and Practice, 23(5), 487–495. 10.1097/PHH.0000000000000468 [DOI] [PubMed] [Google Scholar]
- Willis GB (2004). Cognitive interviewing: A tool for improving questionnaire design (1st ed.). Sage Publications. [Google Scholar]
- Wolf EJ, Harrington KM, Clark SL, & Miller MW (2013). Sample Size Requirements for Structural Equation Models: An Evaluation of Power, Bias, and Solution Propriety. Educational and Psychological Measurement, 73(6), 913–934. 10.1177/0013164413495237 [DOI] [Google Scholar]
- Zimet GD, Dahlem NW, Zimet SG, & Farley GK (1988). The Multidimensional Scale of Perceived Social Support. Journal of Personality Assessment, 52(1), 30–41. 10.1207/s15327752jpa5201_2 [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.