Transgender Dehumanization and Mental Health: Microaggressions, Sexual Objectification, and Shame

Abstract

Using structural equation modeling in a national, nonprobabilistic sample of 292 transgender women and men, this project extends the pantheoretical dehumanization framework by testing direct and indirect relations between dehumanization (i.e., a higher-order construct from experiences of transgender microaggressions and sexual objectification), internalization processes (i.e., internalized transnegativity, self-objectification), shame, and general mental health. The model explained 55% of the variance in general mental health. Direct relations between dehumanization and all internalization processes were positive and significant. Internalized transnegativity and shame were significant, negative, direct predictors of mental health, but neither dehumanization nor self-objectification was a significant direct predictor of transgender mental health. Both self-objectification and internalized transnegativity directly predicted more feelings of shame. However, only shame yielded a significant indirect pathway from dehumanization to mental health. The indirect relations from self-objectification and internalized transnegativity to mental health through shame were significant. Research, advocacy, and clinical implications are discussed.

Compared to their cisgender peers (i.e., people whose gender matches social expectations of their assigned sex at birth), transgender people (i.e., individuals whose gender does not match the social expectations connected to their sex assigned at birth) disproportionately experience poorer mental health (Austin & Goodman, 2017; Bockting et al., 2013; Bouman et al., 2017; Bradford et al., 2013). While researchers have investigated this adverse outcome by attending to discrimination (e.g., Valentine & Shipherd, 2018), fewer have examined dehumanization, an intergroup phenomenon in which minority groups are denied uniquely human characteristics (i.e., experience discrimination) and their human nature (i.e., subjected to sexual objectification; Moradi, 2013). For example, a transgender woman is dehumanized when a person is unfriendly towards her because of how she dresses (discrimination) and when another person makes rude sexual remarks about her body (sexual objectification). The simultaneous investigation of discrimination and sexual objectification is important because transgender people in the United States (U.S.) routinely experience both processes. At the institutional level, for instance, 27 states have harmful or bare minimum protections against discrimination based on one’s gender identity (Movement Advancement Project, 2020). At the interpersonal level, transgender people experience high rates of rape and sexual harassment (e.g., Tebbe et al., 2016), which are forms of sexual violence whose acceptance is associated with objectified bodies (Seabrook et al., 2019). Given this oppressive context, sequalae flowing from environmental discrimination and sexual objectification may explain the variance in transgender mental health.

Pantheoretical Dehumanization: A Model

Pantheoretical dehumanization is the unification of various stigma-based, group-specific, and oppression-oriented frameworks within counseling psychology to explain the affective, cognitive, and physiological consequences of being dehumanized in a single model (Moradi, 2013). Two notable frameworks with relevance to transgender people are discrimination theories (e.g., minority stress and microaggressions; Meyer, 2003; Nadal, 2019a) and objectification theory (Fredrickson & Roberts, 1997). This paper focuses on the ways in which environmental experiences of discrimination and sexual objectification cohere to predict deleterious outcomes through internalization processes. That is, when a transgender person is dehumanized, Moradi’s (2013) pantheoretical framework asserts that: (a) discrimination and sexual objectification are often co-occurring in the person’s social environment (i.e., their lived experience); (b) internalization processes specified by discrimination theories and objectification theory are set in motion; (c) internalization processes interact within the person to trigger affective/cognitive responses; and (d) adverse mental health outcomes result. Accordingly, drawing on Moradi’s (2013) theoretical postulates, we argue that (a) when sexual objectification and microaggressions, or brief everyday forms of discrimination reflecting implicit biases (Nadal et al., 2014), occur in a transgender person’s environment, then (b) self-objectification and internalized transnegativity are probable internalization consequences which, in turn, (c) interact with and stimulate feelings of shame and (d) result in poorer mental health.

Given that pantheoretical dehumanization manifests as discriminatory and objectifying experiences (Moradi, 2013), we expected microaggressions and sexual objectification to be accounted for by a higher-order dehumanization factor in our model. Both processes involve rejecting the transgender person’s humanity. Moreover, transgender microaggressions frequently include the exoticization or fetishization of the transgender body (Anzani et al., 2021; Nadal et al., 2016), thereby overlapping with sexual objectification.

Dehumanization as Microaggressions

Researchers have applied pantheoretical dehumanization to transgender women (Brewster et al., 2019) and men (Velez et al., 2016), and focused on minority stress theory specifically (Meyer, 2003). Most research using a minority stress perspective focuses on major discriminatory events (e.g., harassed for using a bathroom consistent with one’s gender; Balsam et al., 2011). However, focusing on microaggressions is important for several reasons. First, because microaggressions are common and pervasive even when blatant discrimination is prohibited by law (Nadal, 2019a), as transgender people gain protected status in the U.S., dehumanization may be more likely to continue via microaggressions. Second, since microaggressions are sometimes unconscious or unintentional (Nadal et al., 2016), perpetrators of microaggressions may not understand how their actions are harmful (Balsam et al., 2011), which may make this form of discrimination more difficult for transgender people to resolve. Third, microaggressions are often subtle, brief, daily occurrences (Nadal et al., 2014), suggesting that a transgender person is more likely to experience them as opposed to major discriminatory events. Finally, microaggressions have conceptual significance in the pantheoretical model (Moradi, 2013), but are yet untested despite their health-eroding potential (Nadal, 2019b; Nadal et al., 2012), and thus warrant investigation.

Minority stressors in general, and microaggressions in particular, are linked to poor mental health among sexual and gender minorities (SGM; Bockting et al., 2013; Bradford et al., 2013; Nadal, 2019a; Nadal et al., 2011, 2016). Microaggressions targeting sexual minorities are related to depression (Salim et al., 2019), anxiety and distress (Scharer & Taylor, 2018; Woodford et al., 2015), and less behavioral control (Scharer & Taylor, 2018). A longitudinal study where 26.2% of participants identified as transgender found that more frequent experiences of microaggressions were associated with greater psychiatric symptoms (Dyar et al., 2020). Moreover, misgendering (i.e., a type of microaggression) also is associated with psychological distress (McLemore, 2018). Thus, if microaggressions are part of a higher-order dehumanization factor (Moradi, 2013), then a negative relationship with transgender mental health is expected.

Dehumanization as Sexual Objectification

Examining sexual objectification concurrently with microaggressions distinguishes pantheoretical dehumanization from frameworks focusing solely on discrimination. Among transgender people, objectification-based dehumanization has been linked to less behavioral control (Brewster et al., 2019; Velez et al., 2016). However, sexual objectification also is associated with poorer mental health generally. In objectification theory, Fredrickson and Roberts (1997) argued that “psychological consequences […] spring from objectifying treatment” (p. 174). Among cisgender people (Fredrickson & Roberts, 1997), for example, evidence suggests that anxiety (Moradi & Huang, 2008) and depression (Szymanski, 2020) spring from environmental experiences of sexual objectification. Flores et al. (2018) found that transgender people of color were concerned with both appearance anxiety and physical safety anxiety. Evidence also suggests that sexual objectification increases distress for some transgender people (Anzani et al., 2021). Thus, if sexual objectification is part of a higher-order dehumanization factor as hypothesized (Moradi, 2013), then it may detract from transgender mental health.

Internalization Processes

Internalized Transnegativity

Previous studies using pantheoretical dehumanization tested models with an emphasis on objectification-related internalization variables (e.g., body surveillance; Brewster et al., 2019; Velez et al., 2016). Therefore, to test whether discrimination-based internalization processes drive outcomes in the pantheoretical model, this study focuses on internalized transnegativity, or the translation of negative societal messages into debasing self-perceptions of one’s gender identity (Israel et al., 2020), an internalization process predictive of general mental health (Bockting et al., 2020).

While dehumanization as a higher-order construct of microaggressions and sexual objectification has yet to be linked to internalized transnegativity empirically, transgender microaggressions are an expression of societal stigma (Austin & Goodman, 2017) and function as distal stressors (i.e., minority-specific stress in the environment; Arayasirikul & Wilson, 2019; Nadal et al., 2016) which then become internalized proximal structures (i.e., habitual affective and cognitive responses; Meyer, 2003). Therefore, the link is conceptually evident albeit empirically untested. We expected a significant pathway from dehumanization to internalized transnegativity because, as transgender people encounter subtle discrimination (e.g., called by the wrong pronouns) and sexual objectification (e.g., intrusive questions about their genitalia; Anzani et al., 2021), they may start to feel uncomfortable disclosing or embarrassed about their identity (i.e., aspects of internalized transnegativity; Bockting et al., 2020). Internalizing these dehumanizing societal attitudes was expected to detract from mental health given that negative, identity-based self-perceptions increase adverse symptoms (Bockting et al., 2013; Meyer, 2003; Puckett et al., 2020; Puckett & Levitt, 2015). Consequently, internalized transnegativity might partially explain the link between dehumanization and transgender mental health.

Self-Objectification

One intermediate process in the link between objectification-based dehumanization and mental health is self-objectification (Fredrickson & Roberts, 1997; Moradi, 2013). Self-objectification emerges from “internalizing an observer’s perspective” (Fredrickson & Roberts, 1997, p. 180). The main reason nonbinary transgender people were excluded from the present study is due to the potential differences in how self-objectification occurs in transgender nonbinary versus binary groups. Nonbinary people, for instance, may resist sociocultural standards of attractiveness (Cusack & Galupo, 2021), possibly buffering them against self-objectification. On the other hand, binary transgender people might be uniquely susceptible to treating the body as the self (i.e., valuing physical appearance over other personal attributes; Lindner & Tantleff-Dunn, 2017) because (a) there are clear social expectations for female and male bodies, (b) passing as cisgender is a common expectation, and (c) these expectations are frequently signaled by others (Anzani et al., 2021; Bockting et al., 2020; Flores et al., 2018).

Previous research using pantheoretical dehumanization found support for objectification-based internalization processes (Brewster et al., 2019; Velez et al., 2016). Both studies examined body surveillance (i.e., a habit of monitoring bodily appearance; Fredrickson & Roberts, 1997), which is substantively distinct from self-objectification (Lindner & Tantleff-Dunn, 2017; Moradi & Huang, 2008). Since transgender people report internalizing their objectifying experiences (Flores et al., 2018), self-objectification warrants further examination as a direct and indirect variable in the pantheoretical framework for several reasons. First, self-objectification makes it more difficult to value internal characteristics (e.g., intelligence, creativity; Fredrickson & Roberts, 1997), which may result in greater distress. Second, fetishizing microaggressions may translate into treating the body as the self (e.g., “settl[ing] for being the fetish”; Anzani et al., 2021, p. 7). Third, self-objectification is a significant mediator of sexual objectification and mental health in cisgender people (e.g., Jones & Griffiths, 2015).

Affective Response: Feelings of Shame

We included shame (i.e., a painful negative emotion based on the self as undesirable; Gilbert, 1998) as an affective mechanism in the pantheoretical framework because shame is “a health-corrosive emotion” (Scheer et al., 2020, p. 139). General processes like shame flow from discrimination (Hatzenbuehler, 2009), such as when SGM adults report feeling ashamed following microaggressive experiences (Nadal et al., 2011). Body shame is a common correlate of sexual objectification (Watson & Dispenza, 2015) and may generalize to global feelings of shame when compounded by microaggressions. That is, transgender people may not only develop shame about their bodies in dehumanizing interactions but may also receive messages about how their gender/sex/identity “fails” to meet cissexist standards. Perceived failure in meeting social standards generates feelings of shame (Fredrickson & Roberts, 1997).

Shame is associated with variables in the pantheoretical model. Prior research found links between shame, minority stressors, and poorer mental health in samples of SGM individuals (Mereish et al., 2019; Mereish & Poteat, 2015; Scheer et al., 2020), but transgender people tend to be underrepresented (Tebbe et al., 2016). In terms of sexual objection, Flores et al. (2018) found that transgender people of color reported feeling ashamed when others comment on or stare at their bodies. While the scope of shame is narrow in prior research with transgender samples, cisgender women prone to feelings of embarrassment and inferiority were likely to report more objectifying experiences and symptoms of depression (Szymanski, 2020). Because shame involves feelings of being helpless, ridiculous, and disgusting (Harder & Zalma, 1990), transgender people who are dehumanized may feel bad about their gender identity or believe that their worth derives from their potential as sexual objects. Consequently, we tested shame (a) as a direct predictor of mental health and (b) as a mediator between the internalization processes and mental health. Since shame may also flow directly from dehumanization (Nadal et al., 2011; Watson & Dispenza, 2015), we examined the direct path from dehumanization to shame as well.

The Present Study

The present study sought to test whether dehumanization would predict general mental health through internalization processes and shame. Five main hypotheses (H) were specified. We predicted that microaggressions and sexual objectification would be significant positive indicators of a higher-order dehumanization factor (H1). We also predicted that greater dehumanization would directly predict higher levels of shame (H2a), self-objectification (H2b), and internalized transnegativity (H2c) but lower levels of mental health (H2d). We expected higher levels of self-objectification to be associated with higher levels of shame (H3a) and lower levels of mental health (H3b); higher levels of internalized transnegativity to be related to higher levels of shame (H3c) and lower levels of mental health (H3d); and greater shame to have a negative relation to mental health (H3e). Further, significant indirect relations between dehumanization and mental health will be observed through shame (H4a), self-objectification (H4b), and internalized transnegativity (H4c). Finally, significant indirect relations will be observed from internalized transnegativity to mental health (H5a) and from self-objectification to mental health (H5b) through shame.

Method

Procedure and Participants

Following best practices with transgender populations (Tebbe & Budge, 2016), the survey instruments were pilot tested with three trans individuals to ensure cultural appropriateness. Feedback was incorporated (e.g., changing the wording of demographic questions, altering recruitment materials) and ethical approval obtained prior to recruiting a nonprobability sample from online communities (e.g., Tumblr, Reddit, Facebook). After passing a reCAPTCHA test and providing consent, participants completed a demographic form and six measures via REDCap, were debriefed, and had an opportunity to enter a raffle to win a $25 Amazon.com gift card. Data were collected from June 2020 to February 2021. In terms of reproducibility, data and materials were hosted on the Open Science Framework (OSF; Cascalheira, 2020); the project was preregistered.

Data were from a larger data set involving the transgender community (Cascalheira et al., in press). The larger data set was used for an instrument validation study. Of the 618 transgender women and men who started the survey, 185 were removed for incomplete data, 77 for failing at least one attention check, 35 for living outside the U.S., 16 for not entering an age, 11 for identifying as cisgender, and 2 for not providing consent. Hence, 292 responses were analyzed, an adequate sample size for structural equation modeling (SEM; Weston & Gore, 2006). Data were missing completely at random (Little’s MCAR χ²[1896] = 1840, p = .817), the amount of missing data was low (0.16%), the sample size was relatively large, and the subscales evinced adequate internal consistency, so mean imputation was used (Parent, 2013).

Inclusion criteria consisted of being over the age of 18, identifying as binary transgender, and living in the U.S. The 292 participants (M_age = 29.1; S_age = 12.5) reportedly identified as trans men (75.7%) and trans women (24.3%). They reported the following sexual orientations: 45.9% bi+, 14% gay, 11.3% heterosexual, 10.3% queer, 8.9% lesbian, 5.5% other, and 3.8% asexual (0.3% no response). In terms of ethnicity, 73.3% of participants identified as European American, 11.0% multiethnic, 5.8% Latin American, 3.4% African / Caribbean American, 3.4% other, and 3.1% Asian American. They lived on the East Coast (28.8%), in the Midwest (24.3%), West Coast (19.5%), Southwest (11.3%), Southeast (8.2%), Rocky Mountains (6.9%), and Alaska / Hawai’i (0.7%; 0.3% no response). Regarding annual income, 48.6% earned less than $20,000, 29.8% earned $20,000 to $44,999, 16.4% earned $45,000 to $139,999, and 3.8% made more than $140,000 (1.4% no response). In terms of their highest education, participants attended some college (34.9%), held a bachelor’s degree (18.8%), graduated high school (15.4%), went to graduate school (12.0%), earned an associate degree (10.3%), did not graduate high school (5.1%), or held a tradeperson certificate (3.1%).

Measures

Experiences of Sexual Objectification

Self-reported incidents of sexual objectification were assessed with the Interpersonal Sexual Objectification Scale (ISOS; Kozee et al., 2007). Kozee and colleagues originally derived a two-factor solution from the set of 15 items with female participants while Davidson et al. (2013) found a three-factor solution with a sample of college men; both models are conceptually consistent with a higher-order factor of interpersonal sexual objectification. Items, rated on a 5-point Likert-type scale ranging from 1 (never) to 5 (almost always), express body evaluation, unwanted explicit sexual advances, and body gazes (e.g., “How often have you noticed someone leering at your body?”). Higher scores indicate more frequent incidents of sexual objectification. The ISOS is stable over a three-week period (r = .90) and, regarding discriminant and convergent validity, ISOS total scores exhibit moderate to strong correlations with measures of sexist events, body consciousness, and body shame among female individuals (Kozee et al., 2007). In a sample of transgender women, an estimate of internal consistency was excellent (α = .95; Brewster et al., 2019). Cronbach’s coefficient alpha for the present sample was .94.

Transgender Microaggressions

Brief, often subtle, everyday experiences of discrimination based on one’s transgender identity were assessed with the Gender Identity Microaggressions Scale (GIMS; Nadal, 2019b). Although the GIMS presently lacks robust validity evidence, it was the only scale assessing transgender microaggressions. Participants responded to 14 items dichotomously (1 = yes, 0 = no) to indicate whether denial of gender identity, misuse of pronouns, invasion of bodily privacy, behavioral discomfort, or denial of societal transnegativity had occurred within the last six months (e.g., “Someone [e.g., family, friend, co-worker] has asked me personal questions about gender reassignment.”). Higher scores indicate more frequent microaggressions. The measure was normed on both transgender women and men. Internal consistency in the original study was .76 (Nadal, 2019b). In the present sample, Kuder-Richardson formula 20 was strong (.83).

Self-Objectification

The extent to which participants internalize objectification was measured with the Self-Objectification Beliefs and Behaviors Scale (SOBBS; Lindner & Tantleff-Dunn, 2017). Fourteen items measure the internalization of interpersonal perspectives (e.g., “I try to imagine what my body looks like to others [i.e., like I am looking at myself from the outside]”) and the valuing physical attributes over internal qualities (e.g., “My physical appearance says more about who I am than my intellect”). Response choices range from 1 (strongly disagree) to 5 (strongly agree); higher scores indicate higher levels of self-objectification. In terms of validity evidence, significant positive correlations between scores on the SOBBS Total Scale and scores on measures of objectifying experiences, body shame, appearance anxiety, disordered eating, and depressive symptoms were observed (Lindner & Tantleff-Dunn, 2017). The test-retest reliability (r = .89) and internal consistency estimate (α = .91) of the SOBBS Total Scale were excellent. Cronbach’s coefficient alpha in the present study was .88.

Internalized Transnegativity

Discomfort with one’s transgender identity as a result of internalizing cisnormative messages was assessed with the Transgender Identity Survey (TIS; Bockting et al., 2020). Factor analysis confirmed a four-factor solution with 26 items, which range from 1 (strongly disagree) to 7 (strongly agree). Participants responded to statements for subscales of Passing (e.g., “Passing is my biggest concern”), Alienation (e.g., “I’m not like other transgender people”), and Shame (e.g., “I envy people who are not transgender”). The Pride subscale (e.g., “Being perceived as transgender by others is okay for me”) is reverse scored. All items inquire about feelings over the last three months and higher scores indicate greater internalized transnegativity. TIS total scores are negatively correlated with outness and self-esteem and positively correlated with depression, anxiety, and felt stigma (Bockting et al., 2020). One-week test-rest reliability (r = .93) and internal consistency (α = .90) for total scores were excellent. Cronbach’s coefficient alpha in the present sample was .94.

Feelings of Shame

The intermediate process of shame was quantified with the Shame subscale of the Personal Feelings Questionnaire-2 (PFQ2-Shame; Harder & Zalma, 1990). This 10-item subscale measures feelings of shame (e.g., feeling laughable) over the last year on a 4-point scale ranging from 0 (you never experience the feeling) to 4 (you experience the feeling continuously or almost continuously). Higher scores indicate greater feelings of shame. In addition to excellent test-retest reliability (r = .91), PFQ2-Shame exhibits significant positive correlations with depression, self-derogation, and anxiety. In the original study, Cronbach’s coefficient alpha was .78 (Harder & Zalma, 1990). However, in community samples of SGMs, internal consistency estimates were above .90 (Mereish & Poteat, 2015; Scheer et al., 2020). Cronbach’s coefficient alpha for the present sample was strong (α = .89).

Mental Health

Mental health, the main outcome variable of interest, was measured with the five-item version of the Mental Health Inventory (MHI-5; Berwick et al., 1991). Participants responded to questions about depression, anxiety, positive affect, and behavioral control over the last month. The MHI-5 ranges from 1 (all of the time) to 6 (none of the time). The items indicating positive affect are reverse scored. Higher scores indicate greater mental health. In a sample of SGM community members where 32% identified as transgender, Cronbach’s alpha for the MHI-5 was .83 (Gonzalez et al., 2018). In the present study, the internal consistency was excellent (α = .91).

Results

Data cleaning, descriptive analyses, and assumption verification were conducted in R Version 3.6.3. Latent variable SEM was executed with the R package lavaan 0.6–7 (Rosseel, 2020). Two latent variable models were constructed (Weston & Gore, 2006). First, the measurement model used confirmatory factors analysis (CFA) to determine the relationship between the latent constructs and the observed indicators. Second, the structural model tested the hypothetical relationships among constructs. No covariates were included in the structural model. The chi-square, comparative fit index (CFI), the root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR) indicated model fit. Given the sample size (N < 500), goodness-of-fit guidelines for chi-square (nonsignificant), CFI (≥ .90), RMSEA (≤ .10), and SRMR (≤ .10) were used (Weston & Gore, 2006).

Descriptive Analyses and Assumption Testing

Table 1 shows descriptive statistics and bivariate correlations among measures. Histograms, Q-Q plots, skewness (< |0.67|), and kurtosis (< |0.88|) provided evidence for univariate normality (Weston & Gore, 2006), however the skew and kurtosis divided by their standard errors revealed evidence of skew for ISOS (4.68), and kurtosis for GIMS (−3.10), PFQ2-Shame (−2.08), and MHI-5 (−2.84). Multivariate normality was assessed with the R package MVN (Korkmaz et al., 2019). A chi-square Q-Q plot using Mahalanobis D² provided evidence for multivariate normality, as did Mardia’s kurtosis (0.22, p = .825), but Mardia’s skewness (78.45, p = .025) and Henze-Zirkler’s test (HZ = 1.06, p = .004) suggested multivariate nonnormality. Ten multivariate outliers were identified using the quantile method (Korkmaz et al., 2019). Given these moderate normality violations, maximum likelihood estimation with robust standard errors was used (i.e., MLM; Rosseel, 2020; Weston & Gore, 2006).

Table 1.

Descriptive Statistics and Bivariate Correlations Among Latent Variables and Measures

Variable	1	2	3	4	5	6	7	M	SD	Range
1. Sexual objectification	—	53**	32**	.01	.33**	−.27**	—	2.14	0.75	1 – 4.6
2. Transgender microaggressions	.61**	—	.36**	.10	.36**	−.32**	—	7.0	3.52	0 – 14
3. Self-objectification	.43**	.52**	—	.33**	.50**	−.35**	—	3.09	0.69	1.14 – 4.86
4. Internalized transnegativity	.12	.14*	.45**	—	.47**	−.46**	—	4.46	1.19	1.38 – 7
5. Feelings of shame	.37**	.45**	.69**	.60**	—	−.66**	—	29.7	8.35	10 – 50
6. Mental health	−.27**	−.32**	−.48**	−.58**	−.72**	—	—	17.6	5.69	5 – 30
7. Dehumanization	.71**	.86**	.60**	.16**	.52**	−.38**	—	2.14	0.75	1 – 4.6

Note. Latent variable correlations are below the diagonal; measurement-level correlations are above the diagonal.

^*p < .05

^**p < .001

Measurement Model

Individual CFAs were conducted to establish robust evidence for the measurement model (Byrne, 2016) prior to fitting an overall measurement model (Weston & Gore, 2006) and the results are presented in the Supplemental Materials. In cases where fit indices were suboptimal, a “model generating” strategy was used to “locate the source of misfit in the model and to determine a model that better describes the sample data” (Byrne, 2016, p. 8). That is, we examined how the measurement model could be modified to improve fit to the data. Based on individual CFAs, it was evident that: (a) the factor structure of the ISOS did not fit this sample, so an exploratory factor analysis (EFA) was conducted, which yielded support for a four-factor solution that was used in the overall measurement model; (b) modification indices showed that allowing the error variances of four items on the TIS to covary improved model fit, so subscales and individual items were used in the overall measurement model; and (c) individual MHI-5 and PFQ2-Shame items were adequate indicators of mental health and feelings of shame. Subscales were used as observable indicators for transgender microaggressions and self-objectification. Similar to Brewster et al. (2019), a higher-order dehumanization factor was specified in the overall measurement model from two latent variables: transgender microaggressions and sexual objectification (see Figure 1).

Figure 1.

Structural Equation Model of Transgender Dehumanization and Mental Health

Note. A diagram of the structural regression depicting direct relations. Standardized coefficients (standard errors [SE]) are from a completely standardized solution. The covariance of self-objectification and internalized transnegativity also was estimated (Ψ = .64, SE = .09, p < .001), but omitted for clarity. * p < .05 ** p < .01 *** p < .001

The overall measurement model met assumptions of nonstandard CFA (Kline, 2015) and exhibited somewhat inadequate fit, Satorra-Bentler corrected χ²(572) = 1,174.682, p < .001, CFI = .893, RMSEA = .063 (90% CI [.058, .068]), SRMR = .076. Modification indices revealed that the covariance between the error variances of the following items would improve the model fit and be theoretically justifiable (Byrne, 2016; Kline, 2015): PFQ2-Shame items 7 and 10; PFQ2-Shame items 8 and 9; and GIMS subscales 3 and 5. The respecified model yielded acceptable model fit, Satorra-Bentler corrected χ²(569) = 1,130.203, p < .001, CFI = .90, RMSEA = .061 (90% CI [.056, .066]), SRMR = .075. This model did not differ from a model without the higher-order factor, ${\chi }_{\text{diff}}^{\text{2}}$(3) = 2.507, p = .474; however, it was significantly better than a model with shame, self-objectification, and internalized transnegativity estimated as a second higher-order factor, ${\chi }_{\text{diff}}^{\text{2}}$(4) = 28.024, p < .001, as well as a model with only the internalization variables as a single higher-order factor, ${\chi }_{\text{diff}}^{\text{2}}$(3) = 26.211, p < .001. Neither a bifactor model with GIMS and ISOS subscales as indicators of dehumanization, χ²(586) = 2,024.034, p < .001, CFI = .744, RMSEA = .097 (90% CI [.090, .099]), SRMR = .222, nor a bifactor model with all variables except for the MHI-5 items as indicators of dehumanization, χ²(564) = 1,548.169, p < .001, CFI = .824, RMSEA = .082 (90% CI [.077, .087]), SRMR = .175, fit the data as well as the model with the higher-order factor. Thus, for theoretical purposes, the model with the single dehumanization factor was retained and H1 was supported.

Table 1 shows bivariate correlations among latent variables. As expected (H2a–d), dehumanization as sexual objectification exhibited a moderate, positive relationship with feelings of shame and self-objectification and a moderate, negative relationship with mental health (for correlation benchmarks, see Cohen, 1992). Similarly, dehumanization as microaggressions was positively correlated with feelings of shame (moderately) and internalized transnegativity (weakly) and exhibited a moderate, negative correlation with mental health. In terms of internalization processes (H3a–e): self-objectification was strongly, positively associated with shame and moderately, negatively related to mental health; internalized transnegativity was strongly, positively related to shame and strongly, negatively related to mental health; and feelings of shame were strongly, negatively correlated with mental health.

Structural Model

In addition to correlations among latent factors, hypothesis testing derived from the structural component of the SEM. The structural regression model yielded adequate fit to the data, Satorra-Bentler corrected χ²(569) = 1,130.201, p < .001, CFI = .90, RMSEA = .061 (90% CI [.056, .066]), SRMR = .075. As shown in Figure 1, the model explained a large proportion of variance in all endogenous variables except for internalized transnegativity (R² ≥ .36; Cohen, 1992): 36% in self-objectification, 62% in feelings of shame, and 55% in mental health.

Alternative structural models were tested (Kline, 2015). Given that self-objectification is a distinct construct from internalized transnegativity (Fredrickson & Roberts, 1997; Israel et al., 2020), a model without estimating the covariance of self-objectification and internalized transnegativity yielded suboptimal goodness-of-fit and was inferior to the retained model, ${\chi }_{\text{diff}}^{\text{2}}$(1) = 45.158, p < .001. Since previous discrimination can influence perceptions of subsequent discrimination (Meyer, 2003), internalization factors were reversed to predict dehumanization and the model fit was equivalent to the retained model, χ²(569) = 1,130.201, p < .001, CFI = .90, RMSEA = .061 (90% CI [.056, .066]), SRMR = .075, unless the covariance of self-objectification and internalized transnegativity was omitted, χ²(570) = 1,130.114, p < .001, CFI = .90, RMSEA = .061 (90% CI [.056, .066]), SRMR = .075. The model presented in Figure 1 was retained because theory should guide model selection, instead of reversing directional predictions, when model equivalence is present (Thoemmes, 2015). Although (a) internalization processes likely influence perceptions of subsequent dehumanization and (b) temporal precedence was not established in this study, dehumanization precedes the internalization of transnegativity, self-objectification, and shame theoretically (Bockting et al., 2013; Fredrickson & Roberts, 1997; Moradi, 2013).

Figure 1 also shows the completely standardized path coefficients of all hypothesized direct relations. When dehumanization was used as an exogenous variable, the direct paths to the internalization processes were consistent with H2a–d: the direct relations between dehumanization and internalized transnegativity, self-objectification, and shame were significant and positive. Support for the internalization hypotheses was also evident: the direct path from self-objectification to shame was significant and positive (H3a), the direct relations between internalized transnegativity and the endogenous variables shame (H3c) and mental health (H3d) were significant and in the hypothesized directions, and the direct relation between shame and mental health was significant and negative (H3e). However, there was no significant direct relation from self-objectification to mental health (H3b). Furthermore, contrary to H2d, the direct path from dehumanization to mental health was not significant.

Indirect pathways were estimated using standardized coefficients (Kline, 2015; Rosseel, 2020) and are presented in Table 2. The indirect pathways from dehumanization to mental health through self-objectification (H4b) and internalized transnegativity (H4c) were not significant. However, H4a was supported: the pathway through shame was significant. Indirect pathways from internalized transnegativity to mental health through shame (H5a) and from self-objectification to mental health through shame (H5b) were significant. Since only the direct path from internalized transnegativity to mental was significant, the total effect was estimated (Rosseel, 2020). The total effect of internalized transnegativity on mental health through shame was significant, β = −.48, SE = .07, 95% CI [−.62, −.34], p < .001.

Table 2.

Indirect Relations of Study Variables

Predictor	Mediator	Criterion	Standardized		Unstandardized		95% CI of Standardized Indirect Relations
			β	SE	B	SE	LL	UL
Dehumanization	Internalized transnegativiy	Mental health	−0.04	0.022	−0.044	0.024	−0.082	0.002
Dehumanization	Self-objectification	Mental health	0.044	0.073	0.048	0.079	−0.099	0.186
Dehumanization	Shame	Mental health	−0.137*	0.055	−0.149	0.06	−0.244	−0.03
Internalized transnegativity	Shame	Mental health	−0.232**	0.046	−0.222	0.044	−0.322	−0.142
Self-objectification	Shame	Mental health	−0.209**	0.068	−0.342	0.12	−0.343	−0.075

^*p < .05

^**p < .001

Discussion

This project extends the predictive power of Moradi’s (2013) pantheoretical dehumanization framework by considering new, conceptually significant constructs. Three main contributions are evident. First, the model explained 55% of the variance in transgender mental health, suggesting that targeting microaggressions, sexual objectification, self-objectification, internalized transnegativity, or shame in treatment has a reasonable chance of decreasing general mental health concerns. Second, microaggressions overlapped with sexual objectification to contribute significantly to the higher-order dehumanization factor (H1), as theorized in other studies (Anzani et al., 2021; Flores et al., 2018; Moradi, 2013; Nadal et al., 2012), and thus advances the pantheoretical model to be inclusive of microaggressions. Finally, seven of the nine direct pathways postulated by the pantheoretical framework were supported, suggesting that pantheoretical dehumanization warrants greater empirical attention in future SGM research.

Results provided support for most hypotheses and corroborated previous findings. The dehumanization factor related directly to internalized transnegativity and self-objectification (H2b, H2c), thereby supporting earlier work using the pantheoretical framework (Brewster et al., 2019; Velez et al., 2016; Watson & Dispenza, 2015). The higher-order factor also directly predicted general feelings of shame (H2a), which aligns with previous work (Nadal et al., 2011; Watson & Dispenza, 2015). As expected, dehumanization evinced moderate-to-strong correlations with all variables (H2a–d). The significant, moderate correlations (a) between microaggressions and shame, (b) as well as among sexual objectification, shame, and self-objectification, align with findings in cisgender samples (Fredrickson & Roberts, 1997; Moradi & Huang, 2008; Nadal et al., 2011; Szymanski, 2020). In other words, this study suggests that discrimination and objectification should be consider as concurrent stress processes among transgender women and men because when everyday, covert indignities compound the distress of being treated as a body part instead of a multidimensional person, transgender people are more likely to appraise their gender identity negatively, objectify themselves, and feel shame.

With respect to other direct and indirect relations, aspects of the current model agreed with previous findings. The significant direct paths from internalized transnegativity and self-objectification to feelings of shame (H3a, H3c) have been observed in correlational studies (Bockting et al., 2020; Fredrickson & Roberts, 1997; Moradi & Huang, 2008; Watson & Dispenza, 2015). Further, the direct relations from internalized transnegativity and shame to mental health (H3d, H3e) substantiate the adverse effects of both constructs (Bockting et al., 2013; Mereish et al., 2019; Mereish & Poteat, 2015; Puckett & Levitt, 2015; Scheer et al., 2020). The indirect pathway from dehumanization to mental health through feelings of shame was significant (H4a), and the indirect relations from internalized transnegativity and self-objectification to mental health through shame (H5a, H5b) emphasizes the importance of this emotion (Mereish & Poteat, 2015; Scheer et al., 2020; Szymanski, 2020), especially in future dehumanization research. Stated differently, our findings indicate that transgender people who avoid perception as transgender or devalue their internal strengths tend towards feeling ashamed which, in turn, exacerbates depressive and anxious symptoms while reducing positive affect.

Three departures from prior research were also evident. First, although the correlation between microaggressions and mental health confirms earlier findings (Dyar et al., 2020; Nadal et al., 2016), the higher-order factor was not directly predictive of mental health in the structural model (H2d). One explanation for non-significance is the ongoing overt discrimination experienced by the transgender community (Human Rights Campaign, 2020; Nadal et al., 2012) which, in comparison to subtle expressions of hostility, may be a greater driver of mental health disparities until more legal protections are enacted (Bradford et al., 2013). Future research should include measures of overt, hostile, physical, and systemic discrimination as manifestations of dehumanization. Another possibility is that the acute, common occurrence of transgender microaggressions may evoke protective responses to mitigate the direct impact on mental health, such as rationalizing the indignity or avoiding similar situations (Nadal et al., 2014). Indeed, the present sample was drawn from affirmative online forums, so it is possible that participants had greater coping resources to respond adaptively to microaggressions (Nadal et al., 2014, 2016). It is also possible that the MHI-5, which was used traditionally as a screener (Berwick et al., 1991), may not be sensitive to the domain of mental health influenced by microaggressions. Thus, future studies should examine coping responses to delineate for whom dehumanization is most problematic and use a more robust measure of mental health. Moreover, since data collection occurred during COVID-19, a history effect may have been observed; perhaps social lockdown reduced exposure to microaggressions in the present sample.

Second, since dehumanization was comprised of sexual objectification as well (H2d), the nonsignificant direct association contrasts previous work with transgender (Brewster et al., 2019) and cisgender (Jones & Griffiths, 2015; Moradi & Huang, 2008; Szymanski, 2020) samples. Further, self-objectification demonstrated a nonsignificant positive trend (H3b). Brewster et al. (2019) explained the positively trending, nonsignificant direct relation from dehumanization to body dissatisfaction by noting how transgender women may interpret hypersexualization as affirmations of gender identity, an observation reported in other work (Anzani et al., 2021). Another explanation for the positive trend is that self-objectification may be a strategy for transgender people to protect themselves from overt discrimination by emphasizing body appearance to pass as cisgender (Flores et al., 2018). Indeed, monitoring the body to avoid harassment has been suggested in a pantheoretical dehumanization study involving sexual minority men (Watson & Dispenza, 2015). Accordingly, our results call for greater attention to how SGM adults adapt to objectification to maintain safety or facilitate approval in a dehumanizing environment.

Finally, the indirect relations between dehumanization and mental health through self-objectification (H4b) and internalized transnegativity (H4c) were not significant. While the latter was untested in previous work, the former contrasts research focusing solely on sexual objectification (Jones & Griffiths, 2015). Lack of significant indirect effects is unsurprising given that the direct effect of self-objectification was nonsignificant and the explained variance of internalized transngeativity was small.

Implications for Practice, Advocacy, Education/Training, and Research

This is the first study to apply the pantheoretical dehumanization framework to general mental health, calling for future research to test discrimination and objectification concurrently. To our knowledge, it is also the first study to provide evidence linking transgender microaggressions to internalized transnegativity. This association confirms previous work indicating that microaggressions are minority stressors (Arayasirikul & Wilson, 2019; Balsam et al., 2011; Nadal et al., 2016). Moreover, the significant direct and indirect pathways indicate the application of pantheoretical dehumanization to phenomena beyond traditional objectification-based mediators and outcomes (Brewster et al., 2019; Velez et al., 2016; Watson & Dispenza, 2015). That is, we found empirical support for microaggressions within the model which, before the present investigation, was theoretical only (Moradi, 2013) or confirmed using a qualitative approach (Flores et al., 2018). These novel contributions to the basic research on the targets of transgender dehumanization are strengthened by a national sample, robust estimation procedures, reliable measurement, and preregistration. However, the sample was comprised predominately of transgender men assigned female at birth and data collection occurred during COVID-19, so external validity of the present findings requires additional empirical support.

It is notable that shame had the strongest correlation with and direct relation to mental health in this sample, thereby calling on future studies using pantheoretical dehumanization as the conceptual core to consider general feelings of shame. Given the larger proportion of variance explained in shame (62%) and significant indirect relations through shame, general shame (versus traditionally examined body shame) may be a primary driver of poorer mental health in transgender populations. Of course, this large proportion of explainable variance may be contributed, in part, to the shame subscale of the TIS (Bockting et al., 2020). Nonetheless, future research might use causal mediation analyses and daily diary studies to confirm these pathways. It would be useful, for example, to determine if feelings of shame fully mediate the pathway from internalized transnegativity to mental health.

Although a nonsignificant direct path from dehumanization to mental health was surprising, the significant indirect relation from dehumanization to mental health through shame indicates that discriminatory and objectifying processes are most detrimental to transgender mental health when they “get under the skin” (Hatzenbuehler, 2009, p. 708). Taken together with the significant indirect relations from internalization processes to mental health through shame, this study suggests greater attention to intermediate reactions to dehumanization is warranted in future research. Advocates and educators might challenge narratives that over-emphasize overt forms of dehumanization by centering internal processes as important drivers of health inequity.

Several intervention and prevention implications with transgender clients can be drawn from these results. First, results suggest that practitioners should address internalized transnegativity. Interventions should illuminate the ways in which contextual dehumanization contributes to negative affective arousal by fusing with self-appraisals about gender identity. Clinicians might help clients develop a rational coping response to this internalized stigma (Puckett & Levitt, 2015). For example, one might help the client accept phrases like “I feel unhappy because people have treated me with less kindness, decency, and respect than I deserve, not because I am transgender” or “I get lots of daily messages that being trans is not okay, so it makes sense that my mind wants to believe that even though it is false.”

Second, Puckett and Levitt (2015) suggest the use of emotion-focused therapy with gender minority clients. Given the significant influence of shame in the present study, targeting this emotion with positive imagery, expressive enactment, or some other Gestalt principle may be useful. For example, a clinician might encourage transgender clients to imagine a personal hero or role model celebrating their gender identity or offering compassion for their dehumanizing treatment. Scheer et al. (2020) suggested that group therapy could help to reduce the health-depleting influence of shame, so referral to this modality may assist transgender clients struggling with general mental health distress. That is, participating in a trans-specific support group might help the transgender client recognize that they are not alone in their struggles, which might increase self-acceptance and reduce feelings of embarrassment, humiliation, and self-consciousness (i.e., aspects of shame; Harder & Zalma, 1990).

Finally, prevention should target microaggressions and sexual objectification in community spaces. For example, behavioral health providers should advocate for transgender-specific sexual harassment and discrimination policies within their institutions to prevent environmental dehumanization and the internalization processes and affective responses that flow from it.

Limitations

Several limitations are evident. First, this study used a cross-sectional, correlational design, so causality cannot be concluded. Relatedly, identical fit indices between the measurement model and the retained structural model indicate that “no possibility exists of unequivocally confirming” that the directional (structural) model is superior to the correlational (measurement) model (Hershberger & Marcoulides, 2006, p. 38). Thus, future research is needed to confirm or refute the evidence of directionality presented in this paper. We also found equivalent structural models. Although theoretically informed model selection is appropriate in this scenario (Thoemmes, 2015), these findings suggest that longitudinal designs are necessary to advance theory-testing research on pantheoretical dehumanization. Our findings suggest that internalization processes can be predicted by dehumanization and vice versa, which is substantively useful: if internalized transnegativity, for instance, increases one’s sensitivity to subsequent experiences of dehumanization, a compounding effect is probable. That is, even if dehumanization precedes internalization as theorized (Moradi, 2013), internalization may intensify subsequent experiences of sexual objectification and microaggressions. Future research should test this plausible compounding effect.

Second, threats to external validity included (a) recruitment of transgender people who participate in specific, trans-affirming online communities and (b) the lack of racial/ethnic diversity. Third, participants may have deduced the aims of the study, choosing to comply or to rebel, an issue of reactivity (Heppner et al., 2016). Fourth, the sample size, although recommended by Weston and Gore (2006), may have limited the statistical power (Wolf et al., 2013). Fifth, threats to construct validity may have occurred because the GIMS currently lacks robust confirmatory evidence. However, it was the only measure of transgender microaggressions at the time of study conception. Relatedly, scores on the SOBBS have not been normed on transgender individuals, thus the factorial structure of scores may have been a concern. Nonetheless, individual CFAs supported the factor structure of both measures. Concerns about the construct validity of the ISOS should be considered as well, although the follow-up EFA provided evidence for the factorial structure of the scores. Additionally, using measures whose scores are normed on cisgender people is not ideal (e.g., ISOS), but it is consistent with previous research with transgender adults (e.g., Velez et al., 2016).

A sixth limitation derives from using modification indices and nonstandard CFA models, which although appropriate, may have introduce effects that are not replicable and sample-dependent (Byrne, 2016; Kline, 2015). Finally, although Weston and Gore (2006) recommend transforming or removing multivariate outliers, we retained them and used a robust estimation procedure (i.e., MLM) which, while consistent with previous research (Brewster et al., 2019; Velez et al., 2016), should be considered while interpreting the present findings. Moreover, although we used Weston and Gore’s (2006) recommendations to evaluate model fit, stricter thresholds exist (e.g., RMSEA ≤ .05; Kline, 2015). Additional research would address these limitations and expand the clinical utility of pantheoretical dehumanization.

In conclusion, dehumanization influences the mental health of transgender people through internalization processes and shame, although more research is needed to confirm or refute the pathways identified in this study. When working with transgender clients, clinicians should target internalization processes and affective responses, most notably shame, to improve mental health. Activists should attend to both discrimination and objectification in policy work.

Significance of the Scholarship to the Public.

This study advances the roles of microaggressions, negative feelings about one’s gender identity, treating the self like an object, and shame in explaining the mental health disparities observed in the transgender community. When working with transgender clients, clinicians should target internalization processes and shame to improve mental health. Activists should attend to both microaggressions and objectifying treatment in policy work.