Mental health self-stigma: links with social self-worth contingencies and ally support

ABSTRACT

Self-stigma (also known as internalized stigma) is common among individuals with mental illness, though less is known about its risk factors. We propose that social self-worth contingencies (over-reliance on others’ approval to maintain self-worth) confer risk for self-stigma, and that community connectedness and ally support build resilience against it. The current study investigates links between self-worth contingencies, community/ally support, and self-stigma among 96 undergraduates (M_Age = 20.33, SD = 1.66) from a medium-sized state university who self-reported one or more mental health conditions. We propose that community/ally support mediate the association between self-worth contingencies and self-stigma. Results support hypothesized correlations between social self-worth contingencies, ally support (participant-defined), and self-stigma, though do not support links with community connectedness or researchers’ definition of ally support. Participant-defined ally support predict self-stigma, controlling for self-worth contingencies, though do not meet all conditions for mediation. Rather, self-worth contingencies and ally support independently predict self-stigma. Participants’ descriptions of stigma and suggestions for addressing it are presented. Results have implications for prevention/intervention by addressing self-worth contingencies and fostering ally support, supporting a multi-prong (individual and community) approach to reduce self-stigma.

Clinical impact statement

It is important to identify intrapersonal and interpersonal predictors of mental health self-stigma that inform treatment. The current study identifies self-worth contingencies and ally support as targets for the prevention and intervention of self-stigma.

KEYWORDS:

• self-stigma
• self-worth contingencies
• ally
• support
• mental health

1.

Prior to the Covid pandemic, approximately 39% of young adults (18–25) in the USA experienced a significant mental health condition (Active Minds, n.d..). Rates remain elevated relative to pre-pandemic levels, with 48% of young adults (18–25) experiencing significant anxiety and depression symptoms and 36% reporting “unmet counseling need” (Adams et al., 2022). Individuals with mental health conditions, as well as other marginalized groups, are targets for stigma (Overton & Medina, 2008). Stigma involves negative attitudes and/or treatment towards marginalized groups (Crabtree et al., 2010) and negatively affects self-worth and interferes with help-seeking behavior (Fung et al., 2008; Olfson et al., 1998; Wade et al., 2011). In contrast, social acceptance of mental health conditions has been linked with lower rates of suicide in a study of 25 European countries (Schomerus et al., 2014).

Among individuals with mental illness, three types of stigma have been identified: public stigma (societal negative attitude toward a group that are implicitly or explicitly transmitted in childhood) (Byrne, 2000), structural stigma (institutional policies that enforce the discrimination of a group), and self-stigma (also referred to as internalized stigma) (P. W. Corrigan & O’Shaughnessy, 2007), with public stigma predicting future self-stigma (Vogel et al., 2013). When faced with negative messages about one’s own marginalized group, self-stigma is common (Conde & Gorman, 2009; David, 2013; Lucksted & Drapalski, 2015) and is linked with severity of symptoms (Devylder et al., 2022; Livingston & Boyd, 2010) and exacerbation of mental and physical health problems (Link et al., 1989; Pearl et al., 2017). Self-stigma has been associated with low self-esteem across numerous mental health conditions (e.g., P. W. Corrigan & Shapiro, 2010), including schizophrenia, comorbid substance abuse and schizophrenia-related diagnoses, PTSD, anxiety, borderline personality disorder, social phobia, and bipolar disorder (Aydemir & Akkaya, 2014; Barr et al., 2019; Huang et al., 2018; Lysaker et al., 2012; Ociskova et al., 2013; Rodrigues et al., 2013; Rüsch et al., 2006). Self-stigma is linked with severity of depression, social avoidance, and demoralization, and inversely predicts mental health management, treatment-seeking, and therapeutic engagement (Fung et al., 2008; Pearl et al., 2017; Schomerus et al., 2009) including decreased motivation to follow through on therapeutic goals (P. W. Corrigan & Rao, 2012; P. Corrigan & Watson, 2002; Ritsher & Phelan, 2004). Conversely, feelings of empowerment and hope predict lower self-stigma (Livingston & Boyd, 2010).

1.1. Proposed predictors of self-stigma

The extant research has established that self-stigma is associated with low self-worth, yet little is known about the processes by which self-stigma develops. We propose one vulnerability to self-stigma is contingent self-worth which involves an over-reliance on external indicators for self-worth maintenance (Crocker & Wolfe, 2001). These self-worth contingences may be yoked to one or more domains: academics, physical appearance, activities, and social approval (e.g., “Whether or not others accept me strongly affects my feelings of worth”; Burwell & Shirk, 2006).¹ The current focus is on the latter, that is, an over-reliance on others’ approval to maintain self-esteem. These social self-worth contingencies have also been described in the literature as dependency (Blatt et al., 1976), need for approval (Weissman & Beck, 1978), and sociotropy (Beck et al., 1983; Clark et al., 1995) and have been linked with depressive symptoms and low self-esteem among adolescents and emerging adults (Burwell & Shirk, 2006; Crocker et al., 2003a). In addition, they are associated with anxiety, eating disorders, stress, aggression, risk sexual behavior, alcohol and drug use, and financial and academic difficulties among undergraduates (Crocker, 2002; Crocker & Knight, 2005; Crocker et al., 2003b; Kim & Williams, 2009; Luhtanen & Crocker, 2005).

Given the dependency on external validation, self-worth contingencies confer risk for vulnerable self-esteem (Crocker, 2002). Lacking a strong internal anchor, self-worth plummets in the face of inevitable failures, and efforts to enhance and/or maintain self-esteem are amplified (Crocker et al., 2006). In contrast, an internal locus of worth (i.e., low self-worth contingencies) is linked with more stable self-esteem (Burwell & Shirk, 2006), even in face of unfavorable external events (e.g., interpersonal rejection). However, previous research has not empirically examined how self-worth contingencies may be a risk for mental health self-stigma. We hypothesize that those high in social self-worth contingencies are prone to avoid situations, such as self-disclosure about their mental health condition, that might endanger the approval they depend on for self-worth.

Whereas social self-worth contingencies may place individuals with mental health conditions at risk for self-stigma, those who connect with community and ally supports may experience less internalized stigma. The Surgeon General’s Advisory on Healing Effects of Social Connection and Community (2023) indicates that social isolation, more common among those struggling with mental and physical health, has twice the impact on mortality than obesity and four times the impact of air pollution. In contrast, social support is broadly linked with positive outcomes across the lifespan (Baumeister & Leary, 1995; Chu et al., 2010; Greenblatt et al., 1982; Office of the Surgeon General, 2023; Song & Fang, 2013), and individuals endorsing high social support from mothers, friends, and romantic partners report higher self-worth (Laursen et al., 2006; Snapp et al., 2015). In a study of 34,653 participants, social support was linked with fewer symptoms of clinical depression, anxiety, and/or social phobia in the face of trauma (Moak & Agrawal, 2010). Longitudinal studies among those with severe mental illness show that social support inversely predicts perceived stigmatization over time (Mueller et al., 2006). Further, social support is negatively correlated with self-stigma in both college students and adults with a mental health condition (Chronister et al., 2013; Denenny et al., 2015).

The first facet of social support we propose is community connectedness which involves a positive relationship with those in one’s marginalized group. Research has found that group identification, a specific form of community connectedness, predicts resilience to stigma and stereotype rejection among participants in a mental health support group (Crabtree et al., 2010). Peer education, consisting of social support, information, and motivation, has been shown to reduce self-stigma in those with a mental health condition (Conner et al., 2015). Community connectedness has been negatively correlated with anxiety and depression for those on the trans-female spectrum (Pflum et al., 2015) and positively linked with mental health among Mexican lesbian and bisexual women (Lozano-Verduzco et al., 2019).

Ally support is a second proposed component of community support. An ally is a “member of the agent group [i.e., that with unearned privilege] who rejects the dominant ideology and takes action against oppression out of a belief that eliminating oppression will benefit agents and targets” (Griffin, 1997, p. 76) (see Gibson, 2014 for a review). Allyship traditionally involves advocacy, taking an active role in challenging inequitable treatment and oppression of under-represented and marginalized groups (Anderson & Middleton, 2011), such as white allies who support people of color, heterosexual and cisgender individuals who fight heterosexism and transphobia, and individuals who do not identify with a mental health issue actively advocating for those who do. The current study applies the aforementioned advocacy lens to conceptualizing allyship towards mental health conditions.

1.2. Current study

The current study examines links between internalized/self-stigma, social self-worth contingencies, and community/ally support. We hypothesize that community connectedness and ally support will be inversely associated with both self-stigma and social self-worth contingencies), and that social self-worth contingencies will be associated with higher self-stigma. We propose a mediational model in which community/ally support partially explain the process by which contingent self-worth is linked with higher self-stigma (see Figure 1). Further, qualitative responses among those endorsing high levels of self-stigma, high social self-worth contingencies, and high ally support are analyzed to highlight experiences of stigma and offer insight into how both intrapersonal (social self-worth contingencies) and interpersonal (ally support) factors may be harnessed to reduce mental health self-stigma among emerging adults.

Figure 1. Proposed model.

2. Methods

2.1. Participants

Participants were 96 undergraduate students aged 18 to 27 (M = 20.33, SD = 1.66) from a medium state university in the northeast USA (84.4% female). Of these, 88.5% identified as White, 8.3% as Hispanic or Latinx, 6.3% as Black or African-American, 3.1% as American Indian or Alaskan Native, 1.0% as Asian, and 1.0% as Native Hawaiian or Other Pacific Islander. The majority of participants identified as cisgender (96.9%) and heterosexual (69.8%). Inclusion criteria involved having one or more mental health conditions based on participants’ selection from a checklist.² Most frequently endorsed were anxiety (83.3% of participants) and depression (64.6%). Additionally, 17.7% indicated Obsessive Compulsive Disorder; 13.5% ADHD/ADD (participants were required to have an additional diagnosis); 13.5% eating disorder; 10.4% Post Traumatic Stress Disorder; 5.2% Bipolar Disorder; 5.2% Borderline Personality Disorder; and 6.3% “Other”. Participants were asked to what extent (slightly; moderately; significantly) their mental health condition impacted their life in academics, relationships and/or work; 94.8% reported “moderately” or “significantly”.

2.2. Procedure

Undergraduate students were recruited through flyers and email to participate in a study requiring identification with at least one mental health condition. Following informed consent, they were directed to a confidential online survey and were offered extra credit for their involvement.³ All participants reported at least one mental health condition, indicating that inclusion criteria were met.

2.3. Measures

2.3.1. Internalized Stigma of Mental Illness Inventory − 10-item version (ISMI-10)

The 10-item abbreviated Internalized Stigma of Mental Illness Inventory (ISMI; Boyd et al., 2014) is a reliable and valid alternative to the original 29-item ISMI scale (Hammer & Toland, 2017; Ritsher et al., 2003). It assesses internalized (self) stigma of mental health conditions in the following domains: discrimination experience, alienation, stereotype endorsement, social withdrawal, and stigma resistance (e.g., “I stay away from social situations in order to protect my family or friends from embarrassment”). Items are rated on a 4-point Likert scale (1 = strongly disagree to 4 = strongly agree), with two items reverse-scored. The mean ISMI (self-stigma) score in the current sample was 1.99 (SD = .44) indicating mild internalized/self-stigma (Boyd et al., 2014). Cronbach’s alpha was .80 in the current study.

2.3.2. Community Connectedness: gender Minority Stress And Resilience Measure (GMSR) – adapted

The original 58-item GMSR measure was developed to assess experiences of discrimination, rejection, victimization, internalized/self-stigma stigma, negative expectations, nondisclosure, pride, and community among gender minorities (Testa et al., 2015). We focused specifically on the community connectedness subscale, adapting items to address mental health (e.g., “I feel part of a community of people who share my mental health condition”) given the paucity of assessments available for this form of internalized stigma. Items that did not apply to mental health were omitted (e.g., “It is okay for me to have people know that my gender identity is different from my sex assigned at birth”.), resulting in 40 questions. The full GMSR was administered, though given the theoretical focus of the study, only the community connectedness subscale was analyzed. Items were rated from 0 (strongly disagree) to 4 (strongly agree); higher summed scores indicated greater mental health community connectedness. The original GMSR has demonstrated good reliability and validity in previous studies (Testa et al., 2015). In the current study, the internal consistency of the adapted community connectedness scale was Cronbach’s alpha = .71.

2.4. Ally support

For the purposes of the current study, we developed five theoretically-supported items to assess ally support. Participants identified a) whether they identify having one or more allies related to their mental health condition (“yes”, “no”, or “uncertain”) (Allies1); b) how they would define an ally, in their own words (open-ended response and a manipulation check to assure participant understanding) (Allies2); and c) how many Allies they have (based on participant definition of Allies) (Allies3). Next, they d) explained why these people were allies (Allies4), and e) reported how many people they knew who do not identify with having a mental health condition, but who support and advocate for people who do (Allies5). Items were analyzed individually, with a focus on the researchers’ (Allies5) and participants’ (Allies3) definitions of allies.

2.5. Self-worth contingencies

Two measures of self-worth contingencies were administered. The first was initially developed for use among adolescents and subsequently validated among emerging adults (Burwell & Shirk, 2006; Burwell et al., 2019), and the second was previously validated among undergraduate samples (Crocker et al., 2003b). We were interested in the relative variance each explained and included both measures to enhance the validity of the construct.

2.5.1. Self-Worth Contingency Questionnaire (SWCQ; Burwell & Shirk, 2006)

The 32-item SWCQ (Burwell & Shirk, 2006) measures the degree to which self-esteem is yoked to external feedback in four domains: social acceptance (e.g., “The way I feel about myself as a person depends a lot on what people in my life think of me”), school performance, physical appearance, and activity performance. Items are rated on a six-point Likert scale (i.e., 1 being “not at all true for me” to 6 being “extremely true for me”) with half reverse-scored. Higher scores on the SWCQ indicate more self-worth contingencies. The social subscale was analyzed given the study focus on social and interpersonal constructs (i.e., ally support and community connectedness). The SWCQ questionnaire has good reliability and validity (Burwell & Shirk, 2006; Burwell et al., 2019). In the current sample, internal consistency of the social contingencies subscale was Cronbach’s alpha = .91.

2.5.2. Contingencies of Self-Worth Scale (CSWS; Crocker et al., 2003b)

The CSWS assesses the extent to which self-worth relies on the following domains: Others’ Approval, Appearance, God’s Love, Virtue, Family, Competition, and Academics (Crockeret al., 2003b). Thirty-five items are measured on a 7-point Likert scale (1= Strongly Disagree to 7= Strongly Agree) with subscale and total scores computed by averaging items within and across the domains, respectively. For theoretical and conceptual reasons, only the Others’ Approval subscale score was examined. The CSWS demonstrates good reliability and internal consistency (Crocker et al., 2003b), with a Cronbach’s alpha = .86 for the Others’ Approval subscale in the current study.

2.6. Open-ended responses

In addition to the aforementioned survey, participants were asked to respond to the following open-ended questions: 1) Please share your most negative experience with stigma aimed at your mental health condition(s); and 2) Please provide any suggestions for ways that the university could provide better support for people with mental health conditions. Comments offered by participants high in self-stigma were analyzed, with a focus on those high in self-worth contingencies and ally support.

2.7. Data analytic approach

Analyses were conducted using SPSS version 28.0.11. Pearson correlations were used to assess zero-order relationships among study variables. To assess mediation, we followed Baron and Kenny’s (1986) steps using hierarchical linear regression to establish links between 1) independent and dependent variables; 2) between independent and proposed mediating variables; and 3) between mediating and dependent variables, controlling for the independent variable. The final step of mediation requires that inclusion of the mediating variable significantly reduces the association between independent and dependent variables. A Sobel test of the indirect effect was used to assess this reduction. Finally, we supplemented this test of mediation with Preacher and Hayes (2008) bootstrapping method which uses replacement sampling to estimate confidence intervals for direct and indirect effects.

3. Results

3.1. Quantitative findings

Correlations, means, and standard deviations are presented in Table 1. Social self-worth contingencies, as measured by both the social SWCQ and the CSWS Approval subscales, were related to self-stigma (ISMI) (r = .29 and .30, p’s < .01, respectively) with higher contingencies linked with higher self-stigma. Although the researchers’ definition of ally (Allies5) was not significantly correlated with self-stigma (r = −.17, ns), participants’ definition of ally (Allies3) was inversely related (r = −.29, p < .001), with higher self-reported allies linked with lower levels of self-stigma.

Table 1. Correlations between community connectedness, ally support, Self-Stigma, and Self-worth contingencies

	Social SWCQ	CSWS Others	#Allies3 (participant definition)	#Allies5 (researcher definition)	Community Connected-ness	ISMI (self-stigma)
Social SWCQ	1	.77***	−.25*	−.15	.12	.29**
CSWS Others		1	−.02	−.04	.15	.30**
#Allies3 (participant definition)			1	.57***	.26*	−.29**
#Allies5 (researcher definition)				1	.06	−.17
Community connectedness					1	.06
ISMI (self-stigma)						1
Mean	31.34	4.71	7.65	11.10	16.13	1.99
SD	9.25	1.22	12.79	21.27	3.69	.44

Note: *** p < .001; ** p < .01; * p < .05.

Note: Social SWCQ = Self-Worth Contingency Questionnaire—Social subscale (Burwell & Shirk, 2006; CSWS Others = Contingent Self-Worth Scale—Others Approval subscale (Crocker et al., 2003b); #Allies3 = number of Allies identified based on participants’ definition of an ally; #Allies5 = number of Allies identified based on researchers’ definition of an ally; Community Connectedness= Gender Minority Stress and Resilience Measure (GMSR; Testa et al., 2015); ISMI (self-stigma) = Internalized Stigma of Mental Illness Inventory (Boyd et al., 2014).

Mediation was assessed following both Baron and Kenny’s (1986) and Preacher and Hayes (2008) guidelines (see Figure 2). Researchers’ definition of ally (Allies5) was not correlated with either contingencies or self-stigma, and thus did not meet criteria for mediation. Because the Crocker CSWS Others’ Approval subscale did not meet the second IV to MV criteria for mediation (i.e., there was no significant link between Crocker’s Others’ Approval and participant-defined ally), it was not further assessed. However, the SWCQ social contingencies scale met initial criteria for mediation; the expected IV (self-worth contingencies) to DV (self-stigma) relationship emerged, as did an IV (self-worth contingencies) to MV (participant definition of ally; Allies3) link. The third step of mediation requiring a significant MV (participant-defined ally) to DV (self-stigma) relationship, while controlling for the IV (self-worth contingencies), was supported using hierarchical linear regression (B = −.22, p < .05). Finally, inclusion of the MV (participant-defined ally) in the model reduced the IV (self-worth contingencies) to DV (self-stigma) relationship from B = .29 (p < .01) to B = .24 (p < .05), though this was not a significant decrease (Sobel t = 1.54, p = 0.12). Preacher and Hayes (2008) less conservative bootstrapping method corroborated these findings. Specifically, social self-worth contingencies predicted both participant-defined ally (coefficient = −.35, se = .14, t = −2.42, p =.018, CI = −.63 to −.06) and self-stigma (coefficient = .01, se = .005, t = 2.26, p =.03, CI = .001 to .021). Participant-defined ally predicted self-stigma, controlling for social self-worth contingencies (coefficient = −.008, se = .004, t = −2.09, p =.04, CI = −.015 to −.0004), but the indirect effect of participant-defined ally was not significant (Bootleg effect = −.0026, se = .002, CI = −.004 to .006), suggesting that while social contingencies and participant-defined ally support both predict self-stigma, they do so independently.

Figure 2. Links between social self-worth contingencies, ally support, and self-stigma.

3.2. Open-ended responses

3.2.1. Ally support

To better understand links between participant-defined allyship and study variables, participants’ open-ended responses were examined. From the question, “Please provide your own definition of an ally”, four categories of response emerged: participants 1) explicitly stated that an ally does not have a mental health condition (6.5%); 2) explicitly stated that an ally does have a mental health condition (1%); 3) indicated that an ally is not defined by whether or not they have a mental health condition, but that they provide general support (79.3%); and 4) the definition could not fit into a category (13%). Thus, almost 4/5 of participants’ definitions of an ally involved someone who provided general support whereas the researchers’ definition involved those individuals who do not have a mental health condition advocating for the rights of those who do, suggesting that support rather than advocacy was central to protecting against self-stigma.

3.2.2. Negative experiences

Participants were invited to describe their “most negative experience with stigma aimed at [their] mental health condition(s)”. Using a median split, those high in self-stigma, and those high in contingencies and/or high in allies, were the focus of analyses. Participants identified the following contexts related to their negative and/or stigmatizing experiences: society/people/“they”; friends; peers; parents/family; partner; mental health services; unspecified; and “can’t remember/secret/none” (see Table 2). Thirty-seven percent of individuals high in self-worth contingencies, and 25% of those above the median in self-reported allies, indicated that the source of stigma was “public/people/societal”. For example:

Many people my age take mental health conditions as a joke, and think that it’s my own fault and that I do what I do on purpose. (high contingencies)

People often think that depression is in your control and that you can just “snap out of it”. I wish it was that easy as I could have more of full and complete life. (high contingencies)

Mostly people telling me that it’s my fault I’m like this, that it’s all in my mind and I’m doing it to myself and that I could fix it if I wanted to. (High Allies3)

Table 2. Interpersonal context of negative stigma experience

	High Social Self-Worth Contingencies^a	High Allies3^a
Society/people/“they”	10 (37%)	5 (25%)
Peers	3 (11%)	3 (15%)
Friends	1 (4%)	2 (10%)
Parents/family	1 (4%)	1 (5%)
Partners	2 (7%)	1 (5%)
Mental health services	1 (4%)	1 (5%)
Unspecified	4 (15%)	4 (20%)
Can’t remember/secret/none	4 (15%)	3 (15 %)

^aNote that these categories (High Contingencies and High Allies) were not mutually exclusive, though quantitative data indicated a significant negative correlation (r = −.25, p < .05) between them.

One participant high in both contingencies and ally support described experiencing stigma while also receiving support: “I posted something on Yik Yak about my mental health and someone called me a freak, but other people defended me”.

A number of participants high in social self-worth contingencies conveyed that their mental health condition had been misunderstood (e.g., “My father doesn’t believe depression exists, despite the fact that my mother and I have both struggled with it a lot”; “My sister, being an RN, thinks that because you can’t see my depression, it doesn’t exist. This hurts me a lot, because depression has been a huge part of my life”.).

Asked about any experience of stigma, a participant scoring high in contingencies reported, “None, I haven’t told many people I have a disorder”, supporting the inverse correlation between contingencies and self-reported allies that emerged in quantitative analyses. Similarly, others conveyed that they did not discuss their mental health condition, though were aware of and affected by the public stigma:

I feel like because I don’t discuss my mental health conditions, I’m not a specific target for stigma but there are always the jokes about depression, self-harm, eating disorders. The snide remarks made in psych classes. Doctors not taking conditions seriously.

In high school I was beat up because they thought I was the kid that would snap because I was quiet and to myself but in actuality I’m a good person I would never snap on anyone even if they hurt me. I don’t feel that violence solves problems.

3.2.3. Recommendations for the institution

Participants offered suggestions for ways the university could provide better support and reduce stigma (see Table 3). Of note, participants high in self-reported allies described more satisfaction with current services than those high in social self-worth contingencies (57% vs. 25%, respectively) (e.g., “So far, I think that [the unnamed university] does a pretty good job of having support services available. I feel safe and welcomed for the most part”).

Table 3. Recommendations to the university

	High Social Self-Worth Contingencies^a	High Allies (Participant-defined)^a
No suggestions/happy with services available	5 (25%)	8 (57%)
Recommends group intervention	6 (30%) ^b	4 (29%)
Better access to services	3 (15%)	2 (14%)
Education	2 (10%)	0
Unsatisfied	3 (15%)	0
Unsure	1 (5%)	0

a Note that these categories (High Contingencies and High Allies) were not mutually exclusive, though quantitative analyses indicated a significant negative correlation (r = −.25, p < .05) between them, with those high in social contingencies reporting less ally support.

b Note: 1 of the 6 high SWCQ participants who recommended group intervention reported they would not themselves want to go (e.g., “Possibly some sort of support group, but I would be too anxious to even go. Might help others though”).

There was a trend towards those high in participant-defined number of allies (Allies3) more likely to recommend group interventions than were those high in contingencies (25% vs. 20%, respectively). As one participant stated, “Support groups are helpful because they provide people with more allies”. A participant high in contingencies provides insight into the need for anonymity (“I don’t use any of the services at [unnamed] University, but being able to stay anonymous when going to get treatment is really nice”). In contrast, several participants high in contingencies (but none high in self-reported Allies) encouraged education as an intervention (e.g., “Providing free alternative types of workshops [such as] breathing, meditation, etc. on how to handle mental disorders without conventional medicine would be nice”). These responses point to the diverse stigma-reducing interventions that may differentially appeal to students depending on their level of contingent self-worth and their perceived ally support.

Several participants shared negative experiences that have implications for faculty within the university (e.g., “When I told one of my professors who would continually call on me in class that I had anxiety, and that her calling on me multiple times every class period, more than others, really triggered my anxiety, she told me that in the real world getting nervous isn’t an excuse to not do your work”).

4. Discussion

We hypothesized that community support (in the form of community connectedness and ally support) would mediate the relationship between self-worth contingencies and self-stigma. Indeed, participant-defined ally support was inversely correlated with self-stigma, consistent with the literature. For example, family caregivers of individuals with severe mental illness experienced less internalized stigma when they had access to social support (Guan et al., 2020). Similarly, Chronister et al. (2013) found that emotional support was inversely related to internalized stigma and together they mediated the relationship between public stigma and mental health recovery.

Our findings indicate that social self-worth contingencies are also positively associated with self-stigma. This is in line with research showing that those high in contingencies experience greater shame, a component of internalized stigma (Ritsher et al., 2003), when they fail to achieve external validation (e.g., Moya-Garófano & Moya, 2019). Indeed, an effective intervention for self-worth contingencies involves self-compassion practice (Neff & Vonk, 2009) which targets shame, protects against psychopathology, and promotes well-being (Boersma et al., 2015; Neff et al., 2018). Acceptance and Commitment Therapy involves a focus on self-compassion which has been linked with decreased internalized stigma in a range of populations including those identifying as sexual minorities (Yadavaia & Hayes, 2012) and individuals with obesity (Lillis et al., 2009). In the few studies examining ACT for internalized mental health stigma, there have been mixed results (Kao et al., 2023; Luoma et al., 2008). For example, Kao et al. (2023) found no interaction between the 5-week ACT intervention and control groups on self-stigma or “believability of stigmatizing thoughts” among participants with self-reported depression, anxiety, bipolar, or psychosis. However, this interaction emerged at one-month follow-up, with the intervention group endorsing lower levels of “believability in stigmatizing thoughts” relative to the control group. This suggests a lag in the salubrious effect of the intervention. Given that acceptance of emotions and experiences are foundational components of ACT, future research might examine experiential avoidance as a key mechanism involved in contingent self-worth and internalized stigma (Skinta et al., 2015).

As expected, self-worth contingencies were negatively correlated with participant-defined ally support. In turn, our findings suggest that highly contingent participants were less inclined to engage in group intervention to reduce internalized stigma than were those high in ally support. These findings fit with research among adolescents which has found that those high in self-worth contingencies report feeling less trust in and are less likely to confide in a parent when distressed (Shirk et al., 2005). They also endorse more ambivalence towards and suppression of emotion than those low in contingencies (Burwell, 2015). Similarly, Ko et al. Ko et al. (2022) found significant links between social self-worth contingencies, loneliness, and low self-disclosure among undergraduate students. These relationships are moderated by cultural beliefs, with more avoidance and social contingencies among those with collectivistic (versus individualistic) world views (Cheng & Kwan, 2008). As such, prevention and intervention attempting to increase ally support should be both culturally sensitive and attuned to participants contingent self-worth, providing diverse modalities (e.g., individual support, app, or web-based) (e.g., Finkelstein & Lapshin, 2007; Hilty et al., 2015) to tackle internalized stigma.

Sherry et al. (2008) found that perceptions of low social support (though not actual support received) help explain the mechanism by which socially prescribed perfectionism (involving perceived expectations of perfection by others) is linked with depression. Our findings similarly suggest that those high in contingent self-worth (sharing features of socially prescribed perfectionism) may perceive less ally support and, in line with the social disconnection model (Hewitt et al., 2006), disengage socially. As a result, they lack the support that otherwise might protect against internalized stigma. Future research might move beyond self-reported measures of ally support to other-informant measures as well as observational or experimental methodologies to distinguish between actual and perceived ally support.

While many of the study hypotheses were supported, others were not. Although participant-defined ally support predicted lower self-stigma, even when controlling for self-worth contingencies, the reduced correlation was insufficient to meet criteria for mediation. Thus, while ally support does not explain the process by which contingencies and self-stigma are related, it is independently associated with self-stigma. The lack of mediation suggests an independent mechanism linking contingencies with self-stigma and, as previously mentioned, points to the need for alternative prevention and interventions besides those based on allyship. Indeed, it may be misguided and/or ineffective to offer more ally support if a subset of individuals (those high in contingencies) are unable to make use of them.

Of our two proposed forms of support, only participant-defined ally support, and not community connectedness, was correlated with study variables. Although internal consistency of the community connectedness measure was acceptable (a = .71), and its association with ally support lends to its convergent validity, future research with larger samples would clarify to what extend this is a power issue. An alternative conceptualization is that community connectedness best functions as a moderator rather than mediator. Wastler et al. (2020) found that among veterans with significant mental health conditions, the sense of belonging (a form of community connectedness) moderated the link between self-stigma and suicidal ideation. Kaniuka et al. (2019) similarly found that community connectedness interacted with perceived stigma to predict both depression and suicidal behavior. In a more complex model, Petruzzella et al. (2019) found that ethnicity, centrality of sexual orientation to identity, and femininity moderated the association between community connectedness and mental health among gay men. This suggests that intersectionality of social identities (e.g., of race, sexual orientation, gender identity) influences the effects of community connectedness among marginalized groups. Larger sample sizes would also permit disaggregation of data to unpack if and how social identities (e.g., related to race/ethnicity, economic status, and gender and sexual orientation) may independently, additively, or interactively affect one’s experience of connection to community and perceptions of ally support.

A surprising finding involved ally support, which was significantly related to self-stigma and self-worth contingencies only when using participants’ definition of allies. Average number of allies (based on the researchers’ definition) was higher than the average number that participants reported, though researcher-defined ally support was not correlated with study variables. Open-ended responses suggest that participant-defined allyship may have been more strongly and negatively correlated with self-stigma and contingencies because, as 79.3% conveyed in their responses, it was characterized as being supportive or understanding whether or not the ally had a mental health condition. In fact, only 6.5% of participants endorsed the researchers’ definition that an ally must not have a mental health condition.

This conceptualization of allyship departs from much of the LGBTQ and racial justice literatures which emphasize that an ally is not a part of the marginalized group yet advocates politically for those who are. For example, the Safe Zone training program, typically conducted at colleges and businesses, educates heterosexual allies about LGBTQ topics to increase awareness, promote advocacy, and effectuate social change to create a LGBTQ-affirming culture. At the conclusion of training, ally participants display a “Safe Zone” placard on their office or dorm room, indicating their political support of the LGBTQ community (Finkel et al., 2003). In an online intervention targeting stigma related to substance abuse disorders, Beeson et al. (2019) found that allyship (involving use of inclusive language, providing access to resources, displaying a symbol of allyship publicly) reduced public stigma among participants. However, this study did not address internalized stigma, and its focus involved political advocacy more than direct supportive listening. In this way, their study methodology and findings diverge from our results indicating that, more than allies’ advocacy, participants benefit from support and understanding around mental health issues.

Our participants’ view of allyship is consistent with LeMaster and Toyosaki (2022) who note that an ally is not required to endorse a marginalized identity nor to politically advocate for those who do. In line with our focus on emotional (versus advocacy) support, a randomized clinical trial of the Honest, Open, and Proud-College (HOP-C) intervention found that undergraduates engaging in appropriate self-disclosure about their mental health reported reduced self-stigma as well as increased confidence in resources to cope with mental health issues (Conley et al., 2020). These results support our conclusion that the key ingredient is not disclosure per se, but sharing with supportive others who they trust will not discriminate against or reject them (Camacho et al., 2020). Emotional support for individuals with mental health issues may be particularly salient given the affective nature of their struggles (without diminishing the challenges faced by LGBTQIA+ and other marginalized groups, and while recognizing their intersectionality).

Aside from fostering ally support in order to reduce self-stigma, another point of intervention involves targeting self-stigma directly in those experiencing mental illness. Among patients with diverse psychiatric issues attending a partial hospitalization treatment program, cognitive therapy targeting participants’ self-stigma was linked with a decrease in psychiatric symptoms and with enhanced daily functioning (Pearl et al., 2017). Similarly, a meta-analysis found that a peer support program significantly decreased self-stigma among individuals experiencing psychosis (Pyle et al., 2018), supporting the benefits of interventions specifically addressing internalized stigma. Similarly, the Tell Me About Your Day peer support group (Morell, 2015) is available on many university campuses. This program involves students’ sharing about their mental health challenges and daily stressors with like-minded peers who similarly experience mental health challenges, and whose only instruction is to listen actively and empathically. Given that self-stigma is linked with lower engagement in treatment among more highly educated individuals (Hack et al., 2020), addressing public and self-stigma would be particularly important on university campuses. We recommend that future research further elucidate specific aspects of a peer support program that may reduce self-stigma (e.g., others openly sharing their experiences; receiving empathy etc.).

And yet, the current findings indicate that groups and/or peer support programs are differentially appealing to those high versus low on contingent self-worth. As noted previously, those high in contingencies may be reluctant to share in a group format or with peers, risking the very approval they seek to maintain their self-worth (Crocker, 2002). With this in mind, prevention and intervention programs would be enhanced by offering diverse modalities aiming to destigmatize mental illness. These modalities might include peer support, individual and group therapy, and self-help programs involving bibliotherapy and/or web-based apps, the latter allowing for anonymity (e.g., Finkelstein et al., 2007; Hilty et al., 2015; Mills et al., 2020).

In addition to honoring empowerment and choice in intervention (Substance Abuse and Mental Health Services Administration, 2014) for those seeking mental health support and/or those looking to challenge internalized stigma (with some preferring self-help and others peer support), we might assist those high in contingencies in developing a stronger internal anchor which can buffer against internalized stigma (Neff et al., 2018). One such intrapersonal approach is compassion-based therapy, which shifts the focus from increasing self-esteem (seen in traditional cognitive behavioral approaches) to compassionate self-relating, especially when they face disapproval, rejection, and perceived failure.

The current study findings are limited by the sample size, which is smaller than recommended for mediational analyses, by the correlational and cross-sectional design that preclude causal interpretations, and by the exclusive use of self-report and resulting potential for common rater and common method bias. In addition, participants’ self-reported mental health issues were not verified by clinical interview. Clinical diagnosis is recommended to accurately capture mental health diagnoses and to rule out sub-syndromal cases that may have attenuated the findings. Moreover, without clinical confirmation, we cannot determine whether the source of stigma relates to a mental health diagnosis (which may not involve observable behaviors and thus easily concealed; Camacho et al., 2020) and/or to maladaptive behaviors that participants attribute to a mental health issue, but may be better understood through a non-clinical (yet other social identity) lens. Future research might consider use of a dimensional assessment of symptoms which would enable examination of symptom severity as a covariate. It is possible that some individuals who are deeply affected by the stigma of mental health conditions might not have felt comfortable participating in this study, resulting in those with the least self-stigma overrepresented. To address this, we recommend targeting those at high risk for self-stigma (e.g., those in inpatient care) and/or recruiting without mention of mental health symptomatology in order to cast a broader net. Further, future research might examine whether prior involvement in therapy affected participation and/or moderated the current findings.

Another limitation to the current study involves its relative lack of racial diversity. Participants were predominantly white (reflective of the institution), yet research has found lower rates of self-worth contingencies among Black relative to White participants (Zeigler-Hill, 2007) and higher rates of mental health conditions among mixed-race (27% point prevalence over 12 months), Indigenous (22%), and LGBT (37%) adults relative to Asian (15%), Black (16%), Hispanic/Latinx (17%), and White (20%) individuals (NAMI, 2023). Minority stress theory notes that individuals from marginalized groups experience greater stressors than those with structural advantage (Sarno et al., 2020), and Black, Asian, Hispanic-American/Latinx, and mixed race/multi-racial individuals are less likely than White individuals to seek out mental health treatment (NAMI, 2023). Further, minority racial groups have less access to mental health care (Murthy, 2022; Office of the Surgeon General US, Center for Mental Health Services US, & National Institute of Mental Health US, 2001) which may further affect self-stigma. On the other hand, self-stigma related to mental health may involve difficult decisions about when and with whom to self-disclose relative to internalized stigma involving observable difference (e.g., race) (Camacho et al., 2020). Given that current participants identified predominately as white, heterosexual, and cisgender, and due to low power to detect such differences in a small sample, we were unable to examine these important questions. We also note that it may be difficult to separate out stigma related to mental illness from that linked with other forms of marginalization given that most participants occupy numerous and interacting social identities. Future studies require larger, more diverse pool of participants to examine possible differences in, and intersectionality related to, current findings.

In sum, we found support for the hypotheses that ally support was negatively associated (and social self-worth contingencies positively linked) with self-stigma in individuals with mental health conditions, with qualitative findings lending further insight into these relationships. The results have implications for prevention and intervention, inviting a two-pronged approach at interpersonal and intrapersonal levels. Specifically, we might foster interpersonal ally support for those comfortable with self-disclosure, which benefits both the marginalized and allies themselves (Mizock & Page, 2016). An intrapersonal approach (e.g., through web-based self-compassion interventions) may empower highly contingent individuals to make choices about self-disclosure without their self-esteem on the line (Crocker, 2002). This does not preclude the need to shift institutional, systemic, and cultural marginalization of those with mental health conditions, though it offers additional pathways to do so (both/and).

Acknowledgements and declaration of interest statement

We would like to acknowledge and are grateful to the individuals who participated in this study. This work was not funded by an external source, and it has not been published elsewhere. The authors report there are no competing interests to declare.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data are available upon request from the first author.

Notes

1. Crocker has examined additional domains not addressed in the current study: competition and family support (external validation) as well as God’s love and virtue (internal validation). The focus of the current study is on the former and specifically focused on social self-worth contingencies. Both measures of contingencies are included to further validate the SWCQ scale among emerging adults.

2. Mental health conditions were self-reported (e.g., “Do you experience any of the following: depression, anxiety … [etc.]”) (yes/no) and were not confirmed with a diagnostic checklist or interview.

3. Students were offered a number of options for extra-credit such that those who did not want to participate in the current study and/or those who did not identify with a mental health condition were able to choose an alternative activity.