The Role of Sense of Belonging in Self-Stigma among People with Serious Mental Illnesses

Abstract

Objective

Self-stigma significantly impacts people with serious mental illnesses. Evidence from other marginalized groups indicates that sense of belonging may buffer these impacts. The purpose of this study was to assess buffering of self-stigma by sense of belonging among this population, including the relationship between these effects and self-identification in other marginalized groups.

Methods

267 adults with serious mental illnesses completed demographic, self-stigma, exposure to stigma, and sense of belonging measures. Regression analyses were conducted to determine whether sense of belonging buffered self-stigma, and, if so, whether those effects vary by race and gender identification.

Results

Sense of belonging buffered self-stigma. Self-identification with other marginalized groups did not impact the buffering effect.

Conclusions and Implications for Practice

Sense of belonging can protect against self-stigma. Self-stigma interventions should integrate components that improve sense of belonging, including community integration. These components should be flexible to meet the cultural context of individual consumers. Key words: Self-stigma, serious mental illness, sense of belonging, race, gender

Self-stigma is the internalization of negative societal beliefs about marginalized groups one belongs to (Dickerson, Somerville, Origoni, Ringel & Parente, 2002; Lucksted et al., 2011). People with serious mental illness face many stereotypes, including that they are dangerous, incompetent, or weak (Link & Phelan, 2001; Rüsch, Angermeyer, & Corrigan, 2005). Internalizing these beliefs can lead to poorer self-esteem and self-efficacy, lower treatment engagement, decreased coping skills, depression, and hopelessness (Drapalski et al., 2013; Yanos, Roe, Markus & Lysaker, 2008).

Internalized stigma is dangerous not only for people with serious mental illness, but for most marginalized groups (e.g., Herek, Gillis & Cogan, 2015; Mossakowski, 2003). Despite diverse efforts to reduce stigmatization at policy, public education, and individual levels (Michaels et al., 2014; Ungar, Knaak & Szeto, 2016; Yanos, Lucksted, Drapalski, Roe, & Lysaker, 2015), stigmatization and internalization remain prevalent.

People experiencing stigma often create psychosocial buffers against internalization (Corrigan et al., 2009; McLaren, Jude, & McLachlan, 2008). For example, sense of belongings within minority ethnic groups helps buffer racism’s effects (Forsyth & Carter, 2012; Iwamoto & Liu, 2010), and is widely associated with positive mental health (Fisher, Overholser, Ridley, Braden & Rosoff, 2015; Sargent, Williams, Hagerty, Lynch-Sauer & Hoyle, 2002; Tew et al., 2011). Thus, determining whether sense of belonging might buffer self-stigma among people with serious mental illness is warranted.

Therefore, this study evaluated how sense of belonging relates to self-stigma among people with serious mental illness, and how that relationship is impacted by self-identification with other marginalized groups. We hypothesized, first, that sense of belonging would buffer self-stigma in people with serious mental illness, and, second, that people with serious mental illness who self-identify with an additional marginalized group (i.e., as African American or female) would experience a larger buffering effect than those who do not.

Method

Data for this study originated from a clinical trial of a group-based self-stigma intervention (Lucksted et al., 2011) using baseline assessments collected pre-randomization. The parent study was pre-approved by the University of Maryland Institutional Review Board, no authors have conflicts to report, and all certify their responsibility for this manuscript.

Participants

The study enrolled 267 adults with serious mental illnesses from five psychosocial rehabilitation programs in urban, suburban, and rural regions of Maryland. The sample was generally congruent with clients of such programs, including 79.1% receiving disability benefits, 55.6% living in supervised housing, 68.7% with ≥12 years of education, and 94.8% unmarried. . See Table 1; diagnoses were obtained via chart review.

Table 1.

Descriptive statistics.

Variable	N	Mean (SD)
Age	268	44.69 (12.34)
	N	Frequency
White	117	43.7%
African-American	124	46.3%
Male	105	39.2%
Female	163	60.8%
Schizophrenia Diagnosis	79	29.5%
Bipolar Disorder Diagnosis	70	26.1%
Schizoaffective Diagnosis	56	20.9%
Depression Diagnosis	23	8.6%
Psychotic Disorder NOS Diagnosis	11	4.1%
Depression with Psychosis Diagnosis	15	15.6%
Other Diagnosis	4	1.5%

Measures

The Self Stigma of Mental Illness Scale (SSMIS; Corrigan, Watson & Barr, 2006; Watson, Corrigan, Larson & Sells, 2007) assesses the social cognitive model of self-stigma’s four components. Our focus on applying stigmatizing beliefs to oneself led us to use only the Application subscale, Cronbach’s alpha = .840.

Sense of Belonging Inventory (SOBI; Hagerty & Patusky, 1995)

The SOBI measures sense of belonging via two subscales: the psychological experience (SOBI-P) and its antecedents (SOBI-A). Since previous studies support SOBI-P’s validity and reliability while SOBI-A shows less consistency (e.g., Hagerty & Williams, 1999), only the 18-item SOBI-P was included. Our analysis indicated that Cronbach’s alpha = .916.

Wahl Stigma and Discrimination Scale (WSD; Wahl, 1999)

The WSD measures stigma and discrimination experiences. We used the WSD total score scale after removing reversed score items, as that stabilized issues with consistency and structure. Our revised WSD had 17 items and Cronbach’s alpha = .861.

Analytic Procedure

Analyses were conducted using SPSS 23.0. First, the psychosocial measures (i.e., SOBI-P, SSMI-Application, and WSD) were tested for bivariate correlations. Second, we tested whether sense of belonging moderated the relationship between exposure to stigma and self-stigma by testing whether sense of belonging (SOBI-P), and exposure to stigma (WSD), predicted self-stigma (SSMI-Application) using multiple regression. In a separate regression step, we tested whether SOBI-P moderated the relationship between the other variables by including the interaction between the SOBI-P and WSD. Third, we tested whether the relationship between SOBI-P and SSMI-Application varied by racial and/or gender identities. These two demographic variables were added to the regression model in the first step of the multiple regression, and were included in an interaction with SOBI-P and WSD in the second step. Last, an interaction between gender and race was added to account for intersection among minority statuses.

Results

See Table 2 for descriptive statistics. All psychosocial measures had significant correlations in expected directions, p < .05 (Table 3).

Table 2.

Target variable descriptive statistics overall and by race and gender.

	WSD (total, reverse items removed)		SSMI- Application subscale		SOBI- Psychological subscale
	M	SD	M	SD	M	SD
Total	20.54	11.22	23.62	13.91	48.72	9.60
White men	20.64	9.91	20.68	12.77	49.11	8.62
White women	20.72	12.76	24.45	16.42	48.60	9.249
Black men	23.30	11.79	24.16	12.02	47.68	11.09
Black women	19.36	9.90	24.66	12.68	49.14	10.08

Table 3.

Correlation matrix.

	SOBI-P	SSMI-Application	WSD
SOBI-P	1	-	-
SSMI- Application	−.289***	1	-
WSD	−.384***	.312***	1

^*p < .05;

^**p < .01;

^***p < .001

Both base and interaction models were significant, p < .001 (Table 4). The interaction model improved the base model (R²Δ= .066, FΔ (1, 205) = 16.411, p < .001. In the interaction model, all three variables were significant, p < .05. The significant interaction between WSD and SOBI-P indicated that at lower levels of exposure to stigma, increased level of sense of belonging is associated with increased self-stigma, but at higher levels of exposure to stigma, increased level of sense of belonging is associated with decreased self-stigma. Analysis of potential gender and race interactions were all nonsignificant, p > .05.

Table 4.

Base and interaction regression models.

Variable	Base Model β(SE)	Interaction Model β(SE)
WSD	.349 (.089)***	1.756 (.358)***
SOBI-P	−.117 (.104)	.533 (.189)**
WSDxSOBI-P	-	−.029 (.007)***
N	209	209
R2	.103	.170

^*p < .05;

^**p < .01;

^***p < .001

Discussion

The bivariate relationships in this data corroborate past research that stigmatizing experiences are associated with increased self-stigma, and that sense of belonging is negatively associated with both, among people with serious mental illness (Brohan, Elgie, Sartorius, Thornicroft & GAMIAN-Europe Study Group, 2010). Thus these constructs are fundamentally relevant to reducing the person-level harm of societal stigmatization, including by enhancing sense of belonging and experiences fostering it.

Our first hypothesis, that sense of belonging moderates the relationship between perceived stigma experiences and self-stigma, was supported. While statistically significant, the moderation effect β weight was small (−.029). Perhaps because of skewness in our sample (SSMI-A scores range 10-90; ours ranged 10–86, M = 23.62, SD = 13.912), the impact of sense of belonging for people with high self-stigma was not fully captured. Given that the relationships among sense of belonging, perceived stigma experiences, and self-stigma varied across levels of these variables, the impact may be different in samples reporting more self-stigma. Yet, our participants’ self-stigma scores echo other studies (Brohan et al., 2010; Linz & Sturm, 2012; Wong, Sands & Solomon, 2010) so our findings may depict typical experiences for people with serious mental illness.

Our second hypothesis, that people with serious mental illness who also identified with another marginalized group would experience a larger buffering effect than people who do not, was not supported; the moderating role of sense of belonging did not vary by race or gender. This underlines considering sense of belonging in all prevention and/or buffering intervention efforts.

Such interventions could help people with serious mental illness optimize positive belonging to maximize buffering against stigma (Cook, Purdie-Vaughns, Meyer & Busch, 2014). For example, by helping participants reflect and act on their cultural identifications, personal values, and community affiliations, participants could increase belonging opportunities and the support and validation they receive from them. Program creators may want to build individualized community integration activities into their stigma interventions.

As more is known about self-stigma moderators, interventions could leverage them to maximize effectiveness (Mora-Rios, Ortega-Ortega, Natera, 2015). For example, helping people with serious mental illness draw on within-community affiliations ((Mossakowski, 2003; Fingerhut, Peplau & Gable, 2010) when relevant could boost buffering effects. For those who have few community ties, interventions could foster identifying communities of choice and ways to increase connections (Wong et al., 2010), including roles in the mental health community.

It is important to consider the “other side” of the moderation-- that for people reporting fewer stigma experiences, sense of belonging is associated with higher self-stigma. This relationship is counterintuitive and has no clear origin. One consideration is that our measures did not asses what community participants were thinking of when responding. Most people have multiple community affiliations (e.g., mental health consumers, neighborhood/geographic, religious, cultural). It is possible that choice of a referent community is salient in predicting self-stigma. For example, a religious community may be less accepting of mental health symptoms than the mental health consumer community. Nonetheless, this puzzling result needs to be clarified.

Conclusions

The parent study had several strengths, including high power and a diverse sample representative of a serious mental illness outpatient population. This study tested a novel question about the relationship of sense of belonging to self-stigma, with clinical, community, and policy applications. However, this secondary analysis used cross-sectional data, so conclusions are not causal. Also, its measures did not require participants to identify a community of reference, so we cannot know how different community affiliations may impact these relationships. Considering this in future studies may clarify the complex relationship between sense of belonging and self-stigma.

Among psychiatric rehabilitation clients with serious mental illness, sense of belonging moderated stigma experiences’ relationship with self stigma, such that for participants highly exposed to stigma, sense of belonging buffered against self-stigma, but among others, sense of belonging and self-stigma were positively associated, with no differential impacts by race or gender. Reducing stigma experiences and self-stigma among people with serious mental illness remains critical; these data suggest strategies of increasing sense of belonging and community connectedness.