Racial-Ethnic Differences in Mental Health Stigma and Changes Over the Course of a Statewide Campaign

Mental health–related stigma is a significant public health concern (1–3). Described as the “most formidable obstacle” to progress in mental health by a former U.S. Surgeon General (3), mental health–related stigma not only can discourage people from initiating and adhering to treatment (4–7) but can also create a climate in which people with mental illness are shunned and deprived of opportunities to become fully contributing members of society (5, 6, 8).

The pernicious effects of stigma may be particularly severe for racial-ethnic minority groups who may be deterred from treatment because of “double stigma”: prejudice and discrimination resulting from the two stigmatized social identities of minority status and having a mental illness (9, 10). Moreover, some have posited that mental illness may be more highly stigmatized in racial-ethnic minority communities, given different cultural conceptualizations of mental illness and treatment (11–14). Remarkably, whether stigma is more prevalent among racial-ethnic minority groups has been subject to only limited investigation. With the exception of one study that found no significant differences between Whites and non-Whites (15), the few studies conducted with nationally representative U.S. samples have consistently found that persons of Black, Latino, or Asian American race-ethnicity harbor more negative beliefs, primarily related to perceptions of dangerousness, toward people with mental illness (16–18).

Perceptions of dangerousness along with social distance (i.e., reluctance to interact with members of devalued groups) are two domains of stigma that have been the most difficult to shift in both the United States and other Western countries (19, 20). To investigate the cumulative impact of local and national stigma-reduction campaigns in the United States, Pescosolido et al. (21) examined whether shifts in stigma had occurred between the years 1996 and 2006 (race-ethnicity variations were not explored). Although mental health treatment attitudes had shifted in a positive direction, concomitant reductions in perceived dangerousness and social distance were not observed.

Whether population “shifts” in stigma differ across racial-ethnic groups has been subject to even less empirical investigation. One of the few studies that has examined differential changes in stigma across racial-ethnic groups involved an evaluation of contact-based educational programs (22). Its findings, based on comparisons of prepresentation and immediate postpresentation surveys, revealed that Asian and Latino American participants showed greater reductions in stigma than did Whites. However, that study was limited to individuals who self-elected to attend educational presentations about mental illness. In England, where a national stigma and discrimination program has been running for more than a decade, positive shifts in stigma were observed in 2017 compared with baseline surveys in 2008 and 2009, but these shifts did not appear to differ across racial-ethnic groups (23).

In a landmark effort to reduce mental illness stigma and discrimination in California, a statewide, multifaceted initiative targeting institutional, societal, and individual factors was funded by the Mental Health Services Act and was managed by the California Mental Health Services Authority (24). The initiative was implemented through community organizations and included social marketing campaigns in English and Spanish, distribution of informational resources, efforts to improve media portrayals of mental illness, and thousands of contact-based educational presentations. At the evaluation baseline (spring 2013), most organizations were still building capacity, and implementation and reach were limited (25).

Statewide surveys conducted at that point and repeated 1 year later indicated that social distance decreased during this period but that perceived public stigma had increased (26). However, it is unknown whether shifts in stigma differed across racial-ethnic groups. According to U.S. Census projections, racial-ethnic minority groups are expected to become the majority population in less than 3 decades (27); in some states such as California, racial-ethnic minority groups already constitute the majority (28). To our knowledge, no U.S. study has tracked longitudinal shifts in stigma across racial-ethnic minority groups in the general public during a population-based antistigma campaign.

The purpose of this study was twofold: to examine whether shifts in mental health–related stigma during a population-based antistigma campaign differed across racial-ethnic groups, and to assess whether racial-ethnic disparities in stigma were present during the start of the California initiative and 1 year later. This study reanalyzed data from the aforementioned surveys as well as the 2013 and 2014 waves of the California Statewide Survey (CASS), a surveillance tool following a longitudinal cohort that was developed to track attitudes, beliefs, and behaviors related to mental illness.

Methods

Sample

Study participants were 1,285 individuals who completed both the 2013 and 2014 waves of the CASS, a longitudinal telephone survey of California adults ages ≥18 years enrolled through random-digit dialing to landlines and cellphones. The Field Research Corporation administered surveys by using their computer-assisted telephone interview system. The baseline CASS was conducted from May to September 2013, with 2,006 individuals enrolled May through June; surveys were administered in English or Spanish. An additional oversampling included 567 Black, Chinese, Vietnamese, Cambodian, and Laotian Americans, whose phone numbers were identified through purchased targeted lists and who were surveyed from August to September 2013. Most Asian Americans of the oversample chose to complete the survey in their native language (i.e., Mandarin, Cantonese, Vietnamese, Khmer, or Hmong). The follow-up 2014 CASS was conducted 1 year later (from May to September 2014); 1,285 adults (50% of baseline participants) were reinterviewed. RAND Human Subjects Protection Committee approval and informed consent were obtained.

Measures

Survey items were largely drawn from previous studies that had developed surveys to track mental health–related stigma at the population level (17, 20, 29–32). The 2013 and 2014 CASSs assessed the following stigma domains: social distance, traditional prejudice, perceptions of dangerousness, treatment carryover, disclosure carryover, perceived public stigma, and courtesy stigma (33).

“Social distance,” the desire to be at distance from persons with stigmatized statuses, was assessed by asking participants to rate their degree of willingness to “move next door to,” “spend an evening socializing with,” and “start working closely on a job with” someone who has a mental illness (21, 33). Response options were dichotomized (0, probably/definitely willing; 1, probably/definitely unwilling).

“Traditional prejudice” (i.e., negative stereotypes or beliefs about people with mental illness) was assessed with the item, “People who have had a mental illness are never going to be able to contribute to society much.” “Perceptions of dangerousness” was assessed with the item, “I believe a person with mental illness is a danger to others” (17).

“Treatment carryover,” a belief that public knowledge that an individual has obtained mental health treatment would diminish one’s status within the community, was measured with the following items (31, 34): “Would you put off seeking treatment for fear of letting others know about your mental health problem?” and “If you had a serious emotional problem, would you go for professional help?” Response options were dichotomized (0, probably/definitely not; 1, probably/definitely would).

“Disclosure carryover,” a belief that disclosing a stigmatized condition such as a mental illness will incur negative responses, was assessed with the following items (31): “Would you try to hide your mental health problem from family or friends?” and “Would you try to hide your mental health problem from coworkers or classmates?” Response options were dichotomized (0, probably/definitely not; 1, probably/definitely would).

“Perceived public stigma” (i.e., beliefs about the public’s attitudes and behaviors toward people with mental illness) was measured with the following items: “People with mental illness experience high levels of prejudice and discrimination,” (31) and “People are generally caring and sympathetic to people with mental illness” (17). The aforementioned measures use a 5-point Likert scale ranging from 1 to 5, and response options were dichotomized (0, strongly/moderately disagree/neither agree nor disagree; 1, moderately/strongly agree).

“Courtesy stigma” occurs when those with social ties to individuals with mental illness incur devalued status, also referred to as “stigma by association” (35–37). Courtesy stigma was measured with the following question (38, 39): “If someone in your family had a mental illness, would you feel ashamed if people knew about it?” Response options were dichotomized (0, probably/definitely not ashamed; 1, definitely/probably ashamed).

In addition to these seven domains, “exposure to the social marketing portion of the antistigma campaign” was assessed with eight items in the 2014 CASS asking respondents whether they had been exposed to different marketing activities during the past 12 months. Endorsement of any of the activities was coded as 1 (exposed), and endorsing none was coded as 0 (not exposed).

On the basis of self-reported race, ethnicity, and preferred language of interview, we categorized participants according to the following “race-ethnicity and language” groups: White, Latino-English, Latino-Spanish, Asian-English, Asian–native language, and Black.

Analyses

Weights were applied to align sample characteristics with the characteristics of the California population. Attrition was higher for all non-White respondents. Inverse probability weights were used to account for these differences. The resulting sample roughly represented the general California adult population, although somewhat fewer Latinos were represented, as indicated by the 2013 U.S. Census (40). We report the weighted percentage of respondents who positively endorsed each stigma domain at baseline and follow-up by race-ethnicity and language.

To test for significant shifts in stigma between the baseline and follow-up surveys, we used logistic regression analyses for each of the racial-ethnic groups, predicting each stigma item at follow-up and controlling for its baseline value. To assess whether racial-ethnic disparities were present at the beginning of the initiative and 1 year later, we conducted separate logistic regression analyses predicting each stigma item at baseline and at follow-up from the racial-ethnic groups, with Whites as the reference group. All significance tests were adjusted to control the familywise error rate at 5% within each stigma domain and wave using Holm’s (41) method. Exploratory analyses examined whether any observed disparities at follow-up remained after controlling for exposure to social marketing aspects of the campaign.

Results

Table 1 shows the demographic characteristics of the 1,285 respondents who completed the 2013 and 2014 CASSs. Changes in mental health–related stigma between baseline and follow-up surveys varied by racial-ethnic language group (Table 2). Stigma decreased in all groups except White and Black respondents. Latino-English, Latino-Spanish, and Asian-English respondents all had reductions in social distance. Asian–native language respondents had decreases in disclosure carryover. Interestingly, increases in stigma were observed for White (perceived public stigma), Latino-English (traditional prejudice), Latino-Spanish (perceptions of dangerousness and disclosure carryover), and Asian-English (disclosure carryover) respondents. (For weighted percentages, standard deviations, chi-square test statistics, and p values, see an online supplement to this article.)

Significant racial-ethnic disparities were present at baseline, with the greatest number of differences from Whites observed among Latino-Spanish and Asian–native language respondents (Table 2). Latino-Spanish and Asian–native language respondents had significantly higher levels of stigma on all three social distance items. Latino-Spanish and Asian–native language respondents were more than twice as likely as White respondents to be unwilling to move next door to, socialize with, or work closely with someone experiencing a mental illness. Compared with White respondents, Latino-Spanish and Asian–native language respondents were also about three times more likely to perceive people with mental illness as dangerous and six to seven times more likely to say that people with mental illness are never going to contribute much to society.

Additionally, Asian–native language respondents expressed greater levels of treatment carryover and courtesy stigma than did White respondents. Reverse disparities were also observed among Latino-Spanish and Asian–native language respondents who had lower levels of disclosure carryover and perceived public stigma than White respondents. No disparities with White respondents were observed for Latino-English respondents. Asian-English respondents showed disparities from Whites in social distance and treatment carryover. Black respondents were more likely to endorse traditional prejudice but less likely to endorse perceived public stigma than were White respondents.

One year later, all of the disparities documented among Latino-Spanish and Asian–native language respondents at the start of the initiative persisted, with the exception of treatment carryover stigma in the Asian–native language group (Table 2). Baseline reverse disparities among Latino-Spanish (i.e., disclosure carryover) and Asian–native language (i.e., disclosure carryover and perceived public stigma) respondents also persisted at the follow-up. For Asian-English respondents, baseline disparities (social distance and treatment carryover) were no longer present at the 1-year follow-up.

Exploratory analyses revealed differential exposure to the social marketing aspect of the campaign, with 47% Latino-English, 48% Latino-Spanish, 26% Asian-English, 28% Asian–native language, 50% Black, and 35% White respondents reporting exposure (χ²=31.54, df=6, p<0.001; percentages were weighted). Asian–native language respondents had lower (p=0.03) and Black respondents higher (p=0.02) exposure to the marketing than did White respondents. After controlling for exposure, only a single disparity at follow-up was no longer significant, namely, the Latino-Spanish disparity on social distance (unwilling to work closely with someone with a mental illness).

Discussion

This study is the first to assess whether shifts in mental health–related stigma during the implementation of a population-based antistigma campaign in the United States differed by race-ethnicity and language. Our findings indicate clear group differences, with stigma decreasing in both language groups of Latino and Asian respondents but not among White or Black respondents between the 2013 and 2014 CASSs. Moreover, Latino (English and Spanish) and Asian (English only) respondents showed a more complex pattern, with both increases and decreases across varying stigma domains. The findings raise questions about whether the campaign might have been less effective for White and Black respondents and for certain stigma domains among Latino and Asian respondents.

Our findings also highlight how associations between stigma domains may differ across racial-ethnic groups. For instance, Latino-Spanish respondents showed significant increases in perceptions of dangerousness but decreases in social distance. Previous studies have shown positive associations between perceptions of dangerousness and social distance, but correlations have ranged from 0.2 to 0.6, indicating that these two stigma domains tap distinct underlying beliefs (42, 43). It is unclear whether this pattern for Latino and Asian respondents showing both improvements and exacerbation in stigma levels indicates racial-ethnic differences in the interrelationships between stigma constructs or in the impact of the campaign across different stigma dimensions.

Nonetheless, persistent disparities across several stigma domains were apparent among Latino-Spanish and Asian–native language respondents who reported greater levels of social distance, perceptions of dangerousness, and traditional prejudice on the 2013 and 2014 CASSs. Latino-Spanish respondents did have reductions in social distance toward people with mental illness but not to a degree that eliminated disparities at the 2014 CASS. These findings underscore the importance of accounting for intragroup differences within racial-ethnic minority groups. Treating interview language as an approximate and imperfect indicator of acculturation, we found that reducing stigma disparities may be particularly challenging among immigrant groups with lower acculturation levels.

Respondents who preferred to complete the survey in their native language likely were immigrants who had arrived in the United States as adults and may have maintained cultural conceptualizations from their native countries that may have shaped their attitudes toward people with mental illness. For instance, conceptualizations of mental illness in Asian and Latin American countries may be limited to more severe forms of mental illness, such as psychotic disorders (44–46), which may be related to perceptions of dangerousness. This finding is consistent with prior research showing that racial-ethnic minority groups in the United States and other Western countries harbor greater perceptions of dangerousness compared with their majority counterparts (47).

None of the racial-ethnic minority groups reported greater levels of disclosure carryover or perceived public stigma than White respondents; in fact, Latino-Spanish and Asian–native language respondents were less likely to endorse these stigma domains. Immigrants may prefer to turn to family and friends first for help with mental health problems than to mental health professionals (48), which may account for non-English–speaking respondents being less likely to endorse potentially hiding a mental illness from others. However, despite the observed reverse disparities, disclosure carryover increased among Latino-Spanish respondents between survey waves, which may reflect acculturation to (more negative) views of mental illness in the United States.

Altogether, our findings underscore the importance of cultural influences on not only the salience of particular stigma domains but also potentially on shifts within these stigma domains. The intransigent disparities observed among non-English–speaking groups in social distance, traditional prejudice, and perceptions of dangerousness may negatively affect the recovery and integration of people with mental illness who reside within these communities. Non-English–speaking respondents likely immigrated from Latin America and Asia, where hospital-based care versus community-based care is more prevalent and where experiences of people with mental illness being fully integrated into society may be more limited (49, 50).

Immigrant groups may need more frequent, higher doses of, or better-quality contact with people who have recovered successfully from a mental illness to override negative conceptualizations of mental illness that may be prevalent in their native country and in ethnic enclaves in the United States where public disclosure of mental health problems may be even less the norm than in the broader society (51). Public stigma may translate into internalized stigma and lead to premature treatment termination; addressing stigma within the course of treatment, especially with racial-ethnic minority groups, may be critical to ensuring successful outcomes (52, 53). Although culture has been identified as a key factor that shapes stigma (9, 49), limited research has been conducted in this arena. Much of the research has documented cultural variations in the prevalence of stigma, but few studies have examined why these differences occur (47); such information could be critical to tailoring effective antistigma interventions for culturally diverse populations.

Certain study limitations should be considered. Our study’s use of single-item measures, although drawn from previous population-based surveys, may have limited reliability. In addition, this study’s sample of Asian and Latino groups was not sufficiently diverse or statistically powered to examine potential intragroup differences. Even though the documented shifts in stigma occurred during the time of the campaign, it is uncertain whether they were a direct result of it. Secular trends in stigma or events occurring during the campaign (e.g., high-profile suicides) may have affected public attitudes.

The statewide initiative featured other components tailored for communities comprising African Americans (e.g., a faith-based initiative to create mental health–friendly congregations), Asians (e.g., in-language public service announcements on the radio), or Latinos (e.g., family forums), which could have contributed to shifts in stigma. Moreover, the baseline CASS was administered while the antistigma statewide initiative had already been under way and may not have captured true estimates of stigma preintervention, resulting in a potential underestimation of change. Future studies may be better able to link changes to campaign efforts by implementing intervention activities in certain regions while withholding activities in others to create a control comparison region, as done in a previous study in Germany (54).

Conclusions

This study documented racial-ethnic variations in shifts across multiple domains of stigma in a population-based sample. Previous population-based studies examining racial-ethnic differences have focused on only a few stigma dimensions (16, 18) and have rarely investigated shifts during an antistigma campaign. Evaluating which components of antistigma campaigns are effective across racial-ethnic minority groups is warranted for future research to better understand whether more targeted efforts are needed, especially in light of the persistent and growing racial-ethnic disparities in mental health care (55).