Unmasking the Aversive Stigmatizer: Integrating Deliberate and Automatic Cognition in the Study of Mental Health Stigma and Perceived Dangerousness

Abstract

Scholars posit that mental health stigma may be the product of both deliberate (i.e., intentional and conscious) and automatic (i.e., effortless and nonconscious) cognitive processes. Yet, few sociological studies have empirically tested this theoretical assumption with valid measures of automatic cognition. This study integrates explicit and implicit stigma measures to investigate the presence of a potentially unique actor in the stigma process, the aversive stigmatizer: individuals who deliberately reject negative cultural stereotypes but still hold implicit mental illness-related biases that may inadvertently influence discriminatory behaviors. Findings suggest that a substantial portion of the sample are aversive stigmatizers and that these actors are disproportionately present among certain social groups. Thus, this study sheds light on a previously hidden population that likely contributes to prejudice and discrimination, as well as advances our theoretical understanding of both public stigma and self-stigma.

Keywords

mental health stigma explicit attitudes implicit bias

The persistence of public stigma, or the attitudes and beliefs that motivate individuals to socially avoid, devalue, or fear people with mental illness, is well documented (Link et al. 1999; Pescosolido, Manago, and Monahan 2019). Scholars posit that mental health stigma may be the product of both deliberate (i.e., intentional and conscious) and automatic (i.e., effortless and nonconscious) cognitive processes (Pescosolido et al. 2008). Yet, few sociological studies have empirically tested this theoretical assumption with valid measures of automatic cognition (Miles, Charron-Chénier, and Schleifer 2019; see, however, Phelan et al. 2019). In the context of mental health scholarship, integrating explicit measures that capture the controlled, conscious responses of individuals with implicit measures that tap into automatically or nonconsciously activated bias could reveal novel insights regarding stigma processes.

The following study thus measures both deliberate (i.e., explicit) and automatic (i.e., implicit) cognition when examining mental health stigma. In doing so, this study theorizes the role of the aversive stigmatizer in the stigma process. Distinct from those who explicitly endorse mental illness stereotypes, aversive stigmatizers deliberately reject stereotypes (i.e., outwardly convey an aversion toward mental health stigma) but still hold implicit mental illness-related biases that may or may not inadvertently influence discriminatory behaviors. Given that implicit biases are malleable and can be shaped by the social environment (e.g., Dasgupta 2013), this study also examines if social positioning potentially influences the likelihood of being an aversive stigmatizer as opposed to an explicit stigmatizer (i.e., exhibits explicit bias) or non-stigmatizer (i.e., exhibits neither explicit nor implicit bias).

More specifically, this study uses an aversive stigmatizer framework to investigate if different types of stigmatizers are evenly distributed across social groups (e.g., gender, race, sexual identity) or if aversive stigmatizers are disproportionately found among certain groups within the sample. It could be the case that all manifestations of bias are more prevalent among certain social groups. But, as discussed below, there is reason to suspect that social groups that are more likely to explicitly reject negative stereotypes may not necessarily exhibit less implicit bias. To explore these possibilities and demonstrate the value of an aversive stigmatizer framework, this study incorporates a cognitive test capturing the strength of implicit associations (i.e., mental illness and dangerousness) and commonly used explicit attitude items (i.e., perceived dangerousness).

In sum, this study aims to (1) integrate two methodological approaches (e.g., traditional survey measures and an Implicit Association Test [IAT]) to examine the presence of aversive stigmatizers in the mental health stigma landscape and (2) survey the potential social patterning of aversive stigmatizers compared to non-stigmatizers and explicit stigmatizers. In doing so, it provides a sociological lens through which to investigate empirical questions in a research space predominantly occupied by psychologists (Spencer and Grace 2016) and has implications for studying relevant micro-, meso-, or macro-level processes discussed in the stigma literature.

Background

Sociologists have long posited that automatic cognition plays an integral role in a variety of social processes. Social psychologists have developed a “dual-process” model that provides a framework for understanding two distinct types of cognition experienced during social interactions: deliberate cognition (intentional and conscious) and automatic cognition (effortless and nonconscious) (Lizardo et al. 2016). The dual-process model proposes that both deliberate and automatic cognition influence behaviors and decision-making that contribute to various outcomes (e.g., social inequality), although the precise mechanisms linking cognition and behavior remain disputed (see Leschziner and Brett 2021). Some scholars contend that deliberate cognition is more central to decision-making but interacts with automatic cognition processes, while others argue that automatic cognition predicts human behavior and deliberate cognition then justifies the behavior produced from nonconscious processes (e.g., Vaisey 2009). Nonetheless, both camps agree that methodological approaches that incorporate implicit measures can contribute to our understanding of sociological processes.

The dual-process model has been integrated into theoretical propositions in a wide range of sociological areas, yet few sociological studies have empirically tested these theoretical assumptions with valid measures of automatic cognition (Miles et al. 2019). For example, the role that implicit bias plays in perpetuating racial inequality is often assumed in sociological work but only occasionally measured (e.g., Melamed et al. 2019). Meanwhile, psychologists interested in understanding racial prejudice have routinely conducted empirical studies designed to test the role of automatic cognition (e.g., Greenwald et al. 2009). Psychologists, for example, developed the aversive racism framework to account for individuals who exhibit racial egalitarian views explicitly but still hold implicit racial bias (Dovidio and Gaertner 2004). This explicit-implicit bias divergence may explain key sociological findings regarding racial discrimination, such as the finding that while employers self-report a willingness to hire Black men with a criminal record (i.e., telephone survey), they are significantly less likely to do so in real life (i.e., experimental audit study) (Pager and Quillian 2005).

Aversive racists and people considered to be non-racists respond similarly to traditional, self-report racial attitude measures but respond differently to methodological techniques that assess implicit attitudes, such as tests that use response latency procedures (Gaertner and McLaughlin 1983). Studies integrating both types of measures find that a substantial portion of the population do in fact fall into the “aversive racist” category (Dovidio, Kawakami, and Beach 2003). With psychologists predominantly engaging in the social scientific work that incorporates explicit and implicit measures (e.g., the aversive racism framework), some sociologists have advocated for greater integration of cognitive measures, such as IATs, into sociological studies (Lamont et al. 2017; Schaap, van der Waal, and de Koster 2019).

Incorporating Explicit and Implicit Measures in the Sociology of Mental Health

Sociologists of mental health also have much to gain by exploring explicit and implicit measures together. Some mental health scholars investigating stigma posit that both deliberate and automatic cognition may play important roles in the stigma process (Pescosolido et al. 2008). Yet, few sociologists have formally incorporated implicit measures into their mental health research (see, however, Phelan et al. 2019). While stigma is considered a multidimensional concept (Pescosolido and Martin 2015) and researchers utilize various measures to capture stigma associated with mental illness, the vast majority of these studies tap into different stigma dimensions and stigma-related concepts with the same methodological tool: surveys that include self-report explicit attitude items (see Link et al. 2004).

Sociologists of mental health, for example, often rely on explicit attitude items to better understand perceptions of dangerousness (e.g., Manago, Pescosolido, and Olafsdottir 2019; Martin et al. 2007). Perceived dangerousness involves the fear that people with a “discredited” condition (i.e., mental illness) are likely to engage in violent or aggressive acts toward themselves or others (Pescosolido and Martin 2015). Perceived dangerousness is a commonly examined construct in stigma research because it is one of the strongest drivers of negative attitudes and discriminatory behavior (Martin et al. 2000; Perry et al. 2022; Pescosolido and Martin 2015). The American public often associates mental illness with violence and fear, a belief that has persisted over the past few decades despite improvements in mental health literacy (Martin et al. 2000; Pescosolido 2013). In 2018, for instance, a higher proportion of American respondents viewed individuals with schizophrenia as likely to be violent toward others and in need of coerced treatment than almost 20 years ago (Pescosolido et al. 2019). Perceptions of dangerousness and support for coercive treatment, along with a desire for social distance from individuals with mental illness, are endorsed despite the fact that individuals with mental illness are significantly more likely to be victims of violent crime than perpetrators (Desmarias et al. 2014). While most people with mental illness are not violent, some people with serious mental illness, particularly those who are also contending with substance abuse, are at increased risk of committing violence (Swanson et al. 1990, 2015). The association between mental illness and violence is weak, however, once risk factors for violence (e.g., demographic factors or childhood exposure to violence) are taken into account, suggesting that individuals with and without mental illness engage in violence for similar reasons (Hiday 1995; Skeem and Mulvey 2020). Yet, negative cultural stereotypes regarding mental illness and violence persist in American culture. For example, American respondents are two to three times more likely than respondents in other Western, industrialized nations to view individuals with mental illness as likely to engage in violence toward others (Manago et al. 2019).

Using survey methods that collect explicit attitude measures to understand perceived dangerousness has clear benefits. Mental illness–related attitude items (i.e. 1996, 2006, and 2018 GSS Stigma Modules), for instance, have contributed to our understanding of how perceived dangerousness and other stigma constructs have changed over time (Pescosolido et al. 2019). But like all methodological tools, self-report surveys have limitations. For example, self-report surveys are susceptible to social desirability biases in which respondents may underreport attitudes they perceive to be socially undesirable (Singleton and Straits 1999).

Meanwhile, methodological tools that measure automatic cognition (e.g., IATs) are less open to social desirability biases and provide little opportunity to intentionally influence performance on a particular task (see Kim 2003). Given that explicit attitudes and implicit bias may concurrently contribute to discriminatory behaviors (Greenwald et al. 2009), studying both could prove fruitful. What can studies that incorporate explicit measures (e.g., direct endorsement of the prejudicial belief that individuals with mental illness are inherently dangerous) and implicit measures (e.g., automatic association between a mental illness label and dangerousness) reveal about mental health stigma?

The Aversive Stigmatizer Framework

Examining both explicit and implicit measures of mental health stigma could reveal a unique actor in the stigma process who consciously rejects negative stereotypes but still holds implicit biases toward those with mental illness: the aversive stigmatizer. Based on commonly used explicit survey measures, people can be theoretically classified into two groups: non-stigmatizers (i.e., those exhibiting no stigma or low levels of stigma) and explicit stigmatizers (i.e., those exhibiting higher levels of stigma). People exhibiting low levels of explicit stigma may, for example, be less likely to report desiring social distance from people with mental illness and less likely to report fearing people with mental illness. However, these explicit measures only tap into one of the two processing types highlighted in the dual-process model and could diverge from implicit measures—potentially masking other pathways toward prejudice.

Research Question 1: To what extent does accounting for explicit and implicit bias reveal the presence of aversive stigmatizers?

It is theoretically possible that a substantial portion of the population are aversive stigmatizers given that psychological studies often find that people can exhibit both lower levels of explicit prejudice and higher levels of implicit bias toward a variety of marginalized and low-status groups (Greenwald and Banaji 1995; Greenwald et al. 2009). Like aversive racists who deliberately reject racial stereotypes but simultaneously hold implicit racial biases (e.g., Kuppens and Spears 2014), aversive stigmatizers may deliberately reject mental illness stereotypes but still hold implicit biases toward those with mental illness. Compared to non-stigmatizers who exhibit neither explicit nor implicit bias, aversive stigmatizers may be more likely to inadvertently contribute to mental health stigma and discrimination in subtle, nonconscious ways. Yet, these two types of actors are theoretically conflated when researchers do not examine automatic cognition. Applying the aversive stigmatizer framework could reveal actors who may have a nuanced propensity for engaging in prejudice or discrimination, especially among people who are sensitive to social desirability norms. If it is the case that aversive stigmatizers exist within the population, then another question remains underexplored.

Research Question 2: To the extent that aversive stigmatizers exist in the population, how are they socially patterned?

Given the impact that social environments can have on implicit racial and gender biases (Dasgupta 2013) and the malleability of implicit attitudes, particularly implicit racial bias, in response to short-term interventions (e.g., Joy-Gaba and Nosek 2010), it is certainly possible that social environments and social positioning may also influence implicit biases related to mental illness. In the case of racial prejudice, evidence suggests that aversive racists are particularly prevalent among some social groups (Kuppens and Spears 2014). Psychologists argue, for example, that while White conservatives are especially likely to exhibit both explicit and implicit racial bias (e.g., Jost, Banaji, and Nosek 2004; Tarman and Sears 2005), aversive racism may better characterize the biases of White liberals who espouse non-prejudiced views and outwardly value racial egalitarianism but still exhibit implicit bias (Dovidio and Gaertner 2004; Nail, Harton, and Decker 2003).

Similarly, aversive stigmatizers may be found more often in some social groups. For example, politically liberal respondents are less likely to report explicit stigma and less likely to endorse stigmatizing attributions (i.e., bad character) compared to their politically conservative counterparts (DeLuca et al. 2018; Watson, Corrigan, and Angell 2005). People who identify as politically liberal could also exhibit less implicit biases, or they could have nonconscious biases that diverge from their self-reported attitudes. Women, White people, and people with mental health issues could also potentially exhibit greater divergence between explicit and implicit stigma given that they are more likely to report knowing more people with mental illness (Perry et al. 2022). While studies find that interpersonal contact is often associated with less explicit stigma (e.g., Felix and Lynn 2022), some research on mental health professionals suggest that it may impact implicit biases differently (Kopera et al. 2015). More research is needed to determine whether political ideology, as well as having other social identities and status characteristics (e.g., gender, race, sexual identity, mental health status), is potentially associated with being an explicit, aversive, or non-stigmatizer. This study aims to take a first step in this exploratory process.

In sum, this study is organized around three tasks: measuring common explicit attitude items related to mental health stigma, measuring implicit biases related to mental health stigma, and collecting relevant respondent characteristics. Given that explicit and implicit measures combined create the best predictors of behavior (Greenwald et al. 2009), this methodological approach can potentially identify respondents who do not perceive people with mental illness as inherently dangerous, either explicitly or implicitly. These actors may be especially unlikely to engage in behaviors that would perpetuate prejudice and discrimination. Meanwhile, aversive stigmatizers may be less likely to engage in discriminatory behaviors compared to respondents who explicitly endorse negative mental illness stereotypes, but could still have the potential to contribute to prejudice and discrimination to a lesser degree. Thus, investigating the presence of aversive stigmatizers could contribute to our understanding of nuances in the stigmatization process. Furthermore, applying this methodological approach to a variety of relevant theoretical propositions in the sociology of mental health could illuminate previously masked stigma processes.

Data and Measures

Data for this study come from a custom-designed survey on mental health. First, respondents are asked eight questions on attitudes toward mental health modeled after the General Social Survey stigma modules. Second, respondents are provided instructions and asked to complete an IAT. Next, respondents are asked several questions regarding their demographic characteristics and mental health. The sample consists of 310 respondents recruited and compensated through Prolific Academic, a web-based data source company. Online survey platforms have been increasingly employed for academic research and Prolific has been found to have higher-quality data compared to other online platforms, such as MTurk (Peer et al. 2017). While Prolific can produce higher-quality data and an online survey is convenient for the administration of an IAT, it is important to note that this is an exploratory study with a small, non-random sample. Respondents were invited to participate in a survey hosted by Qualtrics in March 2022. During data collection, 39 respondents did not complete the study due to failing an early attention check question. Data analysis is conducted on a final sample (N = 295) after 15 respondents were dropped for not meeting the recommended response requirements on the IAT (see Greenwald, Nosek, and Banaji 2003). The 15 respondents who were dropped did not significantly differ on demographic characteristics when compared to respondents with valid IAT scores (see Appendix B). Descriptive statistics of the analytic sample are available in Table 2.

Measures

Dependent variables

Explicit perceptions of dangerousness

This study focuses on an important stigma dimension: perceptions of dangerousness, or fear that people with mental illness are likely to engage in violent or aggressive acts toward themselves or others (see Pescosolido and Martin 2015). To measure perceptions of dangerousness, respondents were asked their level of agreement (1=strongly agree, 6=strongly disagree) with the following statement: “mentally ill people are more likely to do something violent or harmful toward other people.” The explicit dangerousness item was analyzed as an ordinal variable (see Appendix A) and was also recoded to create a dichotomous variable (1=somewhat agree, agree, or strongly agree that mentally ill people are more likely to do something violent or harmful towards other people) for the purpose of examining stigmatizer types as described below.

Implicit perceptions of dangerousness

Implicit bias is captured with an IAT, a laboratory tool which quantifies the strengths of automatic associations between different categories by asking respondents to classify 32 stimulus items (e.g., schizophrenia, diabetes, gentle, aggressive) into four stimuli categories: mental illness, physical illness, harmless, dangerous (Nosek, Greenwald, and Banaji 2005; Teachman, Wilson, and Komarovskaya 2006). IATs can be validly administered through online surveys and online-based findings are consistent with lab-based studies (Gosling et al. 2004; Houben and Wiers 2008). While several studies highlight the validity of IATs more generally (e.g., Gosling et al. 2004; Houben and Wiers 2008), it should also be noted that there is not scientific consensus regarding the validity of the IAT and some scholars critique it as a methodological approach (e.g., Blanton and Jaccard 2008; Kinoshita and Peek-O’Leary 2005). This concern is discussed in more detail at the end of the article. Nonetheless, this study uses the IAT to compare implicit bias across different social groups. This aligns with the use of IATs by social psychologists seeking to compare group differences in implicit bias, such as education-based differences (Kuppens and Spears 2014) or politically based differences (Nail et al. 2003).

During the test blocks (Block 3, Block 4, Block 6, Block 7), 32 stimulus items appear on the screen consecutively and the respondent uses the relevant computer keys to match the stimulus item with its stimuli category. For example, during Block 3, a respondent is expected to hit the left-key if “peaceful” appears on the screen in order to sort it into the “harmless” category appearing on the left side of the screen. During Block 3, the respondent is also expected to hit the left key if “schizophrenia” appears on the screen in order to sort it into the “mentally ill” category appearing on the left side of the screen. In contrast, the respondent is expected to hit the right key during Block 3 when “violent” appears on screen in order to sort it into the “dangerous” category and when “diabetes” appears on screen in order to sort it into the “physically ill” category. Consistent with protocol (Nosek et al. 2005), this IAT utilizes four stimulus items per category. See Table 1 for information on the test blocks.

Table 1.

Sequence of Trial Blocks in the Mental Illness (Harmless vs. Dangerous) IAT.

Block	Number of trials	Function	Items assigned to left-key response	Items assigned to right-key response
1	32	Practice	Harmless words	Dangerous words
2	32	Practice	MI items	PI items
3	32	Test	MI items + Harmless words	PI items + Dangerous words
4	32	Test	MI items + Harmless words	PI items + Dangerous words
5	32	Practice	Dangerous words	Harmless words
6	32	Test	MI items + Dangerous words	PI items + Harmless words
7	32	Test	MI items + Dangerous words	PI items + Harmless words

Note. MI= mental illness PI=physical illness.

The “D” score is the preferred algorithm for calculating IAT scores in terms of (1) internal consistency, (2) resistance to artifact associated with speed of responding, and (3) resistance to known procedural influences (Greenwald et al. 2003). This score describes a respondent’s implicit association between the presented concepts, the direction of association, and the strength of association. D scores theoretically range from −2 to 2, although most are less than 1 in absolute value. Scores between −.15 and .15 signify no preference, scores of .16 to .35 signify slight preference, scores of .36 to .65 signify moderate preference, and scores greater than .65 signify a strong preference. Thus, a score greater than .15 suggests that the respondent associates dangerousness with mental illness, reporting implicit bias. These associations are presumed to reflect personal experiences and societal messages related to mental health issues. D scores were calculated for each respondent using the following equation:

D = \frac{\frac{M (Block 7) - M (Block 4)}{SD (Block 7) + (Block 4)} + \frac{M (Block 6) - M (Block 3)}{SD (Block 6) + (Block 4)}}{2}

Error latencies and latencies that were <300 ms or >10,000 ms were recoded as the block mean + 600 ms (Greenwald et al. 2003). The scores of respondents with tests that are majority errors or majority latencies below 300 ms/above 10,000 ms are dropped from analysis as per recommendations (see Nosek et al. 2005). The D score was analyzed as a continuous measure (see Appendix A) and was recoded to create a dichotomous variable (1=has a D score greater than .15, reflecting a preference for implicit association between mental illness and dangerousness) for the purpose of examining stigmatizer types.

Stigmatizer types

The proposed stigmatizer types were measured using both dependent measures mentioned above. Respondents were grouped as explicit stigmatizers if they endorse the explicit dangerousness item (1 = somewhat agree, agree, or strongly agree that mentally ill people are more likely to do something violent or harmful toward other people) regardless of their D score. Respondents were grouped as aversive stigmatizers if they do not endorse the explicit dangerousness item but they have a D score that is suggestive of implicit bias (1 = has a D score greater than .15, reflecting a preference for implicit association between mental illness and dangerousness).¹ Respondents were grouped as non-stigmatizers if they do not endorse the explicit dangerousness item and they do not exhibit a D score that is suggestive of implicit bias. Thus, respondents fall into one of three possible categories: explicit stigmatizers, aversive stigmatizers, or non-stigmatizers.²

Independent variables

Several respondent characteristics were included in the analysis. Key demographic characteristics were included, such as age (18–80), gender (1 = female, 0 = male), and race (1 = racial minority, 0 = White). Political ideology (1 = liberal, 0 = moderate or conservative) was also included given that liberals and non-conservatives are less likely to explicitly endorse stigmatizing attitudes (DeLuca et al. 2018; Watson et al. 2005), but as found in research on aversive racism, liberals may still hold implicit biases toward those with mental illness. Sexual identity (1 = sexual minority, 0 = straight) was also included as a potential variable of interest. At least partially due to minority stressors (e.g., familial rejection and sexuality-based discrimination) (Meyer 2003), sexual minorities are significantly more likely to experience mental health issues (Bostwick et al. 2010; Cochran, Sullivan, and Mays 2003) compared to their straight counterparts, as well as have more social ties with mental health issues. Lastly, mental health status was included given that people with mental illness are less likely to endorse explicit stigmatizing attitudes. Respondents were considered to have poor mental health if they reported experiencing frequent mental distress (i.e., at least 14 days of poor mental health reported in the past 30 days). Over half of respondents are female (54 percent), approximately 29 percent identify as a racial minority, approximately 26 percent of respondents identity as sexual minorities, 48 percent identify as politically liberal, and approximately one-quarter (24 percent) report poor mental health. The average age of respondents is 38.04 (SD = 14.25). See Table 2 for descriptive statistics.

Table 2.

Descriptive Statistics (N = 295).

Variables	Prop.	M	SD	Range
Stigma measures
Explicit perceived dangerousness		3.23	1.22	1-6
Implicit bias		0.10	0.36	−0.81-0.98
Stigmatizer types
Non-stigmatizer	0.26			0-1
Aversive stigmatizer	0.29			0-1
Explicit stigmatizer	0.44			0-1
Respondent characteristics
Age		38.04	14.25	18-80
Age (1 = 35 years or older)	0.50			0-1
Gender (1 = female)	0.54			0-1
Race (1 = racial minority)	0.29			0-1
Political ideology (1 = Liberal)	0.48		.	0-1
Sexual identity (1 = Sexual Minority)	0.26			0-1
Mental health (1 = Poor)	0.24			0-1

Analytic Approach

Analyses were conducted in two stages to (1) determine whether aversive stigmatizers are present in the sample and (2) understand the potential social patterning of stigmatizer types. During the first stage, I created distinct stigmatizer categories and investigated if a substantial proportion of the sample falls into each stigmatizer type (Figure 1). During the second stage, I initially conducted Pearson’s χ² tests to test for group differences in explicit and implicit stigma (Table 3), as well as for group differences in stigmatizer types (Table 4 and Figure 2). I utilized these basic statistics as a simple but effective way to first explore if aversive stigmatizers are present in the sample. Importantly, I then used multinomial logistic regression to model the relationships between respondent characteristics and stigmatizer types (Table 5, Figures 3 –5, and Appendix C). To summarize the effects of the independent variables, I present average marginal effects (AME) rather than coefficients given their utility (Long and Freese 2014; Mize, Doan, and Long 2019). Throughout both stages, I conducted sensitivity analyses, looked for potential model violations, and tested for model fit.

Table 3.

Percentage Distribution of Explicit and Implicit Stigma Across Social Groups.

	Explicit measures			Implicit measures
Variables	Low stigma	High stigma	Chi-square test p value	Low stigma	High stigma	Chi-square test p value
Respondent characteristics
Age						*
Under 35	59.1	40.9		47.7	52.4
35+	52.1	47.9		60.3	39.7
Gender			***
Female	66.7	33.3		49.1	51.0
Male	42.7	57.3		60.0	44.4
Race
Racial minority	54.7	45.4		51.2	48.8
White	56.0	44.0		55.0	45.0
Political ideology	***
Liberal	67.6	32.4		50.0	50.0
Non-liberal	44.4	55.6		57.5	42.5
Sexual identity			***			*
Sexual minority	72.4	27.6		43.4	56.6
Straight	50.2	49.8		57.6	42.4
Mental health status		**			**
Poor mental health	70.0	30.0		40.0	60.0
Good mental health	51.1	48.9		58.2	41.8

***

p < .001. **p < .01. *p < .05.

Note. The p values are from Pearson’s chi-square test.

Table 4.

Percentage Distribution of Stigmatizer Types Across Social Groups.

Variables	Non-stigmatizer(n = 77)	Aversive stigmatizer(n = 87)	Explicit stigmatizer(n = 131)	Chi-square test p value
Respondent characteristics
Age
Under 35	26.2	32.9	40.9
35+	26.0	26.0	47.9
Gender				***
Female	29.6	37.1	33.3
Male	22.1	20.6	57.4
Race
Racial minority	23.3	31.4	45.4
White	27.3	28.7	44.0
Political ideology				***
Liberal	32.4	35.2	32.4
Non-liberal	20.3	24.2	55.6
Sexual identity				**
Sexual minority	31.6	40.8	27.6
Straight	24.4	25.8	49.8
Mental health status				**
Poor mental health	25.7	44.3	30.0
Good mental health	26.2	24.8	48.9

***

p < .001. **p < .01. *p < .05.

Table 5.

Average Marginal Effects from Multinomial Logistic Regression of Stigmatizer Types.

Variables	Non-stigmatizer	Aversive stigmatizer	Explicit stigmatizer
Age	0.01	−0.01	0.01
	(0.05)	(0.05)	(0.06)
Female	0.07	0.14**	−0.21***
	(0.05)	(0.05)	(0.05)
Racial minority	−0.03	0.06	−0.03
	(0.06)	(0.06)	(0.06)
Politically liberal	0.11*	0.04	−0.15**
	(0.05)	(0.05)	(0.05)
Sexual minority	0.05	0.10	−0.15*
	(0.06)	(0.08)	(0.07)
Poor mental health	−0.05	0.12*	−0.08
	(0.06)	(0.06)	(0.07)
N	295	295	295

***

p < .001. **p < .01. *p < .05.

Note. Standard errors in parentheses. Reference category in multinomial logistic regression = explicit stigmatizer.

Results

To What Extent Does Accounting for Explicit and Implicit Bias Reveal the Presence of Aversive Stigmatizers?

Results suggest that a substantial proportion of respondents are potential aversive stigmatizers—actors who have nuanced attitude structures not captured by explicit measures alone. As Figure 1 demonstrates, the majority of respondents (55.6 percent) appear to be non-stigmatizers when focusing on explicit attitudes alone. When explicit and implicit measures of perceived dangerousness are considered together, however, findings suggest that non-stigmatizers and aversive stigmatizers may be conflated as a monolithic group despite differences in implicit bias. As Figure 1 shows, all three stigmatizer types are potentially present: non-stigmatizers (26.1 percent), aversive stigmatizers (29.5 percent), and explicit stigmatizers (44.4 percent).

Are Aversive Stigmatizers Potentially Socially Patterned?

Findings suggest that the endorsement of stigma is potentially socially patterned. As Table 3 shows, a lower percentage of women (p < .001), liberals (p < .001), sexual minorities (p = .001), and people with poor mental health (p = .005) explicitly perceive people with mental illness as dangerous compared to their male, non-liberal, straight, and mentally healthy counterparts. However, the social patterning of stigma differs when implicit measures are investigated instead. A higher percentage of sexual minorities (p < .05) and people with poor mental health (p = .008) exhibit implicit bias compared to their straight and mentally healthy counterparts. Two sample t-tests produced similar patterns: women, liberals, people with poor mental health, and sexual minorities exhibit lower mean explicit stigma but higher mean implicit bias (see Appendix A).

Table 4 shows the social patterning of three types of stigmatizers when both explicit and implicit measures are considered together. Findings suggest that the likelihood of being a non-stigmatizer, aversive stigmatizer, or explicit stigmatizer is socially patterned by gender (p < .001), political ideology (p < .001), mental health status (p = .004), and sexual minority status (p = .003). The majority of men (57.4 percent) and non-liberal respondents (55.6 percent) fall into the category of explicit stigmatizers, as well as almost half of straight respondents (49.8 percent) and mentally healthy respondents (48.9 percent). In contrast, stigmatizer types are more evenly distributed among other respondents, with aversive stigmatizers being the most prevalent: over one-third of women (37.1 percent), liberals (35.2 percent), sexual minorities (40.8 percent), and people with poor mental health (44.3 percent) are aversive stigmatizers.

As Figure 2 depicts, group differences in the proportion of respondents who are considered non-stigmatizers shrink when explicit and implicit measures are considered together. For example, there would be a 19 percent non-stigmatizer gap based on mental health status if we exclusively focused on explicit dangerousness but less than a 1 percent non-stigmatizer gap when considering explicit and implicit dangerousness together.

Multinomial logistic regression models further examine whether gender, political ideology, sexual minority status, and mental health status play a role in the likelihood of being a certain type of stigmatizer. Table 5 and Figures 3 to 5 include AMEs from multinomial logistic regressions of various respondent characteristics on stigmatizer type grouping. Regression analyses reveal similarities to the previous analyses, as well as a few differences. Compared to men, women have a higher probability of being an aversive stigmatizer (AME = .14, p < .01) and a lower probability of being an explicit stigmatizer (AME = −.21, p < .001). Respondents who identify as politically liberal do not have a significantly lower probability of being an aversive stigmatizer compared to politically conservative or moderate respondents, but do have a higher probability of being a non-stigmatizer (AME = .11, p < .05) and a lower probability of being an explicit stigmatizer (AME = −.15, p < .01). Sexual minorities also have a lower probability of being an explicit stigmatizer (AME = −.16, p < .05) compared to straight respondents.

Finally, Table 5 and Figure 5 convey differences by mental health status. People with poor mental health who report experiencing frequent mental distress are no more likely than people with good mental health to be non-stigmatizers or explicit stigmatizers, but they do have a significantly higher probability of being an aversive stigmatizer (AME = .12, p < .05). People with and without mental health issues, for instance, have similar predicted probabilities of being non-stigmatizers (.23 vs. .28), but people with poor mental health have a significantly greater predicted probability of being an aversive stigmatizer (.40) compared to their mentally healthy counterparts (.26) (see Appendix D).

Discussion

This study measures both deliberate (i.e., explicit) and automatic (i.e., implicit) cognition when examining mental health stigma and in doing so reveals new insights regarding stigma formation processes. Namely, it highlights the presence of a potentially unique actor in the stigma process, the aversive stigmatizer. Almost one-third of respondents in the convenience sample fall into the category of aversive stigmatizers: individuals who deliberately reject negative mental health stereotypes but may still perpetuate prejudice and discrimination given that they exhibit non-deliberate biases. Those categorized as aversive stigmatizers may, for example, outwardly express disdain for mental health stigma and support anti-stigma efforts but may or may not engage in stigmatizing behaviors without consciously realizing it (e.g., employment or housing discrimination driven by implicit biases).

Considering explicit and implicit measures together also reveals potential nuances in the social patterning of public stigma. If these analyses exclusively focused on explicit measures, one may conclude that some social groups in this sample predominantly consist of non-stigmatizers given that they are less likely to be explicit stigmatizers. For example, most women in the sample would appear to be non-stigmatizers based on the explicit measures alone, yet women in the sample are significantly more likely to be aversive stigmatizers when implicit measures are also considered. Women may be more deliberate in their acceptance of people with mental health issues but may still contribute to prejudice and discrimination processes in less deliberate ways.

Moreover, this approach also contributes to our understanding of self-stigma and supports the theoretical propositions of important stigma theories, such as modified labeling theory. While public stigma refers to the negative attitudes and beliefs of people in general, self-stigma reflects the internalization of these negative attitudes and beliefs by people contending with mental illness themselves (Boyd Ritsher et al. 2003; Pescosolido and Martin 2015; Watson et al. 2007). According to modified labeling theory, the devaluation and discrimination that people with mental illness face becomes personally relevant to individuals when they receive a mental illness label and can facilitate negative self-feelings (Link 1987). Being aware of negative stereotypes and the potential for mistreatment can then change interpersonal expectations and result in defensive coping mechanisms, such as withdrawing from social interaction to avoid stigma, that then contribute to negative outcomes (Link et al. 1989). In response to public stigma, some individuals with mental illness internalize and espouse negative cultural beliefs (i.e., self-stigma), while others explicitly reject them and engage in resistance efforts (i.e., deflection, challenging stereotypes, activism) that weaken the impact of stigma on well-being (Thoits 2011; Thoits and Link 2016). These findings further highlight how modified labeling theory and self-stigma may operate: people with poor mental health are significantly more likely to be aversive stigmatizers who internalize negative cultural stereotypes about their group even if they explicitly reject such stereotypes. In fact, results suggest that people with and without poor mental health have similar likelihoods of being non-stigmatizers once implicit bias is taken into account. Given that self-stigma can influence behaviors and negatively impact self-esteem and self-efficacy (Watson et al. 2007), the social distribution of aversive stigmatizers among people with poor mental health is consequential.

Why are stigmatizer types socially patterned this way in the sample? This study cannot definitively answer why there appears to be an uneven distribution of explicit stigmatizers, aversive stigmatizers, and non-stigmatizers among certain social groups in this convenience sample. It could be the case, for example, that certain groups are more susceptible to social desirability biases and are less likely to accurately self-report explicit stigma. It could also be the case that belonging to different social groups elicits different social experiences and cultural expectations that shape deliberate and automatic cognition processes in divergent ways. For example, women and people with mental health issues are more likely to know someone with a mental illness (Perry et al. 2022) and some work focusing on mental health professionals suggests that interpersonal contact may impact explicit and implicit biases differently (Kopera et al. 2015). Interpersonal contact is just one of many potential mechanisms that could shape the association between social groups and stigmatizer types. Future studies should explore these potential mechanisms.

Taken together, these findings support existing research suggesting that implicit biases can be shaped by the social environment (Ravary, Baldwin, and Bartz 2019; Sawyer and Gampa 2018) and can inform sociological theories (Miles et al. 2019). While sociologists may be receptive to culture and cognition scholarship, it has not been fully integrated into the sociology of mental health. For example, key culture and cognition article citations have clustered among three categories of sociology journals (i.e., generalist, culture, and theory journals) and appeared less frequently in other types of sociology journals (Dubash and Brett 2023). In the context of the sociology of mental health scholarship, including automatic cognition in theoretical propositions and incorporating cognitive measures in methodological approaches can further our understanding of mental health stigma and convey a more nuanced conceptualization of stigmatizing actors. Considering both deliberate and automatic cognition reveals three types of actors, two of whom may theoretically contribute to mental health stigma, albeit to varying degrees.

From a policy standpoint, measuring both deliberate and automatic cognition when testing the efficacy of various stigma reduction strategies could prove beneficial given that explicit and implicit measures combined best predict behavior (Greenwald et al. 2009). For example, do anti-stigma campaigns containing inspirational information messages that reduce negative explicit attitudes (e.g., Kroska and Harkness 2021) also reduce implicit biases? Is interpersonal contact as a stigma reduction mechanism (e.g., Felix and Lynn 2022; Perry et al. 2022) effective at tackling both explicit and implicit biases? If certain stigma reduction strategies reduce both explicit endorsement of stigma and implicit biases, they may be even more beneficial for tackling mental health stigma than previously believed. Alternatively, if certain anti-stigma policies or campaigns either increase (or have no effect on) implicit biases, they may unintentionally yield aversive stigmatizers as opposed to non-stigmatizers: individuals who may still engage in stigmatizing behaviors. To be clear, this is not to suggest that if mechanisms only alleviate explicit stigma, they cannot play integral roles in stigma reduction. In theory, converting explicit stigmatizers to aversive stigmatizers could be beneficial; it would just be arguably less beneficial than converting explicit stigmatizers to non-stigmatizers.

While this study is an important first step in applying an aversive stigmatizer framework to the sociology of mental health, it is not without limitations. It is important to note that this study implements an IAT to measure implicit bias. While several studies highlight the validity of IATs (e.g., Gosling et al. 2004; Houben and Wiers 2008) and there are many scholars that defend the implementation of the IAT (e.g., Cunningham et al. 2001; Greenwald et al. 2009; Nosek et al. 2005; Rudman 2008), their validity has been debated. For example, Schimmack (2021) provided a critique of the IAT, which was subsequently countered in a response by Kurdi, Ratliff, and Cunningham (2021). Related to the current study, Schimmack’s (2021) critique states that “there is also good evidence that the IAT can reveal group differences in associations” (p. 397). Some scholars also argue that IATs may be capturing a respondent’s awareness of cultural beliefs or greater familiarity with particular groups (Blanton and Jaccard 2008; Kinoshita and Peek-O’Leary 2005). Importantly though, if these critiques are warranted and if IATs are capturing stereotype awareness rather than stereotype endorsement as some scholars suggest, these findings could still contribute to our understanding of important stigma theories, such as modified labeling theory. If the IAT is in fact capturing internalized familiarity with cultural beliefs rather than implicit biases, results would demonstrate that people with poor mental health are more acutely aware of negative beliefs and these negative beliefs are more cognitively salient. According to modified labeling theory, awareness of stereotypes has real consequences in terms of influencing behaviors (Kroska and Harkness 2011). Nonetheless, future studies should validate these findings with other implicit measures moving forward (see Miles et al. 2019).

Future studies should also address other limitations of the current study. First, this study uses a small, non-representative online sample. Thus, the population prevalence of each stigmatizer type cannot be stated with confidence using this sample and should be determined with a larger, randomized sample. Furthermore, while there are some benefits to web-based data collection approaches, such as reducing social desirability biases (Duffy et al. 2005; Tourangeau and Yan 2007), there are also clear methodological limitations. Given that online samples are often disproportionately White, young, and highly educated compared to the U.S. population (see Levay, Freese, and Druckman 2016), future studies should collect data using nationally representative samples that are more generalizable. Second, this study does not include all potentially relevant measures and future studies should attend to other variables that could shed light on important stigma processes. For example, a limited number of respondent characteristics were gathered and future studies should examine if stigmatizer types are socially patterned across other potentially important characteristics, such as education level or social class.

This study also focuses on a commonly examined construct in stigma research: perceived dangerousness. While perceived dangerousness is often explored in stigma research, other relevant stigma constructs should also be explored. Furthermore, the explicit perceived dangerousness measure used in this study focuses on mental illness more generally. Future studies could build upon this research by implementing vignette survey experiments to also examine if illness type (i.e., depression vs schizophrenia) influences the likelihood of endorsing stigmatizing attitudes toward individuals displaying mental health symptoms. Vignette approaches are also effective strategies for eliciting data that can be used to examine public knowledge and attitudes to sensitive topics, like mental illness, and minimizing social desirability bias (i.e., employing a vignette strategy would be even more effective at distinguishing between the three stigmatizer types than the current study). Finally, the goal of this study is to introduce the concept of an aversive stigmatizer, and while it is a first step, it does not address another potential type of stigmatizer: individuals who may endorse explicit stigmatizing attitudes but simultaneously hold no implicit bias toward those with mental illness. While it is outside of the scope of this article to theorize about these potential social actors, future studies should consider if and why deliberate and automatic processes would diverge in this direction.

Finally, this study does not include behavioral measures to assess whether or not aversive stigmatizers are more or less likely to engage in discriminatory behaviors toward people with mental illness (e.g., employment discrimination, social avoidance) compared to non-stigmatizers and explicit stigmatizers. Given that aversive stigmatizers intentionally reject negative stereotypes, and given that more respondents with poor mental health fall into this category, it could be the case that they engage in public or self-stigma frequently, but it could also be the case that they actively monitor their own behaviors and reflect on any inadvertent discriminatory behaviors to avoid engaging in similar behaviors moving forward. Future studies should incorporate behavioral measures alongside explicit and implicit measures in order to more comprehensively understand these social psychological processes.

Conclusion

In sum, this study integrates two methodological approaches (e.g., traditional survey measures and an IAT) to uncover the presence and potential social patterning of aversive stigmatizers: actors who explicitly reject negative stereotypes but continue to hold implicit bias toward those with mental illness. In doing so, it provides a sociological lens through which to investigate the role of automatic cognition in both public stigma and self-stigma and can contribute to stigma theories (e.g., modified labeling theory). These findings have implications for studying key mechanisms that influence how mental health–related stigmatizing beliefs are formed, sustained, or dismantled over time.

Footnotes

Appendix

Appendix D.

Predicted Probabilities of Being Different Stigmatizer Types.

Variables	Non-stigmatizer(n = 77)	Aversive stigmatizer(n = 87)	Explicit stigmatizer(n = 131)
Respondent characteristics
Age
Under 35	25.0	29.0	46.0
35+	26.5	29.8	43.7
Gender
Female	29.7	36.4	33.9
Male	22.5	21.7	55.7
Race
Racial minority	24.7	34.7	40.7
White	26.8	27.8	45.4
Political ideology
Liberal	32.7	31.8	35.5
Non-liberal	20.7	27.8	51.5
Sexual identity
Sexual minority	29.9	37.7	32.4
Straight	25.1	26.9	48.0
Mental health status
Poor mental health	23.1	40.2	36.6
Good mental health	27.7	26.2	46.1

Note. Results based on multinomial logistic regression models in Appendix C and Table 5.

Author’s Note

I am indebted to Sarah Harkness, Freda Lynn, Bianca Manago, and Brea Perry for their insight and comments on earlier drafts of this paper.

ORCID iD

Elizabeth Felix

Notes

References

Blanton

Hart

Jaccard

James

. 2008. “Unconscious Racism: A Concept in Pursuit of a Measure.” Annual Review of Sociology 34:277–97. doi:10.1146/annurev.soc.33.040406.131632.

Bostwick

Wendy B.

Boyd

Carol J.

Hughes

Tonda L.

McCabe

Sean Esteban

. 2010. “Dimensions of Sexual Orientation and the Prevalence of Mood and Anxiety Disorders in the United States.” American Journal of Public Health 100:468–75. doi:10.2105/AJPH.2008.152942.

Boyd Ritsher

Jennifer

Otilingam

Poorni G.

Grajales

Monica

. 2003. “Internalized Stigma of Mental Illness: Psychometric Properties of a New Measure.” Psychiatry Research 121:31–49. doi: 10.1016/j.psychres.2003.08.008.

Cochran

Susan D.

Sullivan

J.Greer

Mays

Vickie M.

2003. “Prevalence of Mental Disorders, Psychological Distress, and Mental Health Services Use among Lesbian, Gay, and Bisexual Adults in the United States.” Journal of Consulting and Clinical Psychology 71:53–61.

Cunningham

William A.

Preacher

Kristopher J.

Banaji

Mahzarin R.

2001. “Implicit Attitude Measures: Consistency, Stability, and Convergent Validity.” Psychological Science 12:163–70. doi: 10.1111/1467-9280.00328.

Dasgupta

Nilanjana

. 2013. “Chapter Five Implicit Attitudes and Beliefs Adapt to Situations: A Decade of Research on the Malleability of Implicit Prejudice, Stereotypes, and the Self-concept.” Pp. 233–79 in Advances in Experimental Social Psychology. Vol. 47, edited by Devine

Plant

New York: Academic Press.

DeLuca

Joseph S.

Vaccaro

John

Seda

Jenna

Yanos

Philip T.

2018. “Political Attitudes as Predictors of the Multiple Dimensions of Mental Health Stigma.” International Journal of Social Psychiatry 64:459–69. doi:10.1177/0020764018776335.

Desmarais

Sarah L.

Van Dorn

Richard A.

Johnson

Kiersten L.

Grimm

Kevin J.

Douglas

Kevin S.

Swartz

Marvin S.

2014. “Community Violence Perpetration and Victimization Among Adults With Mental Illnesses.” American Journal of Public Health 104:2342–49. doi: 10.2105/AJPH.2013.301680.

Dovidio

John F.

Gaertner

Samuel L.

2004. “Aversive Racism.” Pp. 1–52 in Advances in Experimental Social Psychology, in vol.36, edited by Zanna

San Diego, CA: Elsevier Academic Press.

10.

Dovidio

John F.

Kawakami

Kerry

Beach

Kelly R.

2003. “Implicit and Explicit Attitudes: Examination of the Relationship between Measures of Intergroup Bias.” Pp. 175–97 in Blackwell Handbook of Social Psychology: Intergroup Processes, edited by Brown

Gaertner

Oxford, UK: Blackwell Publishing.

11.

Dubash

Soli

Brett

Gordon

. 2023. “The Diffusion of Culture and Cognition Within and Beyond Sociology, 1997–2021.” Sociological Forum 38:575–84. doi:10.1111/socf.12894.

12.

Duffy

Bobby

Smith

Kate

Terhanian

George

Bremer

John

. 2005. “Comparing Data from Online and Face-to Surveys.” International Journal of Market Research 47:615–39. doi:10.1177/147078530504700602.

13.

Felix

Elizabeth

Lynn

Freda

. 2022. “Mental Health Stigma and Social Contact Revisited: The Role of Network Closeness and Negativity.” Society and Mental Health 12:49–63. doi:10.1177/21568693211043156.

14.

Gaertner

Samuel L.

McLaughlin

John P.

1983. “Racial Stereotypes: Associations and Ascriptions of Positive and Negative Characteristics.” Social Psychology Quarterly 46:23–30. doi:10.2307/3033657.

15.

Gosling

Samuel D.

Vazire

Simine

Srivastava

Sanjay

John

Oliver P.

2004. “Should We Trust Web-Based Studies? A Comparative Analysis of Six Preconceptions About Internet Questionnaires.” American Psychologist 59:93–104. doi: 10.1037/0003-066X.59.2.93.

16.

Greenwald

Anthony G.

Banaji

Mahzarin R.

1995. “Implicit Social Cognition: Attitudes, Self-esteem, and Stereotypes.” Psychological Review 102:4–27. doi:10.1037/0033-295X.102.1.4.

17.

Greenwald

Anthony G.

Nosek

Brian A.

Banaji

Mahzarin R.

2003. “Understanding and Using the Implicit Association Test: I. An Improved Scoring Algorithm.” Journal of Personality and Social Psychology 85:197–216. doi:10.1037/0022-3514.85.2.197.

18.

Greenwald

Anthony G.T.

Poehlman

Andrew

Uhlmann

Eric Luis

Banaji

Mahzarin R.

2009. Understanding and Using the Implicit Association Test: III. Meta-analysis of Predictive Validity. Washington, DC: American Psychological Association.

19.

Hiday

Virginia Aldige

. 1995. “The Social Context of Mental Illness and Violence.” Journal of Health and Social Behavior 36:122–37. doi:10.2307/2137220.

20.

Houben

Katrijn

Wiers

Reinout W.

2008. “Measuring Implicit Alcohol Associations via the Internet: Validation of Web-based Implicit Association Tests.” Behavior Research Methods 40:1134–43. doi:10.3758/BRM.40.4.1134.

21.

Jost

John T.

Banaji

Mahzarin R.

Nosek

Brian A.

2004. “A Decade of System Justification Theory: Accumulated Evidence of Conscious and Unconscious Bolstering of the Status Quo.” Political Psychology 25:881–919. doi:10.1111/j.1467-9221.2004.00402.x.

22.

Joy-Gaba

Jennifer A.

Nosek

Brian A.

2010. “The Surprisingly Limited Malleability of Implicit Racial Evaluations.” Social Psychology 41:137–46.

23.

Kim

Do-Yeong

. 2003. “Voluntary Controllability of the Implicit Association Test (IAT).” Social Psychology Quarterly 66:83–96. doi:10.2307/3090143.

24.

Kinoshita

Sachiko

Peek-O’Leary

Marie

. 2005. “Does the Compatibility Effect in the Race Implicit Association Test Reflect Familiarity or Affect?” Psychonomic Bulletin & Review 12:442–52. doi:10.3758/BF03193786.

25.

Kopera

Maciej

Suszek

Hubert

Bonar

Erin

Myszka

Maciej

Gmaj

Bartłomiej

Ilgen

Mark

Wojnar

Marcin

. 2015. “Evaluating Explicit and Implicit Stigma of Mental Illness in Mental Health Professionals and Medical Students.” Community Mental Health Journal 51:628–34. doi:10.1007/s10597-014-9796-6.

26.

Kroska

Amy

Harkness

Sarah K.

2011. “Coping with the Stigma of Mental Illness: Empirically-grounded Hypotheses from Computer Simulations.” Social Forces 89:1315–39. doi:10.1093/sf/89.4.1315.

27.

Kroska

Amy

Harkness

Sarah K.

2021. “Information vs. Inspiration: Evaluating the Effectiveness of Mental Illness Stigma-reduction Messages.” Social Science Research 96:102543. doi:10.1016/j.ssresearch.2021.102543.

28.

Kuppens

Toon

Spears

Russell

. 2014. “You Don’t Have to Be Well-educated to Be an Aversive Racist, but It Helps.” Social Science Research 45:211–23. doi:10.1016/j.ssresearch.2014.01.006.

29.

Kurdi

Benedek

Ratliff

Kate A.

Cunningham

William A.

2021. “Can the Implicit Association Test Serve as a Valid Measure of Automatic Cognition? A Response to Schimmack (2021).” Perspectives on Psychological Science 16:422–34. doi:10.1177/1745691620904080.

30.

Lamont

Michèle

Adler

Laura

Park

Bo Yun

Xiang

Xin

. 2017. “Bridging Cultural Sociology and Cognitive Psychology in Three Contemporary Research Programmes.” Nature Human Behaviour 1:866–72. doi:10.1038/s41562-017-0242-y.

31.

Leschziner

Vanina

Brett

Gordon

. 2021. “Have Schemas Been Good To Think With?” Sociological Forum 36: 1207–28. doi:10.1111/socf.12767.

32.

Levay

Kevin E.

Freese

Jeremy

Druckman

James N.

2016. “The Demographic and Political Composition of Mechanical Turk Samples.” SAGE Open 6:2158244016636433. doi:10.1177/2158244016636433.

33.

Link

Bruce G.

1987. “Understanding Labeling Effects in the Area of Mental Disorders: An Assessment of the Effects of Expectations of Rejection.” American Sociological Review 52:96–112. doi:10.2307/2095395.

34.

Link

Bruce G.

Cullen

Francis T.

Struening

Elmer

Shrout

Patrick E.

Dohrenwend

Bruce P.

1989. “A Modified Labeling Theory Approach to Mental Disorders: An Empirical Assessment.” American Sociological Review 54:400–23. doi:10.2307/2095613.

35.

Link

Bruce G.

Phelan

J. C.

Bresnahan

Stueve

Pescosolido

B. A.

1999. “Public Conceptions of Mental Illness: Labels, Causes, Dangerousness, and Social Distance.” American Journal of Public Health 89:1328–33. doi:10.2105/AJPH.89.9.1328.

36.

Link

Bruce G.

Yang

Lawrence H.

Phelan

Jo C.

Collins

Pamela Y.

2004. “Measuring Mental Illness Stigma.” Schizophrenia Bulletin 30:511–41. doi:10.1093/oxfordjournals.schbul.a007098.

37.

Lizardo

Omar

Mowry

Robert

Sepulvado

Brandon

Stoltz

Dustin S.

Taylor

Marshall A.

Ness

Justin Van

Wood

Michael

. 2016. “What Are Dual Process Models? Implications for Cultural Analysis in Sociology.” Sociological Theory 34:287–310. doi:10.1177/0735275116675900.

38.

Long

J. Scott

Freese

Jeremy

. 2014. Regression Models for Categorical Dependent Variables Using Stata. 3rd ed. College Station, TX: US: Stata Press.

39.

Manago

Bianca

Pescosolido

Bernice A.

Olafsdottir

Sigrun

. 2019. “Icelandic Inclusion, German Hesitation and American Fear: A Cross-cultural Comparison of Mental-health Stigma and the Media.” Scandinavian Journal of Public Health 47:90–98. doi:10.1177/1403494817750337.

40.

Martin

Jack K.

Pescosolido

Bernice A.

Tuch

Steven A.

2000. “Of Fear and Loathing: The Role of ‘Disturbing Behavior,’ Labels, and Causal Attributions in Shaping Public Attitudes toward People with Mental Illness.” Journal of Health and Social Behavior 41:208–23. doi: 10.2307/2676306.

41.

Martin

Jack K.

Pescosolido

Bernice A.

Olafsdottir

Sigrun

McLeod

Jane D.

2007. “The Construction of Fear: Americans’ Preferences for Social Distance from Children and Adolescents with Mental Health Problems.” Journal of Health and Social Behavior 48:50–67.

42.

Melamed

David

Munn

Christopher W.

Barry

Leanne

Montgomery

Bradley

Okuwobi

Oneya F.

2019. “Status Characteristics, Implicit Bias, and the Production of Racial Inequality.” American Sociological Review 84:1013–36. doi:10.1177/0003122419879101.

43.

Meyer

Ilan H.

2003. “Prejudice as Stress: Conceptual and Measurement Problems.” American Journal of Public Health 93:262–65.

44.

Miles

Andrew

Charron-Chénier

Raphaël

Schleifer

Cyrus

. 2019. “Measuring Automatic Cognition: Advancing Dual-process Research in Sociology.” American Sociological Review 84:308–33. doi:10.1177/0003122419832497.

45.

Mize

Trenton D.

Doan

Long

J. Scott

. 2019. “A General Framework for Comparing Predictions and Marginal Effects across Models.” Sociological Methodology 49:152–89. doi:10.1177/0081175019852763.

46.

Nail

Paul R.

Harton

Helen C.

Decker

Brian P.

2003. “Political Orientation and Modern Versus Aversive Racism: Tests of Dovidio and Gaertner’s (1998) Integrated Model.” Journal of Personality and Social Psychology 84:754–70. doi:10.1037/0022-3514.84.4.754.

47.

Nosek

Brian A.

Greenwald

Anthony G.

Banaji

Mahzarin R.

2005. “Understanding and Using the Implicit Association Test: II. Method Variables and Construct Validity.” Personality and Social Psychology Bulletin 31:166–80. doi:10.1177/0146167204271418.

48.

Pager

Devah

Quillian

Lincoln

. 2005. “Walking the Talk? What Employers Say Versus What They Do.” American Sociological Review 70:355–80.

49.

Peer

Eyal

Brandimarte

Laura

Samat

Sonam

Acquisti

Alessandro

. 2017. “Beyond the Turk: Alternative Platforms for Crowdsourcing Behavioral Research.” Journal of Experimental Social Psychology 70:153–63. doi: 10.1016/j.jesp.2017.01.006.

50.

Perry

Brea L.

Felix

Elizabeth

Bolton

Megan

Pullen

Erin L.

Pescosolido

Bernice A.

2022. “Public Stigma and Personal Networks: Confronting the Limitations of Unidimensional Measures of Social Contact.” Journal of Health and Social Behavior 63:428–45. doi:10.1177/00221465211072311.

51.

Pescosolido

Bernice A.

2013. “The Public Stigma of Mental Illness: What Do We Think; What Do We Know; What Can We Prove?” Journal of Health and Social Behavior 54:1–21. doi: http://dx.doi.org/10.1177/0022146512471197.

52.

Pescosolido

Bernice A.

Manago

Bianca

Monahan

John

. 2019. “Evolving Public Views on the Likelihood of Violence from People with Mental Illness: Stigma and Its Consequences.” Health Affairs 38:1735–43. doi:10.1377/hlthaff.2019.00702.

53.

Pescosolido

Bernice A.

Martin

Jack K.

2015. “The Stigma Complex.” Annual Review of Sociology 41:87–116. doi:10.1146/annurev-soc-071312-145702.

54.

Pescosolido

Bernice A.

Martin

Jack K.

Lang

Annie

Olafsdottir

Sigrun

. 2008. “Rethinking Theoretical Approaches to Stigma: A Framework Integrating Normative Influences on Stigma (FINIS).” Social Science & Medicine 67:431–40. doi:10.1016/j.socscimed.2008.03.018.

55.

Phelan

Jo C.

Lucas

Jeffrey W.

Teachman

Bethany

Braverman

Benjamin H.

Namaky

Nauder

Greenberg

Mollie

. 2019. “A Comparison of Status and Stigma Processes: Explicit and Implicit Appraisals of ‘Mentally Ill People’ and ‘Uneducated People.’” Stigma and Health 4:213–24. doi:10.1037/sah0000106.

56.

Ravary

Amanda

Baldwin

Mark W.

Bartz

Jennifer A.

2019. “Shaping the Body Politic: Mass Media Fat-shaming Affects Implicit Anti-fat Attitudes.” Personality and Social Psychology Bulletin 45:1580–89. doi:10.1177/0146167219838550.

57.

Rudman

Laurie A.

2008. “The Validity of the Implicit Association Test Is a Scientific Certainty.” Industrial and Organizational Psychology 1:426–29. doi: 10.1111/j.1754-9434.2008.00081.x.

58.

Sawyer

Jeremy

Gampa

Anup

. 2018. “Implicit and Explicit Racial Attitudes Changed During Black Lives Matter.” Personality and Social Psychology Bulletin 44:1039–59. doi:10.1177/0146167218757454.

59.

Schaap

Julian

Waal

Jeroen van der

Koster

Willem de

. 2019. “Improving Empirical Scrutiny of the Habitus: A Plea for Incorporating Implicit Association Tests in Sociological Research.” Sociology 53:967–76. doi:10.1177/0038038519846417.

60.

Schimmack

Ulrich

. 2021. “The Implicit Association Test: A Method in Search of a Construct.” Perspectives on Psychological Science 16:396–414. doi:10.1177/1745691619863798.

61.

Singleton

Royce

Straits

Bruce

. 1999. Approaches to Social Research. Oxford, England: Oxford University Press.

62.

Skeem

Jennifer

Mulvey

Edward

. 2020. “What Role Does Serious Mental Illness Play in Mass Shootings, and How Should We Address It?” Criminology & Public Policy 19:85–108. doi:10.1111/1745-9133.12473.

63.

Spencer

Karen Lutfey

Grace

Matthew

. 2016. “Social Foundations of Health Care Inequality and Treatment Bias.” Annual Review of Sociology 42:101–20. doi:10.1146/annurev-soc-081715-074226.

64.

Swanson

Jeffrey W.

Holzer

Charles E.

Ganju

Vijay K.

Jono

Robert Tsutomu

. 1990. “Violence and Psychiatric Disorder in the Community: Evidence From the Epidemiologic Catchment Area Surveys.” Psychiatric Services 41:761–70. doi: 10.1176/ps.41.7.761.

65.

Swanson

Jeffrey W.

McGinty

E. Elizabeth

Fazel

Seena

Mays

Vickie M.

2015. “Mental Illness and Reduction of Gun Violence and Suicide: Bringing Epidemiologic Research to Policy.” Annals of Epidemiology 250:366–76. doi: 10.1016/j.annepidem.2014.03.004.

66.

Tarman

Christopher

Sears

David O.

2005. “The Conceptualization and Measurement of Symbolic Racism.” The Journal of Politics 67:731–61. doi:10.1111/j.1468-2508.2005.00337.x.

67.

Teachman

Bethany A.

Wilson

Joel G.

Komarovskaya

Irina

. 2006. “Implicit and Explicit Stigma of Mental Illness in Diagnosed and Healthy Samples.” Journal of Social and Clinical Psychology 25:75–95.

68.

Thoits

Peggy A.

2011. “Resisting the Stigma of Mental Illness.” Social Psychology Quarterly 74:6–28. doi:10.1177/0190272511398019.

69.

Thoits

Peggy A.

Link

Bruce G.

2016. “Stigma Resistance and Well-being among People in Treatment for Psychosis.” Society and Mental Health 6:1–20. doi:10.1177/2156869315591367.

70.

Tourangeau

Roger

Yan

Ting

. 2007. “Sensitive Questions in Surveys.” Psychological Bulletin 133:859.

71.

Vaisey

Stephen

. 2009. “Motivation and Justification: A Dual-process Model of Culture in Action.” American Journal of Sociology 114:1675–715. doi:10.1086/597179.

72.

Watson

Amy C.

Corrigan

Patrick W.

Angell

Beth

. 2005. “What Motivates Public Support for Legally Mandated Mental Health Treatment?” Social Work Research 29:87–94. doi:10.1093/swr/29.2.87.

73.

Watson

Amy C.

Corrigan

Patrick

Larson

Jonathon E.

Sells

Molly

. 2007. “Self-Stigma in People With Mental Illness.” Schizophrenia Bulletin 33:1312–18. doi: 10.1093/schbul/sbl076.