Patterns of Perceived Hostility and Identity Concealment among Self-Identified Atheists

Abstract

Researchers have produced important findings regarding the types of stigma associated with nonreligion, particularly atheism. However, while numerous studies analyze who is more or less likely to identify as an atheist given that stigma, less is known about how self-identified atheists manage the stigma associated with their identity. This study uses new survey data from a nationally representative sample of US adults, with an oversample of individuals identifying as atheists, to examine the predictors of and connections between atheists’ perceptions of hostility toward their identities and whether they conceal those identities. Contrary to our expectations, we find no association between atheists’ perceived hostility toward their identity and concealment of that identity. We do find, however, that atheists in some social locations report higher levels of identity concealment, particularly those who identify as women, those who identify as Republican, those who live in the South, and those who were raised in a religion or still attend religious services. Our findings suggest that atheists who feel like social or institutional outsiders are more likely to conceal their identity. Our findings also suggest that affirming an atheist identity may buffer some of the negative effects of atheist stigma. These findings have implications for how researchers understand the context-specific nature of religious discrimination, as well as implications for research on stigma management and the ways that the shifting religious and political landscape in the United States shapes the expression of atheist identities.

Introduction

The percentage of people who claim no religious affiliation in the United States has more than tripled in the last four decades. Growing from just 7 percent of the population in the 1980s to almost 25 percent in recent years, the nonreligious now make up one of the largest “religious” groups in the United States (Pew 2015). However, despite their growing numbers and their increased influence on US politics and culture, the nonreligious are still highly stigmatized in the United States, particularly those who are atheists. Atheists are one of the least liked and most distrusted minority groups in the United States (Edgell et al. 2016; Gervais, Shariff, and Norenzayan 2011), and atheists report discrimination on the basis of their atheism in a variety of contexts, including in their social networks, at school, and at work (Cragun et al. 2012; Hammer et al. 2012; Scheitle and Ecklund 2020).

Atheism is a highly politicized identity in America, and it has come to mean more than simply being “without theism.” Atheism was portrayed as anti-American and procommunist in the 1950s during the Cold War, it was demonized as immoral and anti-family in the 1970s and 80s during the rise of the Religious Right, and it became the basis for the “New Atheist” social and political movement in the mid-2000s (Jacoby 2006; Kettell 2014; Schmidt 2018). In this context, to take on the label of “atheist” is often a political choice that is about more than just expressing a lack of theistic beliefs, and many nonreligious people choose to take on less charged labels such as “agnostic,” “nonreligious,” or “secular” to avoid the heightened stigma that comes with being an atheist (Baker 2020; Edgell, Frost, and Stewart 2017; Scheitle, Corcoran, and Hudnall 2019).

While around 25 percent of the US population is nonreligious, only 4 percent identify as atheist (Pew 2019). Even among individuals who say they do not believe in a god, there is a significant gap between those who are nonbelieving, or what some call “atheistic,” and those who choose to identify as an “atheist” (Stewart 2016; Scheitle, Corcoran, and Hudnall 2019). As a result, much of the research on nonreligious stigma focuses on determining which nonreligious people are more or less likely to take on the atheist label (e.g., Edgell, Frost, and Stewart 2017; Scheitle, Corcoran, and Hudnall 2019).

There has been very little quantitative research, however, on the stigma management strategies of those who have already decided to take on the atheist label and whether different social locations might shape these strategies. That is, we have a solid understanding of who is more likely to identify as atheist, but what happens once someone takes on that label? What social factors determine whether atheists perceive stigma or hostility from others for being atheist? What factors determine whether atheists attempt to conceal their atheism from others because of that stigma? And what is the relationship between perceived stigma and concealment of identity among atheists?

Moreover, much of the foundational sociological research on atheist stigma and discrimination is based on surveys that were fielded over 10 years ago (e.g., Cragun et al. 2012; Edgell, Gerteis, and Hartmann 2006; Hammer et al. 2012), and there have been important changes in the religious and political makeup of the country since those studies were conducted. Since 2010, the percentage of nonreligious people in the United States has risen from 17 to 26 percent and the percentage of atheists has doubled, from 2 to 4 percent (Pew 2019). As the nonreligious make up an increasingly larger share of the population, it is possible that atheists feel more accepted and less stigmatized. The United States, however, has also become more politically polarized, and much of that polarization stems from perceptions of a disconnect between “religious conservatives” and “secular liberals” (Castle and Stepp 2021; Noy and O’Brien 2016). This could contribute to atheists feeling more stigmatized by religious conservatives, but it may also mean that atheists feel less stigmatized by religious and nonreligious liberals.

In short, it remains unclear the extent to which atheists continue to perceive stigma surrounding their atheism, whether they are concealing their atheism to avoid that stigma, and how social location might shape these experiences. To address these gaps, we draw on survey data collected in 2019 from a nationally representative sample of US adults with an oversample of atheists. We utilize structural equation models to investigate how social locations such as race, gender, political affiliation, and religious background shape both perceived hostility toward respondents’ atheist identity as well as respondents’ concealment of their atheist identity, and we discuss the implications of our findings for research on stigma management and the ways that the shifting religious and political landscape in the United States shapes the expression of atheist identities.

Atheists and Stigma Management

Many Americans consider religion to be the basis for morality and citizenship in American life and, as a result, the nonreligious are often seen as “moral outsiders” and excluded from conceptions of the “good American” (Edgell, Gerteis, and Hartmann 2006). This is particularly true for atheists, the most stigmatized group within the larger nonreligious population, and numerous studies have documented the stigma associated with atheism in America (Cook, Cottrell, and Webster 2015; Edgell et al. 2016; Gervais, Shariff, and Norenzayan 2011; Wright and Nichols 2014) and the discrimination that atheists experience as a result (Cragun et al. 2012; Hammer et al. 2012).

Stigmatized individuals often use strategies to avoid or lessen the impacts of their stigma. Many of these strategies fall into two broad categories: affirming the stigmatized identity more strongly to gain the benefits of strong in-group identification or hiding the identity in ways that reduce others’ knowledge of it (Branscombe et al. 2011; Ellemers, Spears, and Doosje 2002; Tajfel and Turner 1986). For individuals with a visible stigmatized trait or identity—such as a stigmatized skin color or physical disability—hiding that identity or “passing” is not typically an option. But for those with concealable identities—such as atheism—there is often an option to “pass” by hiding the stigmatized trait or disengaging with that trait altogether (Goffman 1963).

A number of factors determine how someone manages stigma, including how important their stigmatized trait or identity is to their sense of self (centrality), how often it matters for their daily social interactions (salience), and whether or not they have other stigmatized identities and the extent to which these stigmatized identities combine to create new categories of othering (intersectionality) (Ellemers, Spears, and Doosje 2002; Mackey et al. 2021; Quinn and Chaudoir 2009). Intersectionality is particularly important, as numerous studies have shown that people with more power and status typically have more resources for coping with stigma or avoiding it altogether (Link and Phelan 2001; Major and O’Brien 2005).

When it comes to concealable stigmatized identities such as atheism, there are costs and benefits to both identity affirmation and identity concealment. On the one hand, the benefit of concealment is typically that others are unaware of the stigmatized identity and thus are not able to directly discriminate based on that identity. Research also shows, however, that hiding a concealable identity can produce a sense of shame or guilt and lead to poor health and well-being (Ellemers, Spears, and Doosje 2002; Newheiser and Barreto 2014; Smart and Wegner 1999). On the other hand, while affirmation in public can lead to heightened stigma, research shows that strongly identifying with a stigmatized label or “coming out” as a member of a stigmatized group can help stigmatized individuals gain a sense of belonging and pride, especially if coming out takes place in a supportive group environment (Corrigan, Kosyluk, and Rusch 2013; Quinn and Chaudoir 2009; Tajfel and Turner 1986).

Perceived Hostility and Identity Concealment among Atheists

As atheism is a concealable identity, atheists often have the option to either affirm or conceal their identity depending on the context. For example, an atheist might attend religious services as a way of appearing religious in the eyes of their friends and family, or they might say they believe in a god in order to avoid being stigmatized for their nonbelief (Zuckerman 2012). In the United States, where being an atheist is more highly stigmatized than other nonreligious identities such as agnostic, spiritual but not religious, or just nonreligious more generally (Cragun et al. 2012; Edgell et al. 2016), one stigma management strategy that people who do not believe in a god use is to simply not take on the atheist label (Abbot and Mollen 2018; Mackey et al. 2021; Scheitle, Corcoran, and Hudnall 2019; Stewart 2016). Quantitative research shows that the gap between what Stewart (2016) calls the “atheist” and the “atheistic” in the United States is significant. For example, Cragun et al. (2012) found that while over 9 percent of their sample reported no belief in a god when asked about their beliefs, only 4 percent of individuals self-identified as atheist (see also Scheitle, Corcoran, and Hudnall 2019). Qualitative research also shows that many nonbelievers avoid the atheist label as a way to reduce stigma (Manning 2015; McClure 2017; Zuckerman 2012).

There are a few recent quantitative studies that investigate the relationship between stigma consciousness, “outness,” and discrimination among atheists. “Outness” is measured in a few ways in these studies, typically either by atheist identification (i.e., whether or not a nonbeliever identifies with the atheist label) and/or concealment (e.g., whether or not an atheist tells others about their atheism). These studies find that higher levels of anticipated stigma associate with less disclosure of, or association with, an atheist identity (Abbot and Mollen 2018; Mackey et al. 2021), that self-identified and “out” atheists report more discrimination than “closeted” atheists and other nonreligious people (Doane and Elliot 2015; Cragun et al. 2012; Hammer et al. 2012; Pfaff et al. 2021), and that perceived atheist stigma and concealing an atheist identity are both associated with negative mental and physical health outcomes (Abbott and Mollen 2018; Brewster et al. 2020; Doane and Elliot 2015). However, these studies also find that self-identified and out atheists often have higher in-group ties with other atheists, which results in more positive emotions surrounding atheist identity and more social support to help buffer the effects of discrimination (Abbott and Mollen 2018; Brewster et al. 2020; Doane and Elliot 2015). In fact, Brewster et al. (2020) find that atheists who report involvement in atheist groups reported higher levels of “outness” and lower levels of discrimination than atheists not involved in atheist groups, likely because those involved in atheist groups are more often surrounded by other atheists who are not as likely to discriminate against them. Taken together, however, research on the relationship between perceptions of hostility and identity concealment among atheists is inconsistent and requires further investigation.

As is the case with managing any kind of stigma, social context matters for how people manage the stigma of atheism and who chooses “outness” over concealment. For example, family background and parental status matter for perceived stigma and concealment among atheists. Cragun et al. (2012) find that atheists who have religious parents or have parents of two different religions are more likely to report discrimination in the family context than atheists with nonreligious parents (see also Hammer et al. 2012). Thus, atheists who were raised by religious parents may be more likely to perceive hostility and conceal their identity as a result. Parental status as an adult also matters. Atheists who have children often perceive higher levels of stigma because they are in charge of the moral upbringing of their children, something that many Americans’ believe requires religion (Ecklund and Lee 2011; Manning 2015). As a result, atheists are more likely to conceal their atheism when they become parents (McClure 2017) and more likely to return to religious institutions as a way of reducing that stigma and help in the socialization of their children (Ecklund and Lee 2011; Manning 2015). Thus, we predict that atheists who were raised by religious parents and/or are currently involved in a religious organization will perceive more hostility and be more likely to conceal their atheist identity.

Hypothesis 1: Past and current interaction with religion will be positively associated with atheists’ perceived hostility toward their atheist identity and their concealment of their atheist identity.

Social locations such as race and gender also shape atheists’ strategies for stigma management, as people with a minority status are more likely to be stigmatized and less likely to take on more stigma if they can help it (Link and Phelan 2001). Research shows that people of color are less likely to be both atheist and atheistic than white people in the United States (Baker and Smith 2015; Edgell, Frost, and Stewart 2017; Scheitle, Corcoran, and Hudnall 2019). For example, given the important historical role of religion in the Black American community, Black atheists face high levels of stigma and are thus less likely to be out as atheist and more likely to perceive hostility for their atheism than white Americans (Baker 2020; Hutchinson 2020; Pinn 2012; Swann 2020).

Gender is also a key factor in determining atheist stigma management strategies. Women in the United States are more likely to be discriminated against than men for being atheist, women are less likely to be both atheist and atheistic than men, and atheistic women are less likely to be out as atheists than are atheistic men (Edgell, Frost, and Stewart 2017; Scheitle, Corcoran, and Hudnall 2019; Stewart 2016). Given this prior research, we predict that atheists who hold a second minority status, such as identifying as female or as a person of color, will be more likely to perceive hostility for and conceal their atheist identity.

Hypothesis 2: Holding a second minority status will be positively associated with atheists’ perceived hostility toward their atheist identity and their concealment of their atheist identity.

Political affiliation may also influence perceived stigma and stigma management strategies among atheists. Atheists are more likely to be politically liberal and identify as Democrat than those who are religious (Baker and Smith 2015; Schwadel 2020), and some argue that it has been the increasing association between political conservativism and religious conservativism in the United States that has driven the increase in religious disaffiliation over the past few decades (Hout and Fischer 2014). Edgell, Gerteis, and Hartmann (2006) and Edgell et al. (2016) also find that Republicans are less accepting of atheists, less likely to say that atheists share their visions of American society, and less like to say they would support their child marrying an atheist. Thus, atheists who are politically conservative may be more likely to perceive stigma for their atheism and conceal their atheism from others. Finally, geographic location also matters. A number of studies show that atheists in the United States are more likely to be discriminated against in the South than they are in other parts of the country because of the highly religious nature of Southern culture (Mackey et al. 2021; Scheitle and Corcoran 2018; Wallace, Wright, and Hyde 2014). Thus, we predict that atheists who are part of atheist-hostile social locations, like being a Republican or living in the South, will be more likely to perceive hostility for and conceal their atheist identity.

Hypothesis 3: Being a part of an atheist-hostile social location will be positively associated with atheists’ perceived hostility toward their atheist identity and their concealment of their atheist identity.

Data

The data for this study come from the 2019 Experiences with Religious Discrimination Study (ERDS). The ERDS project was supported by a grant from the National Science Foundation and aimed to measure the prevalence, predictors, and consequences of individuals’ experiences with interpersonal hostility, organizational and institutional discrimination, and criminal victimization. The ERDS project consisted of a survey of a nationally representative sample of US adults.

The sample for the survey comes from the Gallup Panel, which is a probability sample of about 80,000 US adults recruited through random digit dialing and address-based sampling methods. Panelists complete surveys either online or, if they lack access to the internet, through the mail. For the ERDS survey, Gallup selected 10,198 individuals to be invited to take the survey. This represented a random general population sample of 5,131 individuals and oversamples of individuals from key religious groups. Specifically, individuals who had indicated on prior panel surveys that they were Muslim, Jewish, Buddhist, Hindu, or atheist were oversampled. In all, 4,774 individuals who were invited to complete the survey did so. More details concerning the ERDS survey can be found in (Authors). Weights are utilized to account for the oversampling of groups and patterns of nonresponse so the estimates mirror the US adult population.

For the analysis presented here, we limit the data to 721 individuals who identified as atheist. This identity was provided in response to a question that asked, “Religiously, do you consider yourself to be Protestant, Catholic, Jewish, Muslim, Hindu, Buddhist, atheist, or something else? If more than one, mark the one that best describes you.” Respondents were provided 24 response options, including a write-in option for “something else, specify.” Responses for “no religion” and “agnostic” were also explicitly offered, so the sample considered in this analysis represents those individuals who identify as atheist even when potentially less stigmatized identities were offered.

Measures

We focus on two latent outcomes in this study: (1) perceived hostility toward one’s atheist identity and (2) concealment of one’s atheist identity. The former may be seen as an individual’s assessment of how those who do not share an identity might react to one’s identity, while the latter might be seen as a strategy for avoiding or managing those reactions or the individual’s internal feelings about those reactions. Each of these latent outcomes is measured by three indicators. We first describe these indicators below, and then present a confirmatory factor analysis to evaluate the observed measures and their relationship to the latent factors (table 3).

Outcome: Perceived hostility toward atheist identity

The ERDS survey contains three items that assess the hostility individuals perceive from others due to their religion. These items began with the instruction, “The following questions ask about how other people react to your religion.” An additional note was provided for atheist and other individuals who did not identify with a religion (emphasis on instrument): “Note: If you identify as an atheist, agnostic, or otherwise do not have a religion, please respond to these questions to tell us how other people react to these  identities or how people react to you not having a religion.”

Individuals were then asked, “Thinking about experiences you have had in the past year, how often have you experienced the following?” Three statements were provided: (1) I sense hostility from others because of my religion, (2) I felt disrespected because of my religion, and (3) People assumed things about me because of my religion. Possible responses to these items were: never, rarely, sometimes, frequently, and always. These items were adapted from Allen et al.’s Religious Discrimination Scale (2020).

Outcome: Concealment of atheist identity

The ERDS survey contained three items meant to assess identity concealment. These items began with the instruction, “This next group of questions asks about how you express, or do not express, your religious identity.” An additional note was provided for atheist and other individuals who did not identify with a religion (emphasis on instrument): “Note: If you identify as an atheist, agnostic, or otherwise do not have a religion, please respond to these questions to tell us how you express these identities or   how you express not having a religion.”

Individuals were then asked to “please indicate your level of agreement with the following statements.” The three provided statements were: (1) I refrain from talking about my religious identity with people outside of my religion, (2) I conceal or camouflage signs of my religious identity when in public, and (3) Outside of my closest friends and family, no one knows how important my religious identity is to me. These items were adapted from Madera, King, and Hebl (2012). Offered responses were strongly agree, agree, neither agree nor disagree, disagree, or strongly disagree. For the analysis here, we reverse coded responses so that higher values represent agreement that the individual conceals their identity.

While these measures do have some limitations, in that the wording of these questions center “religious” identities rather than nonreligious identities, the fact that this survey was distributed both online and via mail meant that the question wordings had to be as general as possible because response options cannot be changed to be specific to each respondent’s identity in mail-based surveys.

Predictors: Past and current interactions with religion

We measure atheists’ past and current interaction with religion in two ways. First, we include a group of indicators representing the religious background of our sample. The ERDS instrument included a question asking, “At age 16, what was your religion? Were you Protestant, Catholic, Jewish, Muslim, Hindu, Buddhist, atheist, or something else? If more than one, mark the one that best describes you:” The offered responses mirrored those for the item asking about respondents’ current religion. From this question, we created four categories for atheist individuals’ religious origins: (1) raised in a religion, (2) raised no religion, (3) raised agnostic, and (4) raised atheist.

Second, we include a measure representing the frequency of individuals’ current religious service attendance. The ERDS survey asked, “How often do you attend religious services?” Nine responses were offered, ranging from never to more than once a week. Although 75.7 percent of atheist individuals report never attending religious services, this means that about one in four attends at least occasionally.

Predictors: Second minority status

We measure atheists’ additional minority statuses in two ways. First, we include indicators representing respondents’ gender. This comes from a question asking, “What is your gender?” The responses of man, woman, nonbinary, or other (specify) were offered. Given the small number of cases (n = 3) in the last response among atheist individuals, we combine this into the nonbinary group to represent a broader “other gender identity” category.

Next, we include a group of indicators representing respondents’ racial or ethnic identity. The ERDS instrument asked (emphasis on instrument), “Which of the following best represents your race or ethnicity? You may mark more than   one.” Offered responses were: (1) White, Caucasian, European, (2) Black, African, Caribbean, (3) Hispanic, Latino, (4) Middle Eastern, Central Asian, North African, Arab, (5) East Asian (Chinese, Japanese, Korean, Taiwanese, etc.), (6) South Asian (Indian, Pakistani, Bangladeshi, etc.), (7) Native American, American Indian, (8) Pacific Islander, and (9) Other, specify. Some of these categories have relatively small numbers of cases, particularly among atheists. Given this, we recoded the responses into the following four categories: (1) White, (2) Hispanic, (3) Black, and (4) Other or multiple races or ethnicities.

Predictors: Atheist-hostile social locations

We include two measures for our concept of atheist hostile social locations. The first represents the political identity of the individual. We include a group of indicators for whether the individual identifies as Republican, Democrat, Independent, or identifies with another political party. Second, we include a group of indicators representing the region of the United States in which the individual resides.

Controls

In addition to the focal predictors just described, we include several other control measures in our analyses. We include a continuous measure for age of respondent. A second control indicates level of education measured on a seven-point scale ranging from (1) less than a high school diploma to (8) postgraduate or professional degree. Two other control measures represent whether the respondent is currently married (1) or not (0), and whether the respondent has any children (1) or not (0).

Figure 1 provides a conceptual summary of our measures in relation to the previously stated hypotheses.

Figure 1

Conceptual model

Analytical strategy

We utilize structural equation modeling (SEM) for this study or, more specifically, a Multiple Indicator Multiple Cause model. This type of analysis has a few particular advantages for our purposes. First, it allows us to directly model the outcomes of interest as latent variables with multiple observed indicators, which strengthens the reliability of the outcomes by accounting for measurement error across the indicators. Second, SEM allows us to simultaneously model both outcomes—perceived hostility and identity concealment—while accounting for any potential relationship between the two. We do this by allowing the error terms for the perceived hostility and identity concealment latent outcomes to covary.¹ Finally, within our SEM models, we utilize a maximum likelihood with missing values (MLMV) method, which has some advantages relative to other ways of addressing missing data (Allison 2009; 2012). All analyses are conducted using Stata SE 15.1 software using its structural equation modeling command (sem) and its complex survey command (svy) to account for the data’s weighting.²

We examine four SEM models, the first three representing reduced versions of the fourth and final model. The first three models focus individually on our focal groups of predictors: past and current interactions with religion, second minority status, and atheist-hostile social locations. The final model considers all of these groupings together alongside the control measures. We allow the control measures to predict both of the latent factors. Traditional indicators of model fit are typically considered inappropriate or inapplicable when utilizing complex survey data (Bollen, Tueller, and Oberski 2013; Stata Corp LLC 2021; Williams 2021). However, our final model does indicate good model fit when estimated without taking into account the data weights (e.g., model chi-square = 125.27, degrees of freedom = 84, p < .01; comparative fit index = .98; Tucker-Lewis index = .97; root mean squared error of approximation = .02).

Results

Table 1 presents descriptive statistics for all the observed measures in this analysis. We see that 41.2 percent of atheists grew up in a religion, 6.4 percent say that they had no religion at age 16, 13.8 percent say that they were agnostic at age 16, and 32.7 percent report being an atheist at age 16. Four out of five atheists in the ERDS data identify as white, while 4.9 percent identify as Hispanic, 5.3 percent identify as Black, and 9.9 percent identify with another race or ethnicity or with multiple races and ethnicities. Atheists are also disproportionately men (61.2 percent) and largely identify as Democrat (55.2 percent) or as politically independent (33.6 percent).

Table 1

Descriptive Statistics

	Weighted Mean or Percentage	N
Perceived Hostility Towards Atheist Identity
I sensed hostility from others because of my religion.	2.3	720
I felt disrespected because of my religion.	2.2	720
People assumed things about me because of my religion.	2.6	720
Concealment of Atheist Identity
I refrain from talking about my religious identity with people outside of my religion.	2.7	720
I conceal or camouflage signs of my religious identity when in public.	2.5	719
Outside of my closest friends and family, no one knows how important my religious identity is to me.	2.7	719
Past and Current Interaction with Religion
Religious origins		720
Raised in a religion	41.2%
Raised “no religion”	6.4%
Raised agnostic	13.8%
Raised atheist	32.7%
Current religious service attendance	1.4	721
Second Minority Status
Race and ethnicity		709
White	79.9%
Hispanic	4.9%
Black	5.3%
Other or multiple	9.9%
Gender		706
Man	61.2%
Woman	35.0%
Other\Nonbinary	3.7%
Atheist Hostile Social Locations
Political party		692
Democrat	55.2%
Republican	5.2%
Independent	33.6%
Other	6.1%
Region of residence		712
Northeast	22.4%
Midwest	21.4%
South	28.8%
West	27.4%
Controls
Age	39.1	721
Education	4.4	720
Currently married	42.4%	705
Parent	24.9%	704

Note: 2019 Experiences with Religious Discrimination Survey.

Although our focus is on individuals who identify as atheist as opposed to those who identify with another nonreligious labels, we did assess means on the perceived hostility and identity concealment items across the other major nonreligious groups in our sample. Table 2 presents the overall means across the three perceived hostility items and the three identity concealment items for those who identified as “no religion,” agnostic, and atheist. We see that perceived hostility increases consistently as we move from no religion to agnostic to atheist. This is in line with what we would expect given past research highlighting variations in stigma between different nonreligious groups and the heightened stigma that comes with atheism. Looking at the concealment scores, though, we find that atheists’ mean concealment score is significantly lower than both the no religion group and the agnostic group. In sum, atheists appear to perceive more hostility but conceal their identity less than other nonreligious groups. This suggests that the dynamics among atheists may be unique and provides further justification for our focus on atheists in this analysis.³

Table 2

Mean Perceived Hostility and Identity Concealment Scale Scores across Nonreligious Groups

	No religion	Agnostic	Atheist
Mean Perceived Hostility Scale Score [95% confidence interval]	1.75b,c [1.65–1.85]	2.01a,c [1.91–2.11]	2.36a,b [2.26–2.45]
Mean Concealment Scale Score [95% confidence interval]	2.76 [2.67–2.85]	2.83 [2.73–2.94]	2.63a,b [2.54–2.72]
N	556	619	721

a

Significantly different from No religion group at p < .05.

b

Significantly different from Agnostic group at p < .05.

c

Significantly different from Atheist group at p < .05.

Table 3 presents a confirmatory factor analysis for our two latent outcomes.⁴ This is simply a measurement model and does not include any paths from our hypothesized predictors to our two latent factors. The three indicators of perceived hostility all show strong loadings on the latent concept, with all of the coefficients over .80. Similarly, the three indicators of identity concealment show good loading on the latent concept, with all the path coefficients over .70. In short, the observed measures of our two latent concepts appear to be reliable. A confirmatory factor analysis that does not take into account the survey weights and removes cases with missing data indicates an acceptable to good model fit (e.g., model chi-square = 38.66, degrees of freedom = 9, p < .01; comparative fit index = .98; Tucker-Lewis index = .97; root mean squared error of approximation = .07).

Table 3

Standardized Path Loadings for Observed Measures of Latent Concepts

	Latent Concept
Observed Measures	Perceived Hostility to Atheist Identity	Concealment of Atheist Identity
I sensed hostility from others because of my religion.	.89**	—
I felt disrespected because of my religion.	.88**	—
People assumed things about me because of my religion.	.85**	—
I refrain from talking about my religious identity with people outside of my religion.	—	.77**
I conceal or camouflage signs of my religious identity when in public.	—	.73**
Outside of my closest friends and family, no one knows how important my religious identity is to me.	—	.72**
Correlation Between Latent Concepts	.01

Note: 2019 Experiences with Religious Discrimination Survey; MLMV method; N = 721.

^‡p < .10.

^*p < .05.

^**p < .01.

At the bottom of table 3, we see the covariance between the two latent concepts. Somewhat surprisingly, we find practically no association between perceived hostility toward atheist identity and concealment of that atheist identity. As noted earlier, research and theory is somewhat ambivalent on the relationship between perceived hostility and concealment. We might have expected that individuals who perceived hostility toward their atheist identity would be more likely to conceal their identity as a way to mitigate or avoid experiencing such hostility. We might have also reasonably expected that atheists who are more open and “out” (i.e., not concealing their identity) might perceive more hostility, either because their openness leads them to encounter such hostility or because their openness is meant to be a counter to such perceived hostility. This measurement model, though, does not find support for either of these patterns.⁵

Table 4 presents our full SEM analysis. Model 1 includes the predictors measuring atheists’ past and current interaction with religion. We see that, relative to those raised atheist, the other religious origin groups do not significantly differ in their perception of hostility toward their atheist identity. Similarly, we do not find a significant association between atheists’ current religious service attendance and their perception of hostility toward their atheist identity. Hypothesis 1, then, does not seem to have initial support for this outcome.

Table 4

Structural Equation Models (Unstandardized Coefficients)

	Outcomes
	Perceived Hostility to Atheist Identity				Concealment of Atheist Identity
Predictors	Model 1	Model 2	Model 3	Model 4	Model 1	Model 2	Model 3	Model 4
Past and Current Interaction with Religion
Religious origins
Raised in a religion	.16	—	—	.16	.30**	—	—	.27*
Raised “no religion”	−.16	—	—	−.18	.35	—	—	.17
Raised agnostic	.30‡	—	—	.25	.53**	—	—	.45**
Raised atheist (ref.)	—	—	—	—	—	—	—	—
Current religious service attendance	.01	—	—	.02	.13*	—	—	.13*
Second Minority Status
Race and ethnicity
White (ref.)	—	—	—	—	—	—	—	—
Hispanic	—	.20	—	.20	—	−.09	—	−.13
Black	—	−.27	—	−.40	—	.19	—	.10
Other or multiple	—	−.13	—	−.11	—	−.17	—	−.17
Gender
Man (ref.)	—	—	—	—	—	—	—	—
Woman	—	.18‡	—	.25*	—	.16‡	—	.20*
Other\Nonbinary	—	−.14	—	.004	—	.05	—	.14
Atheist Hostile Social Locations
Political party
Democrat (ref.)	—	—	—	—	—	—	—	—
Republican	—	—	.05	.17	—	—	.69**	.81**
Independent	—	—	.24*	.30**	—	—	−.10	−.04
Other	—	—	−.08	−.02	—	—	.08	.14
Region of residence
Northeast	—	—	−.11	−.07	—	—	−.20	−.18
Midwest	—	—	.01	.02	—	—	−.06	−.09
South	—	—	.11	.15	—	—	.31*	.26*
West (ref.)	—	—	—	—	—	—	—	—
Controls
Age	−.01**	−.01**	−.01**	−.01**	−.001	−.002	−.002	−.003
Education	−.03	−.03	−.01	−.02	.02	.02	.03	.02
Currently married	.18‡	.18‡	.20*	.15	.20‡	.24*	.23*	.20*
Parent	.20‡	.20‡	.19‡	.18‡	−.18	−.18	−.16	−.21*
Error Covariance with Other Outcome	−.01	.001	−.002	−.03	−.01	.001	−.002	−.03
R-squared	.07	.07	.07	.11	.07	.03	.09	.14

Note: 2019 Experiences with Religious Discrimination Survey; MLMV method; N = 721.

^‡p < .10.

^*p < .05.

^**p < .01.

Notably, atheists who were raised in a religion or who raised agnostic are significantly more likely to say they conceal their identity relative to those individuals who have always been atheist. Similarly, the analysis shows that religious service attendance is a positive predictor of identity concealment among atheists. Overall, then, past and current interaction with religion does appear to be associated with concealment of atheist identity in the ways we expected in Hypothesis 1.

Model 2 focuses on our indicators of atheists’ gender and racial or ethnic identities. Looking first at the results for perceived hostility toward individuals’ atheist identity, we do not find any statistically significant differences across racial or ethnic groups relative to atheists who identify as white. Turning to gender differences, we do find some evidence that women might report greater perception of stigma related to their atheist identity relative to men, although this difference is of borderline statistical significance (p = .07). Hypothesis 2, then, receives some inconsistent and weak support in this initial model.

We find similar patterns when looking at the concealment of atheist identity outcome. While there are no statistically significant differences across the racial or ethnic groups relative to the white group, we again find a borderline statistically significant difference (p = .09) between men and women. Specifically, we find that atheist women report concealing their atheist identity more than atheist men. This equates to weak and mixed initial support for Hypothesis 2.

Model 3 examines our indicators of atheist-hostile social locations. Looking first at the political party measures, we find that Republican atheists do not significantly differ from Democratic atheists in their perception of hostility toward their atheist identity. This is counter to our expectation based on Hypothesis 3. We find that politically independent atheists do perceive significantly more hostility toward their atheist identity than atheist Democrats. Turning to the concealment of atheist identity outcome, we find that Republican atheists are significantly more likely to say that they conceal their atheist identity than Democratic atheists. This is what was expected based on Hypothesis 3.

This model also includes our indicators representing the region that these atheist individuals are residing in. Model 3 shows support for Hypotheses 3 in that we see, relative to atheists residing in the West, atheists residing in the South are significantly more likely to say that they conceal their atheist identity. However, we do not find any significant difference between atheists in the South and atheists in the West in their perception of hostility toward their atheist identity. Note that there is a borderline (p < .08) statistical difference between the South and the Northeast regions if the latter were to be considered the comparison group.

Having examined our focal predictors in isolation, Model 4 considers all of the predictors’ simultaneously. Many of our findings are the same as those found in the reduced models, with the exception of Model 2, which reveals the difference between men’s and women’s perception of hostility toward their atheist identity was of borderline statistical significance. In this full model, though, the difference is significant at the p < .05 level, with atheist women reporting more perceived hostility toward their identity than atheist men. A similar finding occurs with identity concealment outcome.

To put these findings into scale, we computed predicted scores on the perceived hostility and identity concealment latent measures. We found that the standard deviation on the hostility variable is .32 and the standard deviation on the concealment variable is .34. Based on this, we can see in Model 4 that atheists raised in a religion (B = .27) and atheists raised agnostic (B = .45) conceal their identities around a standard deviation more on the concealment variable relative to atheists who were raised as atheists. Similarly, atheists who identify as a Republican (B = .81) conceal their identities about two standard deviations more relative to atheists who identify as a Democrat. Atheists in the South (B = .26) conceal their identities a little less than one standard deviation relative to atheists residing in the West.

Although not our focal predictors, it is worth examining the control measures in Model 4 before discussing these findings in more detail. We see that age is associated with reduced perception of hostility among atheists. Education, on the other hand, has no significant association with perceived hostility toward individuals’ atheist identity. In earlier models, atheists who are currently married showed some greater perception of hostility compared with unmarried atheists, but in Model 4, this difference is not significant. There is a borderline significant positive association (p = .09) between whether an atheist is a parent and their perception of hostility.

Turning to the concealment outcome, we see that there is no significant association between age or education and atheists’ reported concealment of their identity. However, marital status and parental status both have borderline significant associations with concealment of atheist identity. Those atheists who are currently married appear to be more likely to conceal their atheist identity, while those atheists who are parents appear less likely to conceal their atheist identity.

Discussion

Atheism has long been a stigmatized identity in American society, and studies have shown that atheists report stigma and discrimination in a variety of contexts, including at home, at work, and at school (Cragun et al. 2012; Hammer et al. 2012). As a result, the atheistic often conceal their atheism or refrain from adopting the atheist label to avoid this stigma (Abbott and Mollen 2018; Brewster et al. 2020; Doane and Elliot 2015; Ecklund and Johnson 2021). However, the religious landscape in the United States is shifting, and the number of self-identified atheists has doubled in the past 10 years (Pew 2019). Alongside a growing population of other nonreligious people, including agnostics and the “spiritual but not religious,” the amount and type of stigma that atheists perceive may be shifting as well. And while past research has identified who among the nonreligious is more or less likely to identify as an atheist despite the associated stigma, we know less about the rates at which atheists themselves perceive stigma, whether or not they conceal their atheism from others, and how social locations like race, gender, and political affiliation shape these experiences.

In this paper, we address these gaps using 2019 survey data from a nationally representative sample of US adults, which included an over-sample of atheists, to examine the predictors of and connections between atheists’ perceptions of hostility toward their atheist identity and concealment of that identity. We chose to focus on atheists specifically because of the heightened stigma surrounding atheism in the United States and the lack of quantitative research on the social contexts of atheist stigma management specifically. Our initial analyses also revealed that atheists in our sample are the most likely nonreligious group to perceive hostility for their nonreligion but they are the least likely to conceal their identities. Thus, we set out to investigate whether or not this trend held across different social and geographic locations and to better understand the relationship between perceived hostility and identity concealment among atheists specifically.

In line with previous research on stigma management and atheist identities, we find that atheists who are already marginalized, such as women, and atheists who are part of atheist-hostile social locations and contexts, such as the Republican Party or the Southern United States, are more likely to conceal their atheism from others. Previous research has indicated that those who are part of potentially atheist-hostile social groups are less likely to become atheist. We find that those who are members of these groups are also the most likely to conceal their atheist identity when they choose to take on the label.

Furthermore, atheists who were raised in a religion, raised as agnostic, or who still attend religious services are more likely to conceal their identity than individuals who were raised as atheist or who do not attend religious services. This finding highlights the importance of religious upbringing in how atheists perceive stigma, but it also shows the importance of distinguishing between different nonreligious identities in our surveys. To be raised in an agnostic home appears to lead to more perceived atheist stigma than being raised in an affirmatively atheist home, potentially revealing the importance of social ties to stigma management. Interestingly, about twice as many of our atheist respondents were raised in an atheist home than were raised in an agnostic home. These dynamics deserve further investigation, especially as younger generations, who are more likely to be nonreligious, start having children and raising them without religion (see Baker and Smith 2015; Manning 2015).

However, some of our findings diverge from what previous research led us to expect. To start, respondents in social locations that we expected to find high rates of concealment and perceived hostility, including among those who are racial minorities and those who are parents, were not significantly more likely to perceive hostility or conceal their atheism. While previous research has found that people of color, particular Black Americans, are less likely to take on the atheist label than white Americans due to heightened perceived stigma (Baker 2020; Scheitle, Corcoran, and Hudnall 2019’ Pinn 2012; Swann 2020), we find no significant racial differences regarding perceptions of stigma or identity concealment among individuals in these groups who have already taken on that atheist label. While we do find some support for past research suggesting that atheist parents are more likely to perceive stigma due to expectations around moral parenting (Manning 2015; McClure 2017), our analysis finds that atheist parents are in fact less likely to conceal their atheism than atheists who are not parents. It is possible that parenting leads atheists to advocate for their children in contexts that may be more hostile toward their atheist identity, such as schools, which leads to less identity concealment. Finally, in contrast to previous studies that find perceived hostility leads to identity concealment among atheists (Abbot and Mollen 2018; Mackey et al. 2021), we find no association between perceived hostility toward individuals’ atheist identity and concealment of their atheist identity.

We argue that these unexpected findings are the result of our specific focus on atheists. They suggest that the shifting religious and political landscape in the United States has shaped the perception and management of stigma among American atheists. As more people disaffiliate from religion, it is possible that atheists are more likely to encounter other atheists in their daily lives and that the stigma of atheism is on the decline. Numerous atheist organizations, such as the American Atheists and the Freedom From Religion Foundation, have been espousing a popular form of identity politics in which they call on atheists to “come out of the closet” and help reduce atheist stigma by publicly affirming atheism (Anspach, Coe, and Thurlow 2007; Frost 2019; Kettell 2014). It is possible that these stigma reducing strategies are working, which may explain why we see so few significant differences in perceptions of hostility among atheists. While we are not suggesting that atheist stigma is now a thing of the past, it is possible that atheist stigma is reducing as a result of demographic and cultural shifts away from religion.

These atheist organizations may also make it easier for atheists to choose identity affirmation over concealment, even in the face of perceived hostility. Social psychological research suggests that strongly identifying with a stigmatized label and “coming out” as a member of a stigmatized group can help stigmatized individuals gain a sense of belonging and pride, which can buffer the negative effects of stigma and discrimination (Quinn and Chaudoir 2009; Corrigan, Kosyluk, and Rusch 2013). There are now a substantial number of social and political atheist organizations in the United States that have been created to affirm atheist identities and values (Cragun, Manning, and Fazzino 2017; García and Blankholm 2016), which means that atheists now have supportive spaces in which to affirm their atheist identities.

This may help explain why we do not find a significant relationship between perceived hostility and identity concealment among the atheists in our sample. Given that prior research has found that having close friends that are atheists is significantly associated with self-identifying as an atheist (Scheitle, Corcoran, and Hudnall 2019), and that having supportive social networks can reduce the negative impacts of discrimination for atheists (Brewster et al. 2020), this is a plausible explanation. It could be that some of the groups in which we expected to see high levels of perceived stigma and concealment but did not, such as people of color and parents, are more able to form social ties with other atheists once they come out as atheist. And once they form these relationships, they are perhaps less likely to encounter hostility in their daily lives or feel a need to conceal their atheism. Our quantitative data limit our ability to understand individual reasoning for perceiving hostility or concealing an identity or how people relate them in their mind, and we need more qualitative work in this area to understand the lived experience of atheist stigma.

It may also be the case that there is much more social differentiation in atheists’ identity concealment than in their perception of hostility. That is, while atheists in our sample are the most likely nonreligious group to perceive hostility, there are few significant differences in these perceptions across different social locations. However, there are significant differences across social locations regarding whether atheists conceal their identity. This means that social location matters for how people manage the stigma associated with atheism and shapes if and when atheists feel comfortable sharing their atheism with others.

Relatedly, the increased polarization of American society may also be a contributing factor here. Research shows that Americans are increasingly likely to live in their own news and social media “bubbles” and to surround themselves with like-minded friends (Bail 2021). As the divide between “religious conservatives” and “secular liberals” grows, it could be that atheists do continue to perceive hostility from the religious right, but these divides allow them to surround themselves with like-minded people as a way to reduce the impact of that stigma and to feel morally superior to those doing the stigmatizing. This would explain why atheist Republicans, atheists in the South, and atheists who still attend church are more likely to conceal their identities—it is harder for them to build a network of like-minded people because atheism is less accepted in these social contexts.

An important limitation to our study is that we do not have a measure of whether respondents are members of a supportive atheist organization or whether they have social networks comprised of other atheists. This means we cannot directly investigate whether our findings may be the result of engagement with such groups or social networks. As is evident from our SEM models, while we find that social and geographic locations do shape perceived hostility and identity concealment among atheists, there is still a lot of variation to be explained. Social network and group involvement measures would likely account for some of this remaining variation, and future studies should do more to investigate these relationships (see Brewster et al. 2020). These measures may also have helped explain some of our findings. For example, the fact that age is associated with less perceived stigma could be because, as people age and become more comfortable with sharing their atheism with others, they are able to build supportive social networks in ways that younger atheists have not done yet.

Another limitation of our study is that we lack measures of identity centrality and salience, which have been highlighted as being important factors that shape stigma management strategies, and future research should investigate these relationships further. Future studies should also investigate how perceived hostility and identity concealment work across different social contexts. Are atheists more likely conceal their atheism at work versus around their family? Is the kind of hostility perceived from an atheist’s co-workers different than the kind of hostility perceived by their family members? These are questions that our survey cannot address but that future research should investigate.

In the wake of a rapidly changing religious landscape in the United States in which more and more people are opting out of religion, these findings provide new insights into how atheists manage the continued stigma of atheism. While atheists do continue to perceive stigma, though this stigma may be reducing overall, we find that many chose to disclose their atheism to others anyways and that social and geographic location shape who is more likely to do so. These findings have implications for how researchers understand the context-specific nature of religious discrimination and stigma management more broadly.

About the Authors

Jacqui Frost is a Postdoctoral Research Fellow in the Department of Sociology and Religion and Public Life Program at Rice University. She is a cultural sociologist whose research investigates the causes and consequences of religious disaffiliation, the relationships between religion and health, and the social dynamics between religion and science. Her research has been published in American Sociological Review, Poetics, Journal for the Scientific Study of Religion, Nonprofit and Voluntary Sector Quarterly, and Social Currents.

Christopher P. Scheitle is an Associate Professor of Sociology at West Virginia University. He has published three books and over eighty articles examining issues related to religious discrimination and victimization, the social dynamics between religion and science, the organizational structure of American religion, and the nature of individuals’ religious identity.

Elaine Howard Ecklund is the Herbert S. Autrey Chair in Social Sciences, Professor of Sociology, and Director of the Religion and Public Life Program at Rice University. Over the past several years, Elaine’s research has explored how scientists in different nations understand religion, ethics, and gender; religion at work; and the overlap between racial and religious discrimination in workplaces. Elaine is the author of seven books, over 100 research articles, and numerous op-eds.

Endnotes

1

We chose not to specify a causal direction between perceived hostility and identity concealment because research suggests that concealment can both fuel perceived hostility and be a response to perceived hostility. Given this causal complexity, we simply allow the error terms to covary.

2

We did examine alternative models using Stata’s generalized structural equation model command (gsem), which allowed us to specify the measures as ordinal in nature (ologit estimation method). Our substantive findings were largely the same, but the gsem command does not allow for use of the maximum likelihood with missing values method, which reduces our sample size (N = 665). Ordinal logistic models also have disadvantages when it comes to presentation and interpretation. Given these disadvantages and the similarity in findings, we present the sem analysis treating the observed measures as continuous.

3

Muslims report the greatest perceived hostility across the major religious groups (mean = 2.84). Atheist and Jewish individuals’ report the next highest levels of perceived hostility (2.36). All other religious groups (Christian, Buddhist, Hindu, and something else) are significantly lower than these groups (ranging from a mean of 1.67 for Hindu to 2.05 for the something else group). For identity concealment, Christian and Jewish individuals have the lowest means (2.43 and 2.40, respectively). Concealment rises somewhat among Muslim individuals (2.55) but is higher among Buddhist, Hindu, something else, and the nonreligious groups.

4

We also conducted an exploratory factor analysis outside of the context of structural equation modeling (principal factors method, unrotated) that indicated that these six items represent two underlying factors. Factor 1 (Concealment) eigenvalue = 2.16; Factor 2 (Perceived Hostility) eigenvalue = 1.51. No item’s factor loading was greater than .20 on the other factor in this exploratory factor analysis.

5

We did examine the possibility of a curvilinear relationship between perceived hostility and identity concealment among atheists, but we did not find any evidence supporting this.