Forms , Frequency , and Correlates of Perceived Anti-Atheist Discrimination

The nationally representative 2008 American Religious Identification Survey found that 41% of self-identified atheists reported experiencing discrimination in the last 5 years due to their lack of religious identification. This mixed-method study explored the forms and frequency of discrimination reported by 796 self-identified atheists living in the United States. Participants reported experiencing different types of discrimination to varying degrees, including slander; coercion; social ostracism; denial of opportunities, goods, and services; and hate crimes. Similar to other minority individuals with concealable stigmatized identities, atheists who more strongly identified with their atheism, who were “out” about their atheism to more people, and who grew up with stricter familial religious expectations reported experiencing more frequent discrimination. Implications for future research tied to the ongoing religion/spirituality-health debate are discussed.


Introduction
Social acceptance of minority groups is often a struggle in U.S. culture.Racial/ethnic minorities, women, and people with disabilities, among other marginalized groups, continue to report experiencing discrimination (Sue & Sue, 2008).Discrimination is generally defined as "practices and actions which have a differential and negative effect on minority group members" (e.g., Allport, 1954, Feagin & Eckberg, 1980).Pager and Sheperd (2008) draw the distinction between differential treatment and disparate impact.Differential treatment is the unequal treatment of individuals because of their perceived minority group membership, while disparate impact results when ostensibly equal treatment results in the favoring of members of one group over another (e.g., a nation-wide flat tax rate of 14% applied to working poor and billionaires alike).Similarly, Sue et al. (2008) present a typology of discrimination which includes microassaults (blatant derogation; e.g., racial slurs), microinsults (statements or actions, often unintentional, that demean minority group members; e.g., telling an Asian American man that he "speaks good English"), and microinvalidations (actions that negate the thoughts, feelings, or experiences of minority group members; e.g., assuming all Black individuals were raised in urban areas).Other scholars have further distinguished between overt versus covert discrimination (e.g., Sue et al., 2008); routine discrimination versus discrimination associated with major life events (Essed, 1990(Essed, , 1991)); and discrimination at the individual, institutional, and structural levels (Pincus, 1996).
Recognizing the importance of understanding discrimination in order to reduce its prevalence (Stuber, 2008), social scientists have studied discrimination across a wide range of minority populations.Unfortunately, the experiences of some marginalized groups have received far less empirical attention than others.
In particular, very little research has focused on religious discrimination (Malik, 2001;Nadal, Issa, Griffin, Hamit, & Lysons 2010), despite the fact that anti-religious hate crimes are the secondmost common type of hate crime reported to the FBI (Federal Bureau of Investigation, 2010).In their review of this literature, Nadal and colleagues observed that most of this research has, understandably, focused on the experiences of Jews and Muslims.Of the 1,552 anti-religious hate crime victims reported to the FBI in 2010, 67.0% were against Jews and 12.7% against Muslims (Federal Bureau of Investigation, 2010).These hate crimes involved, among other things, physical assault, verbal intimidation, and property damage.In addition to hate crimes, individuals from these two groups have reported being teased; stared at; unfairly stereotyped in person and in the media; and treated unfairly in employment, housing, education, and places of business (Allen & Nielsen, 2002;Dunn, Klocker, & Salabay, 2007;Driedger & Mezoff, 1981;Krämer, 2006;Sheridan, 2006;Weinrach, 2002).While social scientists are starting to understand the nature and impact of discrimination perpetrated against Jews and Muslims, the experiences of other (non)religious groups have been almost entirely overlooked in the academic literature.U.S. atheists have received particularly little attention (Bainbridge, 2005;Pasquale, 2007;Vernon, 1968).

Marginalization of Atheists
Atheists, defined as those who lack a belief in God or believe in the explicit nonexistence of God (Cliteur, 2009), are not popularly considered to be a marginalized group in the U.S. (Thornton, 2007), but recent research has called this position into question.According to a national survey, 78.6% of Americans believe that atheists do not share their vision of American society, and 47.6% would disapprove if their child wanted to marry an atheist (Edgell, Gerteis, & Hartmann, 2006).A recent poll found that half of Americans say they would not vote for an otherwise qualified atheist candidate for president (Gallup, 2011).In addition, atheists are often stereotyped as rebellious, immoral, and hedonistic, among other descriptors (Harper, 2007).
Because stereotypes underlie prejudicial attitudes which are believed to motivate acts of discrimination (Dion, 2003;Stangor & Schaller, 1996), it is unsurprising that atheists have reported experiencing discrimination in anecdotal and exploratory qualitative accounts.For instance, author David Mills was directly threatened with violence by local police when he announced his intention to protest a visiting Christian faith healer who encouraged diabetics to stop taking insulin and pray for divine intervention instead (Mills, 2006).In 2007, Army Specialist Jeremy Hall reported being threatened by fellow soldiers upon revealing his atheism (MSNBC, 2007).A male participant in Hunsberger and Altemeyer's (2006) study of American atheists reported, "My kids have been harassed at school, I've been a victim of religious discrimination at work, my car has been vandalized, I've received death threats via email, mail, and under my windshield wipers" (p.49).Other atheists in the same study reported being shunned by family, losing friends, and being fired from jobs.Overall, more than half of the atheists from each of the authors' three samples reported experiencing difficulties with relatives and friends due to their nonbelief.In addition, a recent law review found that divorced parents have had custody rights denied because of their atheism (Volokh, 2006).Discrimination against atheists has also been demonstrated in the laboratory, where participants gave priority to Christian patients over atheist patients when assigning medical resources (Furnham, Nicholas, & McClelland, 1998).Finally, the nationally representative American Religious Identification Survey (ARIS) found that 41% of selfidentified atheists reported personally experiencing discrimination in at least one context (e.g., family, workplace, school) in the past 5 years due to their lack of religious identification (Cragun, Kosmin, Keysar, Hammer, & Nielsen, 2012).While the ARIS provides a sense of the considerable prevalence of discrimination reported by U.S. atheists, the findings did not specify the different forms of discrimination that underlie this figure.
The accounts mentioned above reveal that atheists report experiencing some of the same forms of discrimination that other marginalized minorities report experiencing, including being physically assaulted by peers; having their lives threatened; being rejected by their families; and being denied employment, fair service at area businesses, and membership in community organizations (Downey, 2004;Heiner, 1992;Kaye, 2008;Nussbaum, 1999;Peters, 2009).Research also suggests that atheists report experiencing forms of discrimination that appear unique to the (non)religious milieu, including being subjected to unwanted prayers and proselytizing by hospital staff (Smith-Stoner, 2007), having their wishes for a non-religious funeral superseded by the beliefs of significant others (Saeed & Grant, 2004), having to swear an oath to God before being allowed to join Veterans of Foreign Wars or the American Legion (Heiner, 1992), and being told that their atheism is just a phase of being angry with God and that if they pray to God they may be cured of their disability (Hwang, 2008).
The Anti-Discrimination Support Network, which documents acts of discrimination against the nonreligious, provides an especially diverse account of the slights directed toward atheists, including: being forced to join in prayer; being denied water as a serviceman in the Iraqi desert without first agreeing with the presiding Army Chaplain to be baptized; being told by a public official that, "It is dangerous to even let our children know that your philosophy exists;" having one's atheist group denied the opportunity to advertise events alongside religious organizations; and being told by a teacher that one is a "child of the devil" with "no right to live in America" (M.Downey, personal communication, October 15, 2008).In summary, it appears that discrimination experiences reported by atheists may be more common (Cragun et al., 2012) and heterogeneous than popularly recognized.
While these accounts hint at different forms of discrimination that atheists report experiencing in the U.S., gaining a more systematic empirical understanding is important for two reasons.First, the above narratives suggest that anti-atheist discrimination is a potentially serious and overlooked form of discrimination.Heeding the calls of social science organizations, such as the American Psychological Association, to act to eliminate discrimination based on (non)religion requires an empirical understanding of what forms anti-atheist discrimination takes (American Psychological Association Council of Representatives, 2007).Second, this line of research could contribute to the ongoing debate regarding the impact of religiosity/spirituality (R/S) on well-being (Hwang, Hammer, & Cragun, 2011;Sloan, 2006).There is a substantial body of evidence suggesting that R/S is associated with better physical and mental health (Hill & Pargament, 2008), though the relationship is typically quite small (Koenig, 2009).While recent research has demonstrated the presence of a curvilinear relationship between belief and health (i.e., confidently religious and atheist individuals report higher well-being than religiously uncertain and agnostic individuals; see Galen & Kloet, 2011a, for a review), other research contends that atheists tend to experience more psychological problems and are generally less happy than religious people (Altemeyer, 2010;Reed, 1991;Schumaker, 1992;Steinitz, 1980;see Zuckerman, 2009, for a review).While these findings have been attributed to "constructing a moral system from scratch" or "a lack of accountability to God or a wider community" (Hall, Koenig, & Meador, 2008, p. 371), the possible effects of social marginalization (perceived and/or actual) have not yet been considered.
Persistent and pervasive social marginalization, according to the minority stress model (Meyer, 2003), can lead to negative health consequences independent of other demographic predictors (Clark & Adams, 2004;Hatzenbuehler, Nolen-Hoeksema, & Erickson, 2008;Herek & Garnets, 2007).Thus, the minority stress model predicts that U.S. atheists may experience adverse health consequences if they are socially marginalized by others.However, there is currently insufficient research to determine if social marginalization accounts, in part or in whole, for the suggested health disparity between R/S individuals and atheists living in the U.S. If future research confirmed this prediction, this would have important implications for the R/S-health debate.It would raise the possibility that a secular worldview could be just as conducive to physical and mental well-being as a R/S worldview, provided anti-atheist discrimination and its impact is effectively reduced.If, on the other hand, future research was to find that discrimination plays a negligible role in atheists' lesser well-being, this would nullify the argument that anti-atheist discrimination can account for the supposed health disparity (Cragun et al., 2012;Hwang et al., 2011).It would also suggest that internal factors, such as anger toward God (Exline, Park, Smyth, & Caregy, 2011), lack of life purpose (Emmons, 1999), unclear self-concept (Campbell et al., 1996), and moral relativism (Hall et al., 2008), may better account for the health disparity between religious and non-religious persons.
However, such discrimination research is not possible until a comprehensive accounting of the various forms of anti-atheist discrimination occurs.For example, the content domain of anti-atheist discrimination must be delineated before perceived anti-atheist discrimination can be properly operationalized by a validated self-report measure (Clark & Watson, 1995).It is important to note here that, as with all self-report research on other stigmatized minority groups, self-reports of discrimination experiences are based on individual perception.This perception is shaped both by internal psychological processes (e.g., group identification), and by objective external events (discriminatory acts perpetrated against the individual ;Kaiser & Pratt-Hyatt, 2009).Thus, like the majority of psychological research on the recipients of discrimination (Berg et al., 2011), we focused on perceived discrimination in this investigation.

The Current Study
The current study had two aims.The primary aim was to provide the first comprehensive empirical account of the various forms of perceived anti-atheist discrimination.We documented the frequency with which a volunteer sample of self-identified atheists living in the U.S. reported experiencing 29 different forms of discrimination derived from research on other stigmatized minorities and extant anecdotal or exploratory qualitative accounts of anti-atheist discrimination.We then subjected participants' first-person accounts of discrimination to textual analysis in order to capture novel qualitative themes not accounted for by the quantitative analysis.Furthermore, because the perceived frequency of anti-atheist discrimination has been found to vary across contexts (Cragun et al., 2012), the secondary aim of our study was to test the following three hypotheses, which are described in detail in the following section, regarding the relationship between perceived discrimination and constructs of theoretical relevance: 1. Individuals who more strongly identify with their atheism would report experiencing more frequent discrimination.2. Individuals who are "out" about their atheist identity to more people would report experiencing more frequent discrimination.3. Strictness of familial religious expectations would be significantly positively associated with the frequency of perceived social ostracism by one's family, and this relationship would be significantly moderated by one's "outness".

Theoretical Constructs Associated with Perceived Discrimination
Research with marginalized minority groups often finds that those who identify more strongly with their in-group report experiencing more discrimination.Group identification is defined as the importance of group membership (i.e., the defining characteristic shared with certain others) to selfdefinition (Tropp & Wright, 2001).Some group-identification theories focus on the victim's internal psychological process.The rejection-identification model, for example, posits that as perceived discrimination increases, individuals seek to protect their self-esteem by increasing their identification with the lower-status in-group (Branscombe, Schmitt, & Harvey, 1999).Similarly, Tajfel & Turner (1986) assert that stronger identification with the group encourages individuals to interpret ambiguous events as discriminatory (Major, Quinton, & Schmader, 2003).Other group-identification theories focus on the perpetrator's psychological process.The prejudice-distribution model, for example, argues that members of the dominant group react more negatively towards those minority individuals who appear to be more strongly identified with their in-group because such identification is seen as challenging the dominant worldview of those in power (Kaiser & Pratt-Hyatt, 2009).Similarly, Cragun et al. (2012) found that nonreligious individuals who explicitly identified as atheist or agnostic were more likely to report experiences with discrimination.Cragun and colleagues inferred that the self-identified labels "atheist" and "agnostic" represent a stronger irreligious identification, one that religious individuals may find especially disagreeable.However, these findings could also be taken to support the rejection-identification and social-identity models.In light of these theoretical perspectives, we hypothesized that (H1) individuals who more strongly identify with their atheism would report experiencing more frequent discrimination.In line with prior research examining the relationship between identification and perceived discrimination among racial, sexual, and religious minorities (e.g., Awad, 2010;Fingerhut, Peplau, & Gable, 2010;Jasperse, Ward, & Jose, 2011;Sellers & Shelton, 2003), we anticipated finding a small but significant positive correlation.
Like lesbians, gays, and bisexuals (LGB's), atheists have the option of "staying in the closet" regarding their stigmatized identity.Research suggests that closeting, though often stressful (Miller & Major, 2000), is the most common coping strategy that LGB's use in an effort to defend against serious discrimination (Hetrick & Martin, 1987).When LGB's are successful in hiding their minority status from bigoted others, this reduces the overt discrimination experienced (e.g., receiving personal threats; Jones et al., 1984;Herek & Berrill, 1992), but likely has a negligible effect on non-directed discrimination experienced (e.g., hearing gay slurs in public).Similarly, individuals who disclose their mental illness are more likely to experience discrimination (Corrigan & Watson, 2002).This same dynamic could be expected to apply to atheists as well.Therefore, we hypothesized that (H2) individuals who are "out" about their atheist identity to more people would report experiencing more frequent discrimination.Based on prior research on the relationship between "outness" and perceived discrimination (e.g., Griffith & Hebl, 2002;Jasperse et al., 2011), we anticipated finding a small but significant positive correlation.
Increased stress among LGB's growing up in more devout households has previously been documented (Newman & Muzzonigro, 1993).Likewise, accounts suggest that atheists who grow up in homes with stricter religious expectations are more likely to report discrimination by family members (Altemeyer & Hunsberger, 1997).This may be due to the fact that atheism-a personal rejection of religion-may particularly trouble family members who highly value family members' religious commitment (Stokes & Regnerus, 2009).Furthermore, it is likely that the strength of the relationship between strictness of religious expectations and reported discrimination would depend on how "out" a given atheist is.Accordingly, we hypothesized (H3) a significant positive association between the strictness of familial religious expectations and the frequency of perceived social ostracism by one's family, and that this relationship would be significantly moderated by one's "outness".As we were not able to locate any prior research examining the correlation between strictness of religious expectations and perceived discrimination, we had no a priori expectations regarding the size of the coefficient.

Method Participants and Procedures
A volunteer sample of 1,038 self-identified atheists living in the United States was recruited via a website which focuses on research on the nonreligious (http://www.atheistresearch.org/).The website attracts regular visitors from online search engines (who use search terms such as atheist, research, and studies), secular organization websites, and word of mouth among secular individuals.Thus, the majority of internet survey participants were derived from these sources.In addition, the first author contacted Margaret Downey, president of the Atheists Alliance International, who agreed to distribute an announcement about the study to individuals with whom she has had prior correspondence due to her involvement in the Anti-Discrimination Support Network.Due to the relative scarcity of atheists in the U.S. population, internet recruitment was deemed useful in achieving an adequate sample size.Previous research has noted that online samples are particularly useful when the population of interest is small or otherwise difficult to locate (Gosling, Vazire, Srivastava, & John, 2004;Kraut et al., 2004;Reips, 2000).Additionally, results from Internet data have been found to be consistent with results from paper-and-pencil measures (Gosling et al., 2004).
Procedures were approved by the University of Missouri institutional review board.After providing informed consent, participants completed all survey measures (see Measures section) and demographic questions, as well as other items related to their identity and experiences as an atheist.Discrimination and atheist identity items were counterbalanced to reduce the chance of order effects.Lastly, participants were presented with the debriefing page.
To clean the data, we first removed data from 11 individuals who participated as minors, despite 49 the informed consent restricting participation to adults.We then removed the data for 208 individuals who failed to respond to more than 90% of the items (Bennett, 2001), many of whom discontinued after completing the first page of the survey.In the retained sample (n=817), missing data across variables ranged from a low of 0% for rejection by coworkers or classmates to a high of 6.2% for familial religious expectations.Participation was explicitly limited to those who self-identified as atheists living in the U.S. Specifically, the description of the survey posted to the website stated "Those who consider themselves atheists are eligible to participate in this study."As a validity check, an item asking participants about their perspective on god was included.All cases indicated an absence of belief in god and thus passed this check.Participants' demographic characteristics are displayed in Tables 1 and 2. Due to small cell sizes, we collapsed several of the demographic groups to insure greater statistical power for demographic analyses.

Measures
Perceived discrimination.Turner and Wheaton (1995) suggest that measures of stressful events should be customized to the cultural group under investigation.As no current measure of perceived anti-atheist discrimination existed, we adapted several items from the Gay and Lesbian Oppressive Situations Inventory-Frequency (GALOSI-F; Highlen, Bean, & Sampson, 2000), and added additional items, based on prior literature, to account for 29 different forms of discrimination (see Table 3).These additional items assessed disparate impact and microinvalidations (e.g., "Being expected to participate in religious prayers against my will").Participants were given the following instructions: "Below are some situations that you may have encountered.Think about each situation and indicate how often this situation has occurred."Participants indicated how often each of the 29 situations had occurred in their lifetime using a 4-point Likert-type scale (0 = never, 1 = once, 2 = two or three times, and 3 = four or more times).It is important to note that these 29 forms of discrimination were called "situations" rather than "discrimination experiences" to reduce the potential impact of demand characteristics.
A post-hoc overall perceived discrimination total score was computed by summing the reported frequency of each of the 29 forms of discrimination.Scores on perceived discrimination could range from 0 to 87 (i.e., if participant selects "3 = four or more times" for each of the 29 forms of discrimination).To facilitate discussion of the results, the authors independently grouped the 29 forms into a priori conceptual categories, with the aim of deriving three to eight categories.Since the present investigation aimed to explicate the various forms that anti-atheist discrimination may take, rather than to determine which forms tend to be co-reported, categories derived from factor analysis were not sought.Through group discussion, the authors arrived at a consensus on the number and contents of categories.50 This process resulted in six categories of discrimination: slander; coercion; social ostracism; denial of opportunities, goods, and services; hate crimes; and other (see Table 3 for which items were included in each category).It should be noted that the three forms that constitute the "hate crimes" category (e.g., physically threatened, property damaged, physically assaulted) align with the "offense types" considered hate crimes by the FBI (2010).Identification.To assess the strength of identification with their atheist worldview, participants were asked to indicate how much they agreed (1 = strongly disagree to 7 = strongly agree) with the following statement: "My atheism is an important part of my personal identity."This item was adapted from the Centrality scale of the Multidimensional Inventory of Black Identity (Sellers, Rowley, Chavous, Shelton, & Smith, 1997), which measures the extent to which being Black is central to the respondents' definition of themselves and has been found to significantly predict perceived discrimination (Sellers & Shelton, 2003).
Outness.Participants' level of "outness" was assessed by asking them to indicate whether they were out about their atheism to the following 10 persons or groups of people: mother, father, siblings, relatives, new religious friends, old religious friends, peers/classmates, supervisors/teachers, partner, strangers.This measure is a dichotomized adaptation of Mohr and Fassinger's (2000) Outness Inventory, which was found to be associated with lesbians' concerns about others' misunderstandings and increased stress related to family issues (Lewis, Derlega, Berndt, Morris, & Rose, 2001).A total outness score was derived by adding 1 for every person/group the participant reported being out to, for a minimum score of 0 (completely closeted) and a maximum score of 10 (completely out).Because we intended to derive only a total outness score, we chose to reduce participant burden by dichotomizing the items.
Strictness.Strictness of religious familial expectations was measured with the single item, "Generally, how strict were familial religious expectations during your upbringing?" measured on a 4point scale (1 = no religious expectations at all, 2 = relatively relaxed expectations, 3 = somewhat firm expectations, and 4 = very strict expectations).
Discrimination stress narratives.In addition to the 29 perceived-discrimination survey items, participants were asked to free respond to the following question: "Please describe, in your own words, any stressors and/or hassles that you feel are, or have been, directly related to your atheism."This question was adopted from a study of gay and lesbian stress experiences (Lewis et al., 2001).We anticipated that the responses to this question, in addition to reiterating and corroborating the quantitative survey responses, would indicate other themes that supplement and provide further nuance to our understanding of the discrimination reported by atheists.
We employed textual analysis to develop empirically descriptive qualitative themes from the data.Initially, line-by-line coding (Emerson, 2001) of all 817 responses was employed in order to begin to process and conceptualize the content of the data, sensitize the researcher to the generic patterns, and render the data suitable for additional focused coding (Lofland, Snow, Anderson, & Lofland, 2006) and further qualitative analysis.Starting with the fairly tedious practice of line-by-line coding before conducting a more focused form of analysis has the benefit of reducing the researcher's taken-forgranted assumptions about the material, and allowing the researcher to discover novel aspects of the data (Emerson, 2001).These specific coding techniques, along with separate memo-taking and the construction of a qualitative and general thematic analysis approach (Gubrium & Holstein, 1997;Maxwell, 2005) were continuously employed throughout, until the final themes (outlined below) emerged.Thus, these themes were generated from the empirical data offered by way of the responses of atheists in their own words.The frequency with which these qualitative themes occurred among participants was not assessed, as comparing such frequencies to those derived from the quantitative responses would be misleading in that the methods used to elicit responses differed substantially.

Results
As discussed in greater detail in the limitations section, readers should consider the results of this study in light of its methodology.First, our sample is not representative of the entire atheist population living in the U.S. Our subject pool was limited to those who viewed the data collection website and who selfidentified as atheists.Thus, those demographic groups who are unlikely to navigate to the data collection website (e.g., those without a computer, those not interested in non-religious research) and who are atheists "by belief," but not by self-identification, were unlikely to participate in this survey.The demographic characteristics of the current sample are generally consistent with a demographic profile of atheists reported by nationally representative population surveys (e.g., ARIS 2008, which included 329 atheists; see Kosmin & Keysar, 2008;Kosmin, Keysar, Cragun, and Navarro-Rivera, 2009).While our sample was somewhat different than the sample of atheists assessed in the ARIS in terms of gender (76.3% male in ARIS; 55.8% in our sample), our sample was similar across race (73.4% white in ARIS; 82.9% in our sample), education (13% graduate degree in ARIS; 24% in our sample), income (24.8%earn more than $75,000 per year in ARIS; 33.7% in our sample), region (24.5% live in the Northeast, 18% live in the Midwest, 32.1% in the South, and 25.4% in the West in ARIS; 14.8%, 22.5%, 32.3%, and 24.4% respectively in our sample), and community size (33.3%live in rural areas in ARIS; 29.6% of our sample).Thus, while there are gender composition differences, our sample is generally similar to nationally representative data.
Second, like most studies that assess perceived discrimination among minority group individuals, several of the survey items focused on participants' negative experiences related to their atheist identity.This focus on negative experiences may have biased the events that participants recalled from memory, perhaps leading them to recall more negative experiences than they otherwise would have.

Outlier Removal
We examined the z-scores for each of the overall scales to check for univariate outliers (Tabachnick & Fidell, 2001).No outliers were found for the outness and strictness scores.In one case on atheist discrimination and 18 cases on identification, there were outliers at p < .001(i.e.z-scores above 3.29).Thus, we removed these cases from subsequent analyses.To check for multivariate outliers, we examined Mahalanobis distances among the variables (Tabachnick & Fidell, 2001).One additional case was found to be an outlier at p < .001(Mahalanobis distance ≥ 21.72), and so was dropped from subsequent analyses (n = 796).

Perceived Discrimination
We first documented the reported frequency of 29 different forms of discrimination.Table 3 summarizes the percentage of participants who reported experiencing each of the 29 forms of discrimination assessed.The average participant reported experiencing about 10 of the 29 possible types of discrimination assessed by the perceived discrimination measure (M = 9.73; SD = 5.05).The percentage of participants who reported experiencing at least one form of discrimination in a given category were as follows: slander (96.7%); coercion (92.5%); social ostracism (56.4%); denial of opportunities, goods, and services (15.8%); hate crimes (13.7%); and other (83.7%).The five forms most frequently reported were: witnessing anti-atheist comments in newspapers or on television (94.7%), being expected to participate in religious prayers against one's will (79.1%),being told one's atheism is sinful, wrong, or immoral (75.2%), being asked to attend religious services or participate in religious activities against one's will (74.4%),and being treated differently because of one's atheism (67.5%).Demographic differences in overall perceived discrimination reported by participants are summarized in Table 1.When Bonferroni-adjusted for multiple comparisons, independent-sample ttests, one-way ANOVA tests, and zero-sum correlations (age; not shown in Table 1) did not indicate any significant demographic differences.

Discrimination Stress Narratives
To capture qualitative themes not accounted for by the 29 perceived-discrimination items, we subjected participants' first-person accounts of discrimination to textual analysis.Six novel, and often inter-related, themes emerged from the free responses to the qualitative question, and were identified as important additional sources of discrimination stress for atheists: (a) assumed religiosity, (b) lack of a secular support structure, (c) lack of church and state separation, (d) negative effects on family, (e) unreciprocated tolerance, and (f) anticipatory stress.
Assumed religiosity.Past research has found that the pervasive, seemingly "default" presumption of theism in the United States can be an important social dynamic for atheists (Smith, 2011).Many respondents expressed their frustration with the fact that others would often assume the respondents held a belief in a god or afterlife.Assumed religiosity can emerge in a variety of social exchanges, but a salient example involves the grieving process that follows the loss of intimate others.For instance, one participant, while mourning the death of a loved one, expressed irritation regarding comments such as "he's in a better place now."While participants may recognize that those offering such oft-repeated assertions are well-intentioned, they may also find these statements to be hollow, untrue, and/or an empty source of comfort and meaning.
Lack of a secular support structure.Despite the documented recent growth in secular and atheist groups at both the local and national levels (Baker & Smith, 2009), many respondents reported experiencing stress at the perceived lack of social, group, and community support for their atheist identities and viewpoint.The relative preponderance of religion, and the lack of an alternative atheist counterpart, in volunteer opportunities, charitable work, and community support was a source of frustration for many respondents.Though the social and symbolic boundaries between theism and atheism are important for atheist identification (Smith, 2011), the marginalizing aspects of being an atheist in the United States, including the perceived absence of some of the important social, cultural, and organizational resources which their religious counterparts enjoy, is a clear source of stress for some atheists.The lack of secular/nonreligious, in comparison to religious, holidays was a frequent example.In combination with the "assumed religiosity" discussed above, participants sometimes expressed frustration with not only lacking holidays that recognize non-belief, but also simultaneously having to negotiate and/or avoid the many officially sanctioned religious holidays, and the assumption of everyone's recognition of, and participation in, such events.Interestingly, a few participants even remarked on the additional stress they experienced while trying to "not make trouble" for others during religious holidays or other events.In the words of one respondent "…it seems like my atheism makes OTHERS feel stressed around the religious holidays that I don't share.It's as though they feel like I'm snubbing them because I won't go through the motions on their holidays."Lack of church and state separation.Participants reported distress over what they saw as the declining wall of separation between church and state in the U.S. Comments about "being taxed to support religious programs," witnessing state government attempts to "infuse religion into schools, work, and public places," being subjected to prayer at civic and other non-religious meetings, and experiencing "teachers who worked the importance of worship into classroom discussion as often as possible" were frequent examples of stressors experienced by some participants.The broader social and political issue of church and state separation is of course one that concerns many religious people as well; for participants, the issue is perceived as directly relevant to their marginalized identity status.
Negative effects on family.Stressors were not confined to the perceived affronts against respondents themselves.The effects of their identification with atheism spilled over into other aspects of life, particularly child-rearing.Some participants experienced stress over their children suffering verbal harassment by teachers and classmates.Other respondents experienced stress because their religious family members or other people would question how they, as atheists, could raise their children properly.One participant even remarked that his "son's mother has tried to claim [he] is unfit as a parent because he is an atheist."Given that proper child rearing can be highly sensitive subject, this theme appears to be of particular importance for many atheists.
Unreciprocated tolerance.Religious and other forms of tolerance are generic values in American culture, but these same cultural ideals may not be applied to atheists.As a result, public disposition toward atheists can be a natural source of stress and anxiety for atheists.As noted earlier, studies have shown that atheists are among the most "distrusted" minority groups in the United States (Edgell et al., 2006).The responses of participants indicate that it is not just the intolerance shown toward them that is stressful: they are simultaneously frustrated because religious individuals demand that their views be tolerated and respected without feeling obligated to extend that same tolerance and respect to atheists.As one participant plainly remarked, "I dislike having to tolerate religion when most of the religious folks I'm around have little tolerance or will to understand or learn about atheism." Anticipatory stress.Finally, participants reported concern about future potential stress regarding their atheism.Anticipatory stress is contingent upon the type of social situation encountered.Situations in which the discussion of god or other religiously-themed topics arise are likely to produce greater anticipatory stress.Weddings, funerals, baptisms, and other religiously-infused rites of passage of close relatives and friends were indeed sources of stress for some atheists.Decisions about whether to attend or avoid these events, and how or whether to negotiate and/or disclose one's atheist viewpoint in such social situations are stressors in and of themselves that are felt because of the possibility of the stressor(s) that could be experienced at these and similar events.Participants often felt anticipatory stress because they had already experienced problematic interactions and various stressors in religiously-laden situations.

Theoretical Constructs Associated with Perceived Discrimination
The secondary aim of our study was to test three hypotheses regarding the relationship between perceived discrimination and (a) identification, (b) outness, and (c) family strictness.We first hypothesized that (H1) participants who strongly identified with their atheism would report experiencing more discrimination.Results supported hypothesis 1: there was a small but significant positive correlation between identification and overall perceived discrimination (r = .19,p < .001).
Our second hypothesis proposed that (H2) participants who were out about their atheist identity to more people would report experiencing more discrimination.Results supported hypothesis 2: there was a small but significant positive correlation between outness and overall perceived discrimination (r = .17,p < .001).
Our third hypothesis stated that (H3) there would be a significant positive association between strictness of familial religious expectations and the reported frequency of being socially ostracized by one's family (assessed by the item "Being rejected, avoided, isolated, or ignored by family because of my Atheism"), and that this relationship would be significantly moderated by one's outness.To test this moderator hypothesis, we first centered strictness and outness, and then multiplied them to create the interaction term (Baron & Kenny, 1986).Main effects were entered at Step 1 of a multiple regression analysis, and the interaction was entered at Step 2. The results of this analysis were significant (R 2 = .26),F(3, 746) = 18.14, p < .001.Strictness (β = .24),outness (β = .10),and their interaction (β = .09) were all significant predictors of reported social ostracism by one's family.Thus, results supported hypothesis 3: there was a small but significant positive association between the strictness of familial religious expectations and the frequency of perceived social ostracism by one's family, and this relationship was significantly moderated by one's outness.Specifically, greater levels of outness predicted a stronger relationship between strictness and ostracism.

Discussion
The primary aim of the current study was to provide the first comprehensive empirical account of the various forms of perceived discrimination reported by self-identified atheists living in the U.S. Quantitative and qualitative responses revealed that participants report experiencing a wide variety of (c)overt forms of discrimination, from more frequent, everyday forms of discrimination to less frequent forms of lifetime discrimination (Essed, 1990(Essed, , 1991)).To facilitate interpretation, these forms of perceived discrimination will be discussed in the context of five conceptual categories introduced in the method section, in decreasing order of reported prevalence: slander; coercion; social ostracism; denial of opportunities, goods, and services; and hate crime.
Slander was a nearly ubiquitous form of perceived discrimination.Only 5% of the sample reported never having witnessed anti-atheist rhetoric in print or television media, suggesting that participants found media messages denigrating atheism to be widespread.It was also relatively common for participants to report that they were told their atheism is an immoral choice, reminiscent of the messages directed at LGB's regarding the "immorality" of their sexual desires and/or choices.In the LGB literature, it is thought that exposure to such microassaults can contribute to internalized negative social attitudes towards the self (Meyer, 2003).Whether and how such a process occurs within atheists are questions for future research.
The vast majority of participants reported experiencing coercion due to their atheist identity.Feeling pressure to perform religious behaviors or risk social consequences may be experienced by some atheists as a stressful violation of one's right to express a nonreligious worldview.While having to say "under God" when reciting the U.S. Pledge of Allegiance may result in negligible stress, being asked to hide or even renounce one's atheism may be experienced as a significant violation of one's liberty to freely express oneself.As another example of disparate impact (Pager & Shepherd, 2008), some atheist participants lamented having religion forced on them in ostensibly secular civic spheres (i.e., a lack of church and state separation).
Social ostracism was reported by more than half the sample.Some research suggests that atheists are generally more introverted and less socially active in the community than theists (Bainbridge, 2005;Hout & Fischer, 2002;Barna Research Group, 2007; though see Galen & Kloet, 2011b).While Bainbridge interprets this as indicating that lacking social obligations encourages disbelief in God, one might argue that feeling shunned by family, coworkers, and community members makes social connections, community involvement, and civic participation more difficult.Research with Muslim Americans suggests that individuals who perceive their environment as discriminatory may demonstrate higher levels of subclinical paranoia and hypervigilance, which tend to exacerbate social withdrawal (Rippy & Newman, 2006).In addition to being ostracized, almost half of the sample reported being disapproved of for being part of an affirmatively-secular organization.This creates a dilemma where some atheists may feel stymied from joining general community groups, and at the same time fear derogation for seeking out fellowship and support with fellow secular individuals.Indeed, this bind was explicitly highlighted as a qualitative theme by some participants, who felt cut off from a society they saw as Christian-centric and unable to locate sources of secular support.
Denial of opportunities, goods, and services was much less common; 16% of participants reported at least one such experience.In contrast, 32% of gays and lesbians in a nationally representative sample reported experiencing similar forms of discrimination on the basis of their sexual orientation (Mays & Cochran, 2001).This suggests that institutional and structural discrimination may be less of a concern for most atheists than other marginalized groups, such as racial minorities, LGB's, or people with disabilities (Pincus, 1996).One factor that may contribute to this lower rate of reported discrimination is that one's atheist identity may be less publicly identifiable by external cues than one's racial identity, for example.
Along similar lines, 14% of the participants reported experiencing an anti-atheist hate crime, such as property damage or physical assault.This figure is lower than what has previously been reported in non-nationally representative studies of LGBs (28%; Herek, Gillis, & Cogan, 1999) and Muslims (54%; Rippy & Newman, 2006).Thus, the large majority of atheists report not experiencing the most severe forms of discrimination.

Theoretical Constructs Associated with Perceived Discrimination
The secondary aim of our study was to test three hypotheses regarding the relationship between perceived discrimination and (a) identification, (b) outness, and (c) family strictness.
First, atheists who more strongly identify with their secular identity tended to report more perceived discrimination.However, in line with extant studies focused on other marginalized minorities, the strength of this relationship was modest.These results support the extension of the rejectionidentification, social-identity, and prejudice-distribution models to atheists, and invite future research to determine the relative explanatory power of these three theories for this population (Branscombe, Schmitt, & Harvey, 1999;Kaiser & Pratt-Hyatt, 2009;Major, Quinton, & Schmader, 2003).On one hand, atheists, like all humans, desire to feel accepted and protect their self-esteem, and so may be more likely to identify with atheists when perceiving more prevalent discrimination.On the other hand, those with stronger atheist identities may attribute negative ambiguous events to others' antiatheist prejudice.However, it is also possible that the documented widespread distrust of atheists (Edgell, Gerteis, & Hartmann, 2006) translates into discriminatory behaviors that are directed particularly at those who are most dedicated and vocal about their atheist identity.Future research is needed to test the respective applicability of these theories to account for the atheist identity-perceived discrimination relationship identified in the current study.
Second, atheists who were out about their atheism to more individuals tended to report experi-encing greater discrimination (Corrigan & Watson, 2002;Herek & Berrill, 1992).As with prior research on marginalized minorities, the strength of this relationship was modest.While non-directed discrimination such as hearing anti-atheist comments in the media might confront out and closeted atheists with similar frequency, atheists who stay silent regarding their identity might be less likely to be targeted on account of that identity.Alternatively, out atheists may more readily anticipate being the victims of discrimination and interpret ambiguous events as discriminatory.Either way, there are potential benefits from coming out, including increased self-esteem (Jordan & Deluty, 1998), decreased stress from hiding an important part of one's identity (Rosario et al., 2001), and improved job satisfaction (Day & Schoenrade, 1997).Future research should examine the precursors, process, and effects associated with coming out as an atheist, and how perceived discrimination influences this process (Smith, 2011).Third, atheists who grew up in homes where religious expectations were stricter tended to report experiencing more frequent ostracism by their family.This relationship was moderated by outness: the more out participants were, the stronger the relationship between strictness and reported ostracism.These effects were modest in strength.Just as LGB individuals raised in more devout households tend to experience greater stress over their stigmatized identity (Newman & Muzzonigro, 1993), it may be the case that atheists raised in stricter households are at greater risk for experiencing stress due to conflict over their atheist identity, especially if they are out to others about their identity.Alternatively, atheists who grew up in stricter households may have held more resentment towards their parents, and thus been more prone to feeling ostracized by them.Longitudinal research is needed to examine the impact of familial environment on young atheists' well-being, and how this impacts atheist identity development over time.In addition, because the primary purpose of this study was to explore the forms of perceived anti-atheist discrimination, we purposely chose to limit the extent of our hypothesis testing.We recommend future researchers conduct investigations dedicated to examining the relationship between specific forms of discrimination and theoretical constructs of interest.

Limitations
This study was not without its limitations.First, while Internet-based recruiting may be practical given the relative scarcity of atheists in the general population and their possible reticence to disclose their atheist identity, and that online studies yield results that are similar or better in terms of measures' psychometric properties (Birnbaum, 2004;Gosling et al., 2004), our subject pool was limited to those who owned a computer and visited the website where the study was advertised.Therefore, the findings derived from this non-representative sample of self-identified atheists cannot be assumed to generalize to all atheists living in the United States.In particular, the relative frequencies reported for each of the 29 forms of discrimination may look different among a nationally representative sample of atheists.However, the primary aim of this study was to explore the reported content and frequency of this discrimination, not to claim national representativeness of our sample.Furthermore, only those individuals who self-identified as atheists were asked to participate in the study.Research suggests that the majority of individuals in the U.S. who are atheists "by belief" do not self-identity as atheists (Cragun et al., 2012).Thus, these demographically significant, non-identified atheists were not captured in this sample.Broader population sampling will be necessary to capture the experiences of non-identified atheists.
Second, the majority of the sample was white, educated, and middle to upper class, which largely reflects the profile of atheists in society (Kosmin & Keysar, 2006), but may fail to capture the reality of atheists from diverse backgrounds.The intersection of multiple identities likely plays a significant role in the nature and frequency of perceived discrimination experienced by atheists (Deaux & Perkins, 2001).Therefore, future research studies should examine the prevalence and qualities of discrimination within specific groups of atheists (e.g., atheists of color, gay and bisexual atheists, working-class atheists, atheists with disabilities).
Third, our data collection relied on participant self-report, the limits of which are well known, including the potential for socially desirable responding (Lucas & Baird, 2006).However, the use of anonymous online participation protocols has been found to effectively reduce socially desirable responding (Booth-Kewley, Larsen, & Miyoshi, 2007;McBurney, 1994).Furthermore, like most research on minority group members, the purpose of our study was not to objectively measure discrimination incidents in the real world, but to assess the nature of discrimination perceived by atheists.Field and experimental research is needed to address whether discrimination against atheists is as prevalent as the participants in this study perceived it to be.
Fourth, due to the paucity of literature focused on atheists, it was necessary to adapt several measures for use with this novel population.Therefore, conclusions from our study are limited by the choice of instruments used to operationalize perceived discrimination, atheist identity, outness, and strictness of familial religious expectations.In particular, while anti-atheist prejudice may operate along the same lines as racism or homophobia, there are as of yet no psychometrically sound measures of discrimination specifically tailored to atheists.Therefore, future research efforts should incorporate the development and validation of standardized assessments of these constructs into the study design.
Fifth, as mentioned previously, the "negative focus" of several of the survey items may have led participants to recall more negative experiences than they otherwise would have.While the instructions neutrally referred to the 29 discrimination experiences as "situations", the content of all items was clearly negative, and astute participants may have guessed that the measure was supposed to be assessing anti-atheist discrimination.Also, we included items assessing milder forms of discrimination, such as "Experiencing pressure to say 'under God' during the U.S. Pledge of Allegiance."There was no guarantee that all participants would agree, had they been asked, that such an event constitutes discrimination per se.In fact, simply including such an event in the item list could have implicitly encouraged participants to start perceiving that event as discrimination, where they may not have previously.However, both of these limitations are shared with other studies which have assessed perceived discrimination among minority group individuals.
Sixth, some items included the phrase "against my will" (e.g., Being expected to participate in religious prayers against my will).This phrase does not allow the participant to clarify whether this expectation went against their private or expressed will, thereby creating ambiguity in the interpretation of this data.In particular, it could be argued that in the former case, discrimination is not actually occurring, as the victim had not made known his or her preference.However, Sue et al. (2008) would argue that acting on assumptions derived from statistical probability (e.g., "Most people in the U.S. are religious, so it's likely that any given person I interact with is interested in participating in religious prayers") can lead one to negate the thoughts, feelings, and experiences of a minority group member (i.e., commit a microinvalidation).
Seventh, the current study did not assess positive interpersonal experiences that atheists encounter.Such research will facilitate understanding of how to create an inclusive environment for atheists.Thus, future research should seek to account for these positive experiences.

Conclusion
This study provides the first mixed-method empirical examination of the nature and frequency of various forms of anti-atheist discrimination experienced by a sample of U.S. atheists, and the association between perceived discrimination and theoretically-related factors.Atheists in this sample reported experiencing a wide variety of types of discrimination.Interestingly, specific types of discriminatory events described by the participants appear to mirror events reported by other marginalized minorities, including hate crimes (Herek, Gillis, & Cogan, 1999), denial of services (Kessler, Mickelson, & Williams, 1999), ostracism (Garnets, Herek, & Levy, 1990), and slander (Rippy & Newman, 2006).Additional qualitative themes regarding discrimination stress raised by participants included (a) assumed religiosity, (b) lack of a secular support structure, (c) lack of church and state separation, (d) negative effects on family, (e) unreciprocated tolerance, and (f) anticipatory stress.These findings extend prior research on prejudice and discrimination towards atheists (e.g., Cragun et al., 2012;Edgell et al., 2006), and provide a more systematic understanding of the nature of perceived anti-atheist discrimination.These results offer a starting point for the development of studies which could validly assess the prevalence and impact of perceived discrimination on the physical and psychological well-being of atheists.Such studies would help social scientists determine whether discrimination-related stress accounts for the proposed health disparity between atheists and R/S individuals living in the U.S., which in turn has substantial implications for the ongoing R/S-health debate (Hwang et al., 2011).

Table 1 .
Descriptives, T-Tests, and ANOVAs for Demographic Variables Note.Discrimination = overall perceived discrimination.P-value has been Bonferonni-adjusted for multiple comparisons (p < .003).Demographic data displayed in the table reflects sample sizes after the removal of outliers.aMay not total 100% due to non-response.

Table 2 .
Means and Standard Deviations for Age and Primary Analysis Variables

Table 3 .
Types and Frequency of Perceived Discrimination a May not total 100% due to non-response.