Unpacking the Varieties of Nonbelief within US-Based Marginalized Populations

Retaining Religion and Spirituality

Some nonbelievers are culturally religious and–despite disaffiliating from organized religion–still retain some religious practices associated with their former beliefs or feel culturally connected to a faith system or community (Mrdjenovich 2019). Indeed, many nonbelievers, particularly soft secularists (e.g., spiritual but not religious people, agnostics who are tolerant of religious traditions; Kosmin 2007), still incorporate religious beliefs into their lives. Such nonbelievers report finding pleasure in attending religious services that allow them to stay connected with others and avoid conflict with religious family members (Mrdjenovich 2019). Additionally, nonbelieving individuals who belong to an ethnic religion (e.g., Judaism) may participate in religious practices to honor their culture and ancestors (Benatar 2006).

Nonreligious people also sometimes retain or develop new spiritual beliefs. For example, nonbelievers may describe their spirituality meaningful experiences with the natural world, life/death experiences, and/or deep connection with other people (Wixwat & Saucier 2021). Some nonbelievers hold New Age beliefs (reincarnation, astrology, psychics, and the presence of spiritual energy in physical objects), with 22% of atheists, 56% of agnostics, and 77% of spiritual but not religious people believing in at least one (Lipka & Gecewicz 2017). Thus, although many nonbelievers doubt the existence of a personal god, this is not always akin to the complete elimination of religious/spiritual belief and practice.

Valuing Justice and Diversity

Investment in and valuing social justice and diversity has been widely noted in studies with secular populations, as common among nonbelievers and a catalyst for religious exiting (Van Tongeren et al. 2025a). For example, nonbelievers are more likely to be supportive of gender equality and gay rights compared to religious people (Zuckerman 2009); formerly religious nonbelievers, in particular, become more politically liberal over time as compared to religious and never religious people (Van Tongeren et al. 2025b). Some nonbelievers belong to organizations designed to challenge religion and theism or support secular values (e.g., American Atheists, American Humanist Association). A primary function of these groups is to engage in social justice organizing, which can help individuals who do not belong to religious organizations still actively help their communities. In contrast, nonmembers indicated that joining a collective group was a low priority, and many noted that they did not feel as though their nonbelief was a salient aspect of their identity (Langston et al. 2015). Low-income and working-class nonbelievers may, then, be less likely to engage in secular community as they report low salience of their nonbelief (Abbott et al. 2022). Therefore, nonbelievers appear divided in their involvement in secular community organizations, possibly due in part to their other social locations, despite their commitment to promoting justice.

Minority Stress Among Nonbelievers

Nonbelief is a concealable stigmatized identity, such that nonbelievers can often choose to disclose or conceal their nonbelief when interacting with others due to social stigma (Abbott & Mollen 2018). Due to early colonists of the US seeking religious freedom, religious values have been sewn into the fabric of US society, with belief in god, specifically Christianity, becoming linked with patriotism over time. For example, the US Pledge of Allegiance, which symbolizes loyalty to the country, states that we are “one nation under god.” Given that theism is often presumed to be the standard or default worldview among people in the US and atheists are perceived as immoral, nonreligious people, particularly those with other marginalization (e.g., women, POC) or living in religious parts of the country (i.e., the South) may be pressured to conceal their identity to conform and avoid minority stress and other harmful consequences of disclosure (Frost et al. 2023; Mackey et al. 2020). In a national survey, 41% of self-identified atheists reported experiencing discrimination due to their nonbelief. Previous research has shown that atheists report experiencing discrimination (Brewster et al. 2020) that impacts various domains of life, including but not limited to slander, social ostracism, and hate crimes (Hammer et al. 2012). In addition to interpersonal discrimination, nonreligious people may lack sufficient political representation and experience structural oppression rooted in Christian nationalism (Campbell et al. 2025).

Existing Nonreligious Typologies

Extant studies have made efforts to explore and document the intra-group diversity among nonbelievers by generating discrete types of nonbelief irrespective of how nonbelievers self-identify. Lindeman et al. (2019) utilized latent class analysis to identify types of nonbelievers in European countries and discovered three groups: (1) analytic atheists, (2) spiritual but not religious, and (3) uncertain nonbelievers. Analytic atheists were more likely to be men, largely held no supernatural beliefs, and scored higher in cognitive reflection than the other two groups. Spiritual but not religious individuals believed in supernatural phenomena, except a personal God, viewed religion as less harmful, and relied more on intuitive thinking in their everyday lives than the other two groups. Uncertain nonbelievers exhibited hesitancy about their nonbelief and held ambivalence in their attitudes toward religion.

In another cross-cultural study (McLaughlin et al. 2022), latent profile analysis was used to form profiles of individuals who had left religion, resulting in two distinct profiles: (1) discontinuing religious dones and (2) still practicing religious dones. Still practicing religious dones engaged in more religious and spiritual behaviors but were more likely to experience mental health concerns compared to the discontinuing group. This finding is similar to results from a study with US-based college students in which “dones” had lower levels of interpersonal, meaning, emotionally-oriented, divine, and doubt struggles than those who were “disengaged but religious” or “disengaged but spiritual” (Exline et al. 2022).

Pew Research Center (2018) organized the nonreligious into two categories: Solidly Secular, low in religiousness and spirituality, and Religion Resisters, who retain belief in a higher power or spiritual force but find organized religion harmful to society. In a grounded theory study, Silver et al. (2014) observed six types of nonbelief: academic atheists, activist atheists/agnostics, seeker agnostics, antitheists, non-theists, and ritual atheists. These studies suggest that nonbelievers are diverse individuals with various practices, (non)beliefs, values, and worldviews. However, samples used in these studies do not reflect the growing proportion of nonbelievers who are POC and/or LGBTQ+ in the US.

Participants

All 888 participants were from the US and 18 to 85 years old (M = 40.14, SD = 15.68, Mdn = 36). To participate in the study, all participants affirmed that they were nonbelievers; however, approximately 49% reported that their “preferred” label was atheist, 21% were agnostic, and 30% were another descriptor (e.g., satanist, freethinker, humanist, Unitarian Universalist). In terms of gender, approximately 39% of the sample identified as women, 38% as men, and 22% as another gender (e.g., genderfluid, nonbinary). Regarding race/ethnicity, approximately 57% identified as White, 8% as Hispanic/Latino/a, 14% as African American/Black, 11% as Asian American/Pacific Islander, 4% as Native American/Indigenous American, and 4% another self-specified race (i.e., Arab, South Asian). For sexual orientation, approximately 20% identified as heterosexual, 29% as bisexual+, 30% as gay/lesbian, and 15% as another self-specified identity (i.e., asexual, queer). In terms of education, approximately 32% of participants completed a post-graduate or professional degree, 29% completed college, 24% completed some college, 10% completed some post-graduate work, 4% had only graduated high school, and < 1% had not graduated high school. For employment status, approximately 49% were employed full-time, 25% not employed, and 13% employed part-time. Regarding social class, 44% identified as middle class, 27% as working class, 18% as upper-middle class, 7% as lower class, and 4% as upper class. In terms of geographic region, approximately 40% identified as living in suburban, 44% in urban, or 16% in rural areas.

Measures

Importance of Religiosity and Spirituality. Developed by Cragun et al. (2015), the NonReligious-NonSpiritual Scale (NRNSS) short-form has six items, arranged in two subsections designed to assess spectrums of religiosity and spirituality, assessing both the presence and absence of these factors. The former indicates an affiliation to institutional religiousness (i.e., “When faced with challenges in my life, I look to religion for support”), whereas the latter to spirituality (i.e., “I have a spirit/essence beyond my physical body”). The NRNSS is a 5-point Likert-type scale ranging from 1 (Strongly Disagree) to 5 (Strongly Agree), wherein scores are averaged and high scores reveal strong R/S, whereas low scores represent strong nonbelief. In the original development study, the authors reported a strong internal consistency reliability of .94 (Cragun et al. 2015). In the present study, NRNSS items demonstrated good internal consistency, with a Cronbach’s α of .83 (religious) and .89 (spiritual).

Outness and Concealment of Nonbelief. In order to assess levels of disclosure and concealment of atheist identity, outness was examined via the Nebraska Outness Scale (NOS; Meidlinger & Hope 2014); this measure was originally developed for use with an outness measure originally LGB populations but modified successfully for use with nonbelieving populations (Abbott & Mollen 2018). The NOS is a 10-item measure with two subscales: concealment (NOS-C) and disclosure (NOS-D), wherein participants indicate the percentage of people who are aware of their atheist identity and the frequency with which they avoid indicating their atheist identity in five contexts: Immediate Family, Extended Family, People You Socialise With, People at Work/School, and Strangers. Items are scored on an 11-point Likert-type Scale, with higher scores indicating more disclosure and concealment. The NOS demonstrated positive correlations with other measures of outness and strong internal consistency reliability across diverse atheist samples (Abbott & Mollen 2018; Abbott et al. 2020). In the current study, Cronbach’s α of the items was .72 (outness) and .79 (concealment).

New Age Beliefs. Developed by Granqvist and Hagekull (2001), the New Age Orientation Scale (NAOS) was used to assess participants’ inclinations toward New Age beliefs and practices. The NAOS is a 22-item self-report questionnaire designed to evaluate engagement with and openness to New Age spirituality, including beliefs in astrology, alternative healing practices, and spiritual experiences. Participants rated each item on a 6-point Likert scale ranging from 1 (Strongly Disagree) to 6 (Strongly Agree), with higher scores indicating a stronger orientation toward New Age beliefs. An example item is “The position of the stars at birth affects how one will live one’s life or how one’s personality develops. Previous research has established the construct validity of the NAOS, showing positive associations with new age practices and magical thinking (Farias et al. 2005). In the present study, NAOS items demonstrated good internal consistency, with a Cronbach’s α of .96.

Illusory Beliefs. We used the 24-item Illusory Beliefs Inventory (IBI; Kingdon et al. 2012) to measure magical thinking. Participants rate each item with a 5-point Likert scale, 1 (Strongly Disagree) to 5 (Strongly Agree), and appropriate items are reverse-scored. Item responses are averaged such that higher scores indicate a stronger endorsement of magical thinking and thought/action fusion. Sample items include: “I sometimes perform special rituals for protection” or “Magical forces have impacted my life.” In prior research, the IBI has shown strong internal consistency and convergent validity (Fite & Magee 2022; Kingdon et al. 2012). Cronbach’s α for IBI items with the current sample was .93.

Christian Privilege. We used the Christian Hegemony Awareness subscale to indicate participants’ perceptions of how Christianity is privileged and a dominant force in society. This is a subscale of the Privilege and Oppression Inventory (Hays et al. 2007), a measure that assesses awareness of privilege and oppression across various dimensions. The 8-item subscale used a 6-point Likert scale ranging from 1 (strongly disagree) to 6 (strongly agree), with lower scores indicating less awareness. The subscale is correlated strongly with other indicators of oppression awareness (Hays et al. 2007). An example item is “Christians hold a lot of power because this country is based on their views.” In the present study, the items yielded a Cronbach’s α of .90.

Stigma Consciousness. Nonbelief stigma consciousness was assessed via the Stigma Consciousness Questionnaire (SCQ; Pinel, 1999), a 10-item Likert scale ranging from 1 (disagree strongly) to 7 (agree strongly) that measures awareness and personal salience of social stigma against one’s group. Following prior studies with atheist populations (i.e., Brewster et al. 2016), SCQ items were adapted for use with nonbelieving individuals (e.g., “Most heterosexuals have a problem with viewing homosexuals as equals” was modified to “Most people have a problem with viewing nonbelievers as equals”). Higher scores indicate greater perceived awareness of stigmatization toward nonbelief. In prior research with atheists, SCQ items yielded a Cronbach’s α of .73 and were shown to correlate positively with perceived experiences of anti-atheist discrimination (Brewster et al. 2016). Cronbach’s α for SCQ items with the current sample was .80.

Loneliness. Loneliness was measured via Version 3 of the UCLA loneliness scale (UCLA3; Russell, 1996) and assesses the degree of loneliness and social isolation an individual might feel. A sample item is “How often do you feel left out?”. Items are rated on a Likert scale from 1 (never) to 4 (often), with higher ratings indicating more loneliness. The UCLA3 yielded a Cronbach’s α of .95 in a sample of primarily white atheists (Brewster et al. 2016), and scores from the UCLA3 have been shown to be positively related to indicators of psychological distress (Westefeld et al. 2001). Cronbach’s α for these items in the present sample was .84.

Internalized Negativity. Following prior research with atheist participants (Brewster et al. 2020), the three-item Internalized Homonegativity and three-item Difficult Process subscales of the Lesbian, Gay, and Bisexual Identity Scale (LGBIS; Mohr & Fassinger 2000) were combined and modified to assess respondents’ negative views and feelings about themselves as nonbelievers. Participants used a 7-point Likert scale ranging from 1 (disagree strongly) to 7 (agree strongly), and a sample item is “Admitting to myself that I am a nonbeliever has been a very painful process.” All items were averaged, with higher scores indicating greater internalized negativity of nonbelief. In a sample with atheist participants, Cronbach’s α for the items was .78.

Social Justice Intentions. Social justice commitment (SJC), or an individual’s choice goals or intentions (Lent & Brown 2006) to engage in social justice advocacy in the future, was assessed by the SJC subscale of the Social Issues Questionnaire (SIQ; Miller et al. 2009). Participants rated SJC items (n = 4) on a 10-point Likert-type scale from 0 (strongly disagree) to 9 (strongly agree), wherein higher scores indicated a stronger commitment to social justice engagement in the future. A sample item is “I have a plan of action for ways I will remain or become involved in social justice activities over the next year.” SJC scores have produced high internal consistency reliabilities and theory-consistent relations with social justice self-efficacy, outcome expectations, and interest (Miller et al. 2007). The Cronbach’s α for SJC items in the present study was .90.

Distress. We used the short form of the Kessler Psychological Distress Scale (K6; Kessler et al. 2002) to measure psychological distress. Its six items assess how frequently in the past 30 days participants experienced symptoms of distress (i.e., felt nervous, hopeless, or restless) and are rated on a 5-point Likert-type rating scale ranging from 1 (none of the time) to 5 (all of the time). When responses are added, the total score ranges from 6 to 30, whereby higher scores represent higher levels of psychological distress. The measure is reported to have excellent internal consistency reliability (Kessler et al. 2002). In the present study, K6 items yielded a Cronbach’s α of .88.

Life Satisfaction. A participant’s reported satisfaction with life was assessed with the five-item Satisfaction with Life Scale (Diener et al., 1985). This 7-point Likert-type scale, wherein 1 (strongly disagree) to 7 (strongly agree), was used to indicate satisfaction. A sample item is “In most ways my life is close to ideal” and responses are averaged with higher scores indicating a greater degree of satisfaction with life. Prior research has provided support for high test-retest reliability and moderate to high convergent validity with other scales assessing well-being (Diener et al., 1985; Pavot et al., 1991). Furthermore, high internal consistency reliability for the items was reported in a diverse sample of religious, spiritual, and atheist participants (Brewster et al. 2021). Cronbach’s α for items in the current sample was .87.

Single items

Nonbelief self-assessment. Although all participants identified as nonbelievers (see inclusion criteria), one item from the General Social Survey (GSS; Davis et al. 2019) was used to help describe the type of nonbelief our participants endorsed: I don’t believe in God (atheist); I don’t know whether there is a God and I don’t believe there is any way to find out (agnostic); I don’t believe in a personal God, but I do believe in a Higher Power of some kind (spiritual but not religious). A second item aimed at assessing past beliefs asked participants to select “What best describes your belief?” with the response options of (a) I don’t believe in God now, and I never have, or (b) I don’t believe in God now, but I used to.

Salience of nonbelief. Participants were asked to respond to a single item, “How important is your nonbelief to you?” with a slider from 0 (not at all) to 100 (extremely).

Secular community involvement. Drawing from prior research with atheist individuals (Brewster et al. 2020), participants indicated if they were ever involved with a secular community organization (i.e., Oasis, Society for Ethical Culture, Sunday Assembly, or any other group)?” and asked to report which group. These responses were dichotomized such that Yes (1) and No (0).

Personal religious practices. Building off prior research (Brewster et al. 2021) with diverse secular, religious, and spiritual participants, we asked, “How frequently do you attend religious services?” and “How often do you engage in prayer?” with response choices of 1 (never), 2 (rarely), 3 (sometimes), 4 (regularly, once a week), or 5 (multiple times a week). A prior study asked two similar items about religious behavior, both of which had good test-retest reliability (.86 and .85, respectively; Dollinger & Malmquist 2009).

Procedure

Data analyzed in the present study was collected from 2023 to 2024 as a part of an Institutional Review Board–approved study on the experiences of demographically diverse nonbelieving people in the US. Virtual communities for atheist people, blogs, and listservs (e.g., Facebook pages) were used for recruitment, as well as targeted outreach to secular communities (e.g., ExMuslims of North America, Oasis, Sunday Assembly, Society for Ethical Culture) via personal contacts and mailing lists. To participate, individuals had to (a) be legal adults in their state of residence, (b) identify as a nonbeliever who was either LGBTQ+ identified and/or a POC, and (c) live in the US. The online survey was hosted by Qualtrics. After reading the study description, individuals were asked to provide informed consent and to confirm that they met the inclusion criteria, after which they could continue to complete the survey. Initially, 1,468 entries were recorded wherein consent procedures were followed and inclusion criteria appeared to be met (i.e., nonbelievers), but 206 of these cases were subsequently removed from the data set because they were missing more than 20% of the data (excluding demographic questions; Parent 2013). Another 374 participants were removed for failing validity test items and/or bot screening methods. These data cleaning procedures resulted in a final sample of 888 participants.

Determining the Number of Acculturation Strategies

To identify distinct profiles based on the nonbelief variables of interest, we performed LPAs in Mplus 8.10 using maximum likelihood estimation with robust standard errors (Muthén & Muthén 2017). LPA identifies possible subgroupings of a sample based on the observed relations among a set of continuous and categorical variables. In the current analyses, profiles were estimated using the continuous variables salience of nonbelief, importance of religion, importance of spirituality, stigma consciousness, internalized anti-atheist prejudice, perceptions of Christian privilege, outness, concealment, illusory beliefs, New Age beliefs, religious service attendance, prayer frequency, and social justice intentions. The categorical variables–Beliefs about God (Don’t believe in God; Don’t know if God exists; Belief in a non-personal higher power; or Believe in God sometimes), Prior belief in God (Never believed in God or Used to believe in God), and secular community involvement (yes or no)—were also used to estimate profiles. Profile solutions are estimated and compared stepwise, beginning with a two-profile solution. Based on prior scholarship (Tein et al. 2013), we evaluated model fit using the Bayesian information criteria (BIC), sample-size adjusted BIC (ABIC), entropy, and the Lo- Mendell–Rubin adjusted likelihood ratio test (LM ALRT). Profiles with lower BIC and ABIC values have a better fit. Entropy indicates how distinguishable profiles within a solution are based on participant profile assignment probabilities; values closer to 1 suggest a better model fit. A significant LM ALRT indicates that the solution with k profiles yields a better fit than a solution with k-1 profiles. We also considered profile size (and, by extension, potential replicability) when evaluating solutions (Masyn 2013).

The fit statistics for the one-to six-profile solutions are presented in Table 2. The BIC and AIC values were lowest for the six-profile solution, indicating a potentially better fit. The entropy values were similar across solutions, but were lowest in the five-profile solution. The LM ALRT was significant for each solution except the five-profile solution, suggesting that the five-profile solution does not significantly improve model fit above and beyond the four-profile solution. We also observed that the size of the smallest profile decreased precipitously between the three-profile solution (n = 132, about 15% of the total sample) and the four-profile solution (n = 44, about 5% of the total sample). Given these observations and our concerns that small profiles may not replicate across samples, we decided to retain and explore the three-profile solution.

Characterizing Profiles

Figure 1 presents scores for the three profiles’ 13 continuous nonbelief variables of interest for the three profiles. All continuous nonbelief variables of interest are z-scored so that values of zero reflect the grand mean and +/–1 reflect values one standard deviation above or below the grand mean. Error bars signify the 95% CI of a variable’s mean. When an error bar does not contain zero, the variable’s value significantly differs from the grand mean at p < .05. Supplemental Table A compares the continuous nonbelief variables across the three profiles (rather than relative to the grand mean). Table 3 compares the responses of the three profiles to the three categorical nonbelief variables.

Figure 1

Relative to the grand mean, participants in Profile 1 (n = 549, 61.82%) reported significantly greater salience of nonbelief, stigma consciousness, and perceptions of Christian privilege and reported significantly lower importance of religion, importance of spirituality, internalized anti-atheist prejudice, illusory beliefs, New Age beliefs, religious service attendance, and frequency of prayer. Outness, concealment, and social justice intentions scores in Profile 1 did not differ significantly from the grand mean. In terms of the categorical variables—in comparison with Profiles 2 and 3—participants in Profile 1 were significantly more likely to indicate that they did not believe in God and were less likely to indicate that they did not know if there was a God, to believe in a higher power, or to believe in God sometimes. Participants in Profile 1 were also more likely than those in Profiles 2 and 3 to have previously believed in God. Finally, participants in Profile 1 were significantly more likely to be involved in secular communities than participants in Profile 3; Profiles 1 and 2 did not differ regarding secular involvement. In light of these patterns, Profile 1 was labeled “Adamant Nonbelievers”¹ to reflect the participants’ strong stance on their nonbelief and their active involvement in secular communities.

Relative to the grand mean, participants in Profile 2 (n = 132, 14.86%) reported significantly greater importance of religion and spirituality, internalized anti-atheist prejudice, illusory beliefs, New Age beliefs, religious service attendance, and prayer frequency. Profile 2 also reported significantly lower salience of nonbelief, stigma consciousness, perception of Christian privilege, and social justice intentions. Outness and concealment scores in Profile 2 did not differ significantly from those of the grand mean. Profile 2 was significantly less likely to not believe in God than Profile 1, but more likely to not believe in God than Profile 3. Participants in Profile 2 were more likely to indicate that they do not know if there is a God than those in Profile 1, but they did not differ in this regard from participants in Profile 3. Profile 2 was more likely than Profile 1 to believe in a higher power, but less likely to believe in a higher power than Profile 3. Profile 2 was also more likely to believe in God sometimes than Profile 1, but did not differ in this regard from Profile 3. Profile 2 was significantly less likely than Profiles 1 and 3 to have believed in God previously. Finally, Profile 2 was more likely to be involved in secular communities than Profile 3, but did not differ from Profile 1. Because of its pattern of responses, Profile 2 was labeled “Ambivalent Maybelievers.”

Relative to the grand mean, participants in Profile 3 (n = 207, 23.3%) reported significantly greater importance of spirituality, stigma consciousness, perception of Christian privilege, illusory beliefs, New Age beliefs, prayer frequency, and social justice intentions. Participants in Profile 3 also reported significantly lower salience of nonbelief and importance of religion than the grand mean. Profile 3’s internalized anti-atheist prejudice, outness, concealment, and religious service attendance scores did not differ significantly from the grand mean. Regarding the categorical nonbelief variables, Profile 3 had the lowest proportion of participants who indicated that they do not believe in God. Participants in Profile 3 were more likely than those in Profile 1 to indicate that they do not know if God exists; Profile 3 did not differ from Profile 2. Participants in Profile 3 were significantly more likely to indicate that they believed in a higher power. Participants in Profile 3 were also more likely than participants in Profile 1 to indicate that they believe in God sometimes; Profiles 3 and 2 did not differ in this regard. Profile 3 had the highest proportion of participants who had previously believed in God. Finally, participants in Profile 3 were the least likely to be involved in secular communities. Given these patterns of responses, Profile 3 was labeled “Spiritually Inclined.”

Demographic Correlates of Profile Membership

A univariate analysis of variance (ANOVA) was conducted examining the association of profile membership with participant age. The association was significant, F(2, 772) = 23.89, p < .001, η_p² = .06. Follow-up comparisons indicated that Adamant Nonbelievers (M = 42.70, SD = 16.59) were significantly older than Ambivalent Maybelievers (M = 32.00, SD = 8.41, t(587) = 6.01, p < .001, d = 0.81) and Spiritually Inclined participants (M = 37.08, SD = 13.08, t(682) = 4.12, p < .001, d = 0.37). Spiritually Inclined participants were also significantly older than Ambivalent Maybelievers (t(275) = 3.23, p = .001, d = 0.44).

Table 4 presents the associations of profile membership with the categorical demographic variables gender, gender identity, race/ethnicity, sexual identity, education, employment status, social class, annual household income, and urbanity. The associations of profile membership with each of these demographic variables were significant. Notably, however, 33% of expected counts for race/ethnicity and 22% for education were fewer than five, so these results must be interpreted cautiously.

Concerning gender, participants in the Spiritually Inclined profile were significantly less likely than the other two profiles to identify as men and significantly more likely to identify as nonbinary. In terms of gender identity, Ambivalent Maybelievers had a significantly higher proportion of cisgender participants—and a significantly lower proportion of transgender or nonbinary participants—than the Adamant Nonbelievers and Spiritually Inclined profiles. Regarding race/ethnicity, Ambivalent Maybelievers were significantly more likely than participants in the two other profiles to identify as Black or African American. Ambivalent Maybelievers were also significantly less likely than participants in the other profiles to identify as White or European American.

In terms of sexual identity, Ambivalent Maybelievers were significantly less likely to identify as asexual than Adamant Nonbelievers, but they did not differ from the Spiritually Inclined profile. Ambivalent Maybelievers were significantly less likely to identify as bisexual or pansexual than Adamant Nonbelievers—who in turn were significantly less likely to identify as bisexual than the Spiritually Inclined profile. Adamant Nonbelievers were significantly more likely to identify as gay or lesbian than Ambivalent Maybelievers. The Spiritually Inclined profile had a significantly lower proportion of heterosexual participants than the Adamant Nonbelievers profile, which, in turn, had a significantly lower proportion of heterosexual participants than the Ambivalent Nonbelievers profile.

Regarding education, Ambivalent Maybelievers were more likely than the other two profiles to report having less than a high school education. Ambivalent Maybelievers were also more likely than Adamant Nonbelievers to report that their highest level of education was a high school diploma. Ambivalent Maybelievers were significantly less likely than the other two profiles to report having a graduate degree. In terms of employment status, Ambivalent Maybelievers were significantly less likely than the other two profiles to report being unemployed. Ambivalent Maybelievers were also significantly more likely than the other two profiles to report being employed full-time. With regard to social class, Adamant Nonbelievers were significantly less likely than the other two profiles to identify as working class. In terms of annual household income, the Adamant Nonbelievers profile was significantly more likely than the Spiritually Inclined profile to report earning $160,001 to $190,000 per year. Adamant Nonbelievers were also significantly more likely than Ambivalent Maybelievers to report earning more than $190,000 per year. Regarding urbanity, Ambivalent Maybelievers were significantly more likely than participants in the two other profiles to report living in rural and urban areas. Ambivalent Maybelievers were also significantly less likely than participants in the two other profiles to report living in suburban areas.

The Associations of Profile Membership with Mental Health

To determine the association of profile membership with distress, loneliness, and life satisfaction, a multivariate analysis of variance was conducted with the mental health variables as dependent variables and profile membership as the independent variable. The multivariate effect of the profile was significant, Pillai’s Trace, λ = .09, F(6, 1,646) = 13.07, p < .001, η_p² = .05. Table 5 presents the results of univariate ANOVAs for distress, loneliness, and life satisfaction. There were significant univariate associations of profile membership with distress and life satisfaction. However, the univariate association of profile membership with loneliness was nonsignificant. Follow-up Sidak-adjusted pairwise comparisons indicated that Adamant Nonbelievers reported significantly lower distress than participants in the Spiritually Inclined profile (Cohen’s d = –0.28), who in turn reported significantly lower distress than Ambivalent Nonbelievers (Cohen’s d = –0.41). However, Sidak-adjusted pairwise comparisons indicated no significant differences in life satisfaction across profiles.

Integration with Previous Literature

Our results revealed three distinct profiles of nonbelievers, a finding that has significant implications for understanding secular Americans’ traits, demographics, and psychological well-being. The largest group (61.8%), Adamant Nonbelievers, were more likely not to believe in god(s) but likely to have previously believed. Such a finding suggests that these participants may have had a religious exit or deconversion, processes that are known to be difficult and distressing for many (Brewster 2014; Streib 2021). We called the second largest group (23.3%) the Spiritually Inclined, as they leaned toward agnosticism and were more likely to retain belief in the supernatural or a higher power and sometimes believed in god(s). This group may still find value in or appreciate the spiritual realm or more New Age practices (i.e., astrology, tarot). The Ambivalent Maybelievers comprised 14.9% of our sample and were markedly less likely to have ever believed in god(s); however, they were more open to sometimes believing in god(s). This type of nonbeliever seems to fit prior literature on agnosticism or quest religiosity (Arrowood et al. 2021)—in that they appear to remain resistant to a clear-cut belief system and “open to possibility.”

As mentioned previously, the results of our LPA were very similar to those from a European sample (Lindeman et al. 2019). Their analytic atheists were described as textbook atheists, scoring high in cognitive reflection and analytic thinking while relying less on intuition, which shares similar traits to our Adamant Nonbelievers. Indeed, compared to our other profiles, Adamant Nonbelievers reported the lowest frequency of prayer and lowest levels of R/S importance, new age beliefs, and magical thinking. Furthermore, Lindeman et al.’s uncertain nonbelievers did not believe in god, but expressed more doubt than their other groups, similar to our Ambivalent Maybelievers. Lastly, their spiritual but not religious profile believed in all other supernatural phenomena except God, and they were less negative toward religion, sharing many characteristics with our Spiritually Inclined participants.

There were also demographic trends that emerged in our data with some support in prior research. For example, akin to Lindeman and colleagues’ (2019) analytic atheists, Adamant Nonbelievers were more likely to be White, male, older, and highly educated. Adamant Nonbelievers in our study reflected the demographics—and associated privileges—often captured in sociological and psychological studies of atheists (Zuckerman 2009). This is cogent given that they most closely align with the academic definition of atheism or the absence of belief in god(s) (Bullivant & Ruse 2013). Indeed, these demographic characteristics with their associated power offer the most protection from the dangers of resisting Christian norms and Christian nationalism in the US (Whitehead & Perry 2020). It is notable, however, that this profile also emerged in our sample, given the overrepresentation of POC and LGBTQ+ nonbelievers and our US-based sample; thus, a focus on marginality did not necessarily shift the ‘typologies’ of atheism unearthed across both studies.

Adamant Nonbelievers, who were the most likely to be involved in secular communities and find nonbelief to be a salient aspect of their identity, share traits with a group that was once described as “strong atheism” or actively rejecting religion (Dawkins 2006). Some qualitative research has indicated that people from marginalized racial/ethnic communities may feel less empowered to be outspoken nonbelievers or active members of secular groups due to threats or stigma from within their racial/ethnic group (Baker 2020). Likewise, POC may not wish to align themselves with atheists, who are predominantly White, and risk encountering or inadvertently enabling the perpetuation of the tenets of White supremacy.

Unsurprisingly, considering atheism was a very salient part of their identity, Adamant Nonbelievers had the highest level of nonbelief stigma consciousness and the greatest awareness of Christian hegemony/privilege. It could be that their connection to a secular community provided education about these in-group/out-group dynamics. That being said, they did not internalize societal negative attitudes about nonbelievers; of the three profiles, Adamant Nonbelievers reported the lowest level of distress. Their relatively more dominant cultural positionalities (e.g., more likely to be White and with a higher SES) may provide some protection against the potential harm of anti-atheist stigma and other forms of social marginalization (e.g., racism, classism). However, the certainty of their nonbelief may also contribute to their relative health. Other studies of nonbelief and psychological well-being studies have demonstrated that strong believers and atheists report higher well-being and lower distress than agnostics, religious questers, and others who are less certain of their views (Arrowood et al. 2021; Exline et al. 2022; Galen 2009; Galen 2015; Galen & Kloet 2011; Speed & Hwang 2019; Uzarevic & Coleman 2021), a finding replicated in our sample.

Prior studies (i.e., McLaughlin et al. 2022) have found that “religious dones” who report that they are still practicing elements of their prior religion have more mental health concerns and lower self-control. Akin to this finding, our Ambivalent Maybelievers had higher levels of religious service attendance and prayer, greater perceived importance of R/S, and lower salience of their nonbelief. Similar to prior studies (i.e., Galen & Kloet 2011), these more “fuzzy” nonbelievers had the highest level of distress compared to the other profiles. However, this trend may be due to this profile’s relative belief uncertainty or other demographic characteristics that may contribute to heightened stress; indeed, Ambivalent Maybelievers were more likely to be younger, POC, working class, and less likely to live in the suburbs. Prior studies support that distress is correlated negatively with age (Kalin 2021) and that financial precarity is strongly related to psychological distress (Ettman et al. 2023).

Beyond these factors, experiences of racism linked to being a POC in the US can also relate to distress symptoms (Hankerson et al. 2022). Ambivalent Maybelievers may continue to have greater ties to religion and spiritual practices than the other nonbeliever profiles to maintain connection to racial/ethnic communities; for some POC, religion offers a bulwark against White Supremacy in the US (Abbott et al. 2020; Brewster 2014; Baker 2020). Lower levels of nonbelief salience have also been replicated amongst working-class and low-income atheists in prior studies (Abbott et al. 2022). Like most cross-sectional research, the causality of this association–‘uncertain’ nonbelief to psychological distress versus external stressors to distress to ‘uncertain’ nonbelief–is unclear and warrants future study.

By contrast, our Spiritually Inclined participants were more often women or nonbinary genders and members of a sexual minority group. Prior research suggests that women are more likely to be spiritual than men (Corcoran et al. 2025; Lindeman et al. 2019). This profile, too, was more likely to have previously been a believer. It may be that a tendency toward spirituality is linked to seeking a replacement for prior religious belief/practice (McLaughlin et al. 2022) or a reflection of prior gendered expectations to serve as a “moral compass” within their families or communities (Mahlamäki 2012; Trzebiatowska 2018; 2019). Of course, we do not mean to suggest that personal spirituality is entirely attributable to former religious indoctrination; it likely has many other benefits, including living more congruently with values (Wixwat & Saucier 2021).

The interplay between gender and sexual orientation for the Spiritually Inclined is particularly interesting because this profile had the highest levels of social justice intentions (i.e., plans to engage in collective action or activism). At a time in the US when reproductive rights and LGBTQ+ community protections are under fire, working to support a mission greater than oneself may feel like a spiritual calling for some. Indeed, this group also had high levels of stigma consciousness and awareness of Christian hegemony/privilege. The Spiritually Inclined profile was highest on endorsement of New Age beliefs, magical thinking, and prayer. Some research suggests that divinatory practices and other metaphysical approaches can serve as methods of coping with stressors (Lowrie et al. 2017). As such, disentangling demographic characteristics (i.e., gender), sociopolitical climate (i.e., state-backed misogyny), and more spiritual flavors of nonbelief (i.e., pulling tarot to gain a sense of control in an uncertain world) is a worthy challenge for secularism researchers to continue to tackle. It would also be fruitful for future researchers to measure the frequency of identity-related discrimination experiences (i.e., racist or heterosexist events) to see how these map onto the existing profiles.

Implications for Clinical Practice and Advocacy

Our sample was more diverse in terms of sexual orientation, gender, and race than prior typologies of nonbelievers. Consistent with the deviance/marginalization theory of secularization, we found patterns in the relationship between privilege, marginalization, and nonreligiousness in which those with less power were less inclined toward conventional religious beliefs and more inclined toward other supernatural beliefs (Corcoran et al. 2025). For example, Adamant Nonbelievers were more likely to be gay men or lesbian women, as compared to plurisexual or asexual people who face unique forms of heterosexism and ingroup discrimination from other queer people (Pollitt & Roberts 2021). All marginalized genders (women, nonbinary, and transgender) were more common among the spiritually inclined, whereas cisgender people were more common among the Ambivalent Maybelievers, perhaps suggesting cis-privilege offered space and protection to sit in the personal discomfort of, or other’s discomfort with their, uncertainty or indecision. Mental health practitioners and advocates working with and for nonbelievers can expand their consideration of the dynamics of power in the lives of nonbelievers to include the imbalance of power and resulting ingroup bias within marginalized groups.

In our findings, Ambivalent Maybelievers reported the highest levels of distress symptoms among profiles and higher loneliness than Adamant Nonbelievers. On the other hand, they reported the highest level of life satisfaction. Given prior research suggesting lower levels of certainty in one’s nonbelief (Pöhls 2021) and/or agnosticism (Galen & Kloet 2011) are associated with more negative health outcomes as compared to higher certainty or atheist identification, it is tempting for clinicians to expect more ambivalent nonbelieving clients to be at higher risk of distress. However, it is important to remember that many identity exploration processes (e.g., sexual orientation, vocation) involve periods of doubt and questioning that are emotionally and psychologically difficult but not necessarily pathological in nature. Additionally, uncertainty is not inherently distressing or negative; rather, nonbelievers often find their uncertainty a positive experience or balance positive aspects of their uncertainty with more challenging aspects like isolation. Likewise, certainty is fluid (Frost 2019), such that current Ambivalent Nonbelievers may later become Adamant Nonbelievers, or Adamant Nonbelievers may go through periods of ambivalence or uncertainty. Therefore, clinicians and organizations for nonbelievers can create a nonjudgmental space for nonbelievers to ebb and flow between or even transcend profiles.

Limitations and Future Directions

Findings from our study should be interpreted in light of a few limitations. First, our study used cross-sectional data, and, as a result, we cannot address causal or temporal topics within the data. Although we can infer some developmental patterns from items that addressed whether participants were formerly religious, having an absolute sense of a developmental trajectory is impossible. Most nonbelieving people in the US were not “raised secular” but have a narrative of leaving a particular religious group; thus, future longitudinal studies that assess how these exiting experiences evolve and shift over time are crucial. Like most prior research from the US, most of our participants came from Christian families. It is essential to explore how the development of nonbelief and/or faith exiting narratives vary for individuals raised in other religions such as Hindu, Muslim, or Jewish faiths. Qualitative research using grounded theory (Charmaz 2015) or Interpretive Phenomenological Analysis (Smith 2011) may help to unpack some of these questions, among others.

Demographically, even our intentionally diverse sample was somewhat homogenous when it came to levels of higher education (61% completed college and of these, roughly half earned a graduate degree)—however, this composition roughly parallels prior research with atheist samples (Abbott & Mollen 2018; Brewster et al. 2020; Swan & Heesacker 2012) and may be representative of nonbelieving people in the US more broadly (Zuckerman 2007). The geographic distribution of our sample was also limited, as the vast majority of our sample lived in urban or suburban areas, and only 16% reported being from rural regions. Prior scholars have called attention to the experiences of rural atheists, and some research has begun to amplify their unique stressors and strengths (Abbott et al. 2022; Abbott & Santiago 2023). That being said, findings from this study are likely not representative of populations with lower levels of education from more rural parts of the US and should be extended to such groups with caution.

Another notable, but not uncommon, limitation of our study was that participants were gathered via convenience sampling through networks that serve nonreligious and intentionally secular communities. Moreover, participants who found the call may experience nonbelief as a more salient part of their identity compared to individuals who are not connected with these resources. While we did gather information on participants’ involvement in secular communities, future researchers should also explicitly explore how a connection to online groups or other virtual resources (i.e., American Atheists newsletter) is related to identity salience.

Finally, we would be remiss not to acknowledge the potential role of measurement issues in our study. By using a dichotomous variable to assess prior involvement in secular groups, we may have missed important nuances in experience—for example, as Brewster et al. (2020) posed, “Is there an ‘I only go to church on Easter and Christmas’ equivalent for atheist group participation?” (p. 10). Future researchers should use a graded measure of frequency or ask multiple questions to capture a more nuanced portrait of secular activity across time.