Varieties of Nonreligious Experience: Expanding Understandings of Nonreligious Wellbeing

ABSTRACT

Religious commitment has long been associated with health and happiness in the United States. However, despite a period of drastic decline in religious affiliation among Americans, much less is known about how nonreligious commitments promote wellbeing. We use novel measures that capture variation among the nonreligious to assess whether aspects of nonreligious experience might replicate religion's health‐promoting features. We analyse how wellbeing among the nonreligious is impacted by nonreligious identity (e.g., atheist and agnostic), but also by nonreligious identity duration, involvement in nonreligious organisations and affective orientation to being nonreligious. Using a national survey of Americans (2020), we analyse how these different aspects of nonreligious identity and experience predict three measures of wellbeing: self‐reported health, happiness, and life satisfaction. We find that the primary factor predicting wellbeing among the nonreligious is whether they experience their nonreligion as comforting or anxiety‐producing (affective orientation), and our findings suggest that whether nonreligious people find comfort or anxiety in their nonreligion changes over time and is shaped by their participation in nonreligious organisations. We discuss the implications of our findings for scholarship on religion and wellbeing as well as for future research on variation in wellbeing among nonreligious Americans.

1. Introduction

In the search for social and contextual factors that most effectively promote wellbeing, sociologists have found religion to be an especially powerful predictor of both physical and mental health. Numerous studies link religiosity to increased wellbeing because of the social, psychological and material resources that religion can provide (George et al. 2002; Hackney and Sanders 2003; Ridge et al. 2008; Upenieks and Schafer 2020). Generally, religious people report higher rates of wellbeing than nonreligious people (Ellison 1991; Hayward et al. 2016), and leaving religion is correlated with declines in wellbeing (Scheitle and Adamczyk 2010; Upenieks and Thomas 2021). This raises the question of whether nonreligious people—including atheists, agnostics and other religiously unaffiliated people—are less healthy than religious people because they lack access to the resources religion provides.

Ongoing debates about which aspects of religion are most important for predicting wellbeing help shed light on this question. Some argue that religion's ability to provide certainty‐filled beliefs in a higher power leads to better health (Abu‐Raiya et al. 2015; Ellison 1991), whereas others argue it is the social support gained through regular participation in religious organisations that drives wellbeing (Lim and Putnam 2010; Shor and Roelfs 2013). More generally, sociologists of health have consistently shown that things such as social support, membership in identity‐affirming groups, belief certainty, and regular organisational participation are conducive to wellbeing whether they stem from being religious or not (Burke and Stets 2009; Thoits 2011). These findings suggest that the nonreligious may have access to many of the things that promote health among the religious.

Despite a recent and rapid rise in the number of nonreligious people in the United States (Pew Research Center 2024; Voas and Chaves 2016), sociological research on religion's relationship to health has largely focused on examining how variations in religious identities, beliefs, rituals and social networks impact wellbeing, whereas potential variation among the nonreligious receives little attention. Nonreligious survey respondents are typically collapsed into a reference category that is compared to outcomes for different religious groups in statistical analyses, obscuring how differences in nonreligious beliefs, practices and social networks might impact wellbeing. In recent research that does pay attention to diversity among the nonreligious, researchers have shown that variation in nonreligious identity matters for wellbeing and that those with committed nonreligious identities—such as atheists—report levels of wellbeing similar to those of religious individuals (Baker et al. 2018; Galen and Kloet 2011; May 2018). This line of research has shown that existential certainty and identity coherence can bolster wellbeing for both the religious and the nonreligious.

Beyond differences in nonreligious identity, recent research points to other aspects of nonreligious experience that can affect wellbeing. A handful of studies find that participation in organisations devoted to atheism or humanism can bolster wellbeing among the nonreligious (Brewster et al. 2020; Galen and Kloet 2011; Price and Launay 2018). However, few large‐scale surveys include measures of participation in nonreligious organisations. Similarly, qualitative studies show that although leaving religion can initially be stressful, over time many nonreligious people develop new nonreligious belief systems and social networks that can enhance wellbeing (J. Smith 2011; Zuckerman 2011; Frost 2019). However, few large‐scale surveys measure the length of time someone has been nonreligious.

In short, key aspects of religious commitment that promote health—such as identity coherence, belief certainty, and social support gained through organisational involvement—have potential secular correlates. However, because of inattention to nonreligious beliefs and experiences in large‐scale national surveys, we still know very little about how differences among the nonreligious might shape wellbeing. In this study, we expand understandings of nonreligious wellbeing by using data from a new national survey of Americans to examine how wellbeing is impacted by four distinct measures of nonreligious experience: identity (atheist, agnostic, spiritual but not religious, those who identify as ‘nothing in particular’ and those who self‐identify as simply ‘nonreligious’), community involvement (participation in explicitly secular/atheist/humanist organisations), identity duration (how long someone has been nonreligious), and affective orientation (whether someone feels comfort or anxiety regarding their nonreligion). We analyse how these different facets of nonreligious experience predict self‐reported health, happiness, and life satisfaction among the nonreligious and how the nonreligious compare to the religious across these wellbeing measures.

We find that many nonreligious people are just as happy and healthy as religious people and that many of the factors driving wellbeing among the religious are also strong predictors of wellbeing among the nonreligious. At a time when some scholars (e.g., Boateng et al. 2021) are raising concerns that religious decline is driving both the growing mental health crisis (Schnittker 2025) and an epidemic of loneliness (Yang and Santos 2023), we argue that paying more attention to diversity among the nonreligious is key to better understanding the factors that lead to increased wellbeing for both religious and nonreligious people. Our findings contribute to ongoing debates about the mechanisms through which religion impacts health, and they counter a longstanding assumption among sociologists of religion that religious decline always brings about declines in wellbeing.

2. Religion and Wellbeing

Religious involvement is often linked to good mental, physical, and social health due to the social, psychological, and material resources that religion can provide (George et al. 2002; Hackney and Sanders 2003; Ridge et al. 2008; Shor and Roelfs 2013; Yaden et al. 2022). Religions offer certainty‐filled beliefs about the meaning and purpose of life with clear directions for moral action, religious groups often act as strong social support networks reinforced through regular meetings, and religious belief systems often promote healthy lifestyles by prohibiting things such as drug and alcohol use (George et al. 2002).

However, there continues to be debate about which aspects of religion are most effective at promoting wellbeing. Some studies point to religious beliefs, such as belief in a higher power or reliance on prayer in times of need (Abu‐Raiya et al. 2015; Ellison 1991), as being beneficial for health in and of themselves. Other studies suggest that it is the more latent functions of religious involvement, such as increased social support, increased identity coherence, and the promotion of healthy lifestyles, that drive the positive relationships between religion and health (Gardner et al. 1995; Lim and Putnam 2010; Shor and Roelfs 2013). Studies also show that the health effects of religion may vary by religious tradition and by social locations such as gender, race and class (e.g., Maselko and Kubzansky 2006).

Relatedly, although some studies show that consistent participation in a religious belief system is good for health and that leaving religion is bad for health (Upenieks and Schafer 2020; Upenieks and Thomas 2021), others show that the health effects of leaving religion are significantly shaped by the type and centrality of one's religious identity (Scheitle and Adamczyk 2010). Religion can also have negative health impacts, particularly when it encourages negative coping mechanisms or conflicts with advice from medical professionals (George et al. 2002; Krause and Wulff 2004). Further, some studies find that chronic illnesses or major health crises can disrupt religious beliefs and trigger religious disaffiliation (Cluley et al. 2024; Zuckerman 2011), raising questions about the causal relationship between religious disaffiliation and health.

Thus, although past research points to multiple pathways through which religion can influence wellbeing, there are debates about which mechanisms are the strongest and whether religion's effects on health are linear or universal.

3. Varieties of Nonreligious Experience

3.1. Nonreligious Identities

A large portion of the emerging research on nonreligion and health focuses on examining differences in health outcomes among those who claim different nonreligious identities. Nonreligious people in the United States are diverse, and people use a variety of nonreligious identity labels to categorise themselves. Although atheists are by far the most vocal and visible subgroup, they are only a small minority of the larger nonreligious population (Pew Research Center 2024). There are also agnostics, humanists, sceptics, freethinkers, people who identify simply as ‘nonreligious’, people who identify as ‘nothing in particular’, and a growing group of people who consider themselves to be ‘spiritual but not religious’ (Ammerman 2013).

Although these terms are not necessarily mutually exclusive, they do often translate into distinct nonreligious identities and subcultures. For example, atheists and agnostics tend to be more politically active than other nonreligious people (Schwadel 2020), and atheists join political and social groups devoted to promoting secular values at higher rates than other nonreligious people (Langston et al. 2015). Atheists also often report higher levels of certainty around their nonreligious beliefs (Frost 2019; Galen and Kloet 2011), and they have been found to be more dogmatic and anti‐religious than agnostics and humanists (LeDrew 2015; Schnell et al. 2021). While some nonreligious people reject beliefs in gods and supernatural beings, many still maintain religious beliefs in gods or religious practices such as prayer (Pew Research Center 2024). People who identify as ‘nothing in particular’ are often still highly religious in belief and practice and those who identify as ‘spiritual but not religious’ often do so to distance themselves from organised religion while maintaining beliefs in higher powers (Ammerman 2013). Thus, the identity labels that nonreligious people use are often intentional and sometimes even political choices that correlate with different levels of belief certainty, dogmatism, political activity and embeddedness within nonreligious communities.

As a result, much of the research on the relationship between nonreligion and health examines how claiming a particular nonreligious identity label correlates with health outcomes. For example, Baker et al. (2018) find that atheists report similar, or sometimes better, mental and physical health outcomes than affiliated theists, whereas agnostics and nonaffiliated theists tend to fare worse on wellbeing measures (see also Hayward et al. 2016; Speed and Hwang 2019). Baker et al. (2018) draw on identity theory (Burke and Stets 2009) to argue that atheists have stronger identities, higher levels of belief certainty, and more resources for identity verification and social network building in secular groups than do other nonreligious people, all of which bolster wellbeing. Similarly, Galen and Kloet (2011) find that nonreligious people with high levels of certainty report better mental wellbeing than nonreligious people with lower levels of certainty (see also May 2018).

3.2. Affective Orientations

While prior research has focused on certainty as the primary mediator between nonreligious identity and wellbeing, some studies suggest that the relationship between certainty and wellbeing may be more complex. For example, Frost (2019) interviewed a sample of nonreligious Americans who experienced belief uncertainty as positive and found it a satisfying framework for understanding their nonreligious beliefs and identities (see also J. Smith and Halligan 2021). This raises the question of whether belief uncertainty is experienced in the same way by everybody and whether it always leads to anxiety or poor mental health. It may be that how nonreligious people feel about being certain or uncertain is just as important for predicting wellbeing.

We call this emotional dimension of being nonreligious an individual's ‘affective orientation’, and we argue that it is an understudied but important factor to examine in studies of nonreligion and health. For some people, nonreligion provides an affirming and comforting worldview, whereas for others being nonreligious is stressful and anxiety‐inducing (Brewster et al. 2020; Frost 2019; Zuckerman 2011). While some of this anxiety can come from belief or identity uncertainty, being nonreligious in the United States also often comes with experiences of stigmatisation and discrimination (Cragun et al. 2012), which studies show can increase anxiety around taking on a nonreligious identity (Brewster et al. 2020; Doane and Elliott 2015). However, despite these stigmas and uncertainties, research shows that nonreligious people are just as likely to find positive meaning and comfort in their worldviews as are religious people (Edgell et al. 2023; Schnell and Keenan 2011; J. Smith and Halligan 2021; Zuckerman 2011).

More generally, research in social psychology shows that uncertainty or anxiety around having a minority identity, such as being nonreligious, is not a good proxy measure for general mental health. For example, Morandini et al. (2015) find that there is no direct relationship between identity uncertainty and psychological wellbeing among gay men (see also Quinn and Chaudoir 2009). These lines of research suggest that uncertainty about one's identity does not always lead to anxiety and that anxiety about an identity does not necessarily correlate with anxiety generally. Thus, rather than focusing on certainty, as past research on nonreligious health has done, we examine how a more general affective orientation to being nonreligious shapes wellbeing.

3.3. Nonreligious Communities

Research on religion and wellbeing has consistently found that social support gained through regular church attendance is a primary driver of health among the religious (Lim and Putnam 2010; Shor and Roelfs 2013). However, few studies have examined whether organisational participation among the nonreligious predicts wellbeing in similar ways. As the nonreligious population in the United States has grown, so too has the number of social and political organisations devoted to cultivating atheist/humanist/secular community and activism. There are now over 1300 such groups in the United States (García and Blankholm 2016), and they range from political organisations such as the American Atheists to hobby groups such as Godless Gamers to church‐like organisations such as the Sunday Assembly and the Houston Oasis that cultivate ritual and community for the nonreligious (Frost 2023; LeDrew 2015).

Like religious organisations, these nonreligious organisations can foster wellbeing by affirming shared identities, providing social support, and cultivating certainty‐filled beliefs and values (LeDrew 2015; Frost 2019; J. Smith and Halligan 2021). There are a handful of studies that find participation in these nonreligious organisations is positively associated with wellbeing (Brewster et al. 2020; Galen and Kloet 2011; Price and Launay 2018). However, research in this area is largely qualitative or focused on just one nonreligious organisation because most large‐scale survey studies do not include a measure of participation in explicitly atheist or humanist organisations.

3.4. Identity Duration

The length of time someone has been nonreligious may also matter for wellbeing. Many nonreligious people in the United States have become nonreligious after ‘deconverting’ from some form of religion (Voas and Chaves 2016). Belief change and role exit are stressful processes in general, regardless of the belief being changed or the role being exited (Ebaugh 1988), and research shows that a transition out of religion can cause uncertainty, anxiety, and the loss of core social networks (Krause and Wulff 2004; J. Smith 2011; Zuckerman 2011). This is likely why much of the research on religious disaffiliation and health finds that leaving religion is correlated with declines in health (Scheitle and Adamczyk 2010; Upenieks and Thomas 2021).

However, qualitative research on religious disaffiliation has found that the social and psychological disruption that can accompany leaving religion is often short‐lived. For many nonreligious people, this stressful transitory period gives way to acceptance and the creation of new meaning systems and social networks grounded in nonreligious communities (Frost 2019; J. Smith 2011; J. Smith and Halligan 2021; Zuckerman 2011). For others, this transition leads to the decreased salience of religious‐related issues over time or what some call ‘existential indifference’ (e.g., Schnell 2010). This means that we need to pay more attention to how nonreligious wellbeing might change over time as newly nonreligious individuals may report lower levels of wellbeing when compared to those who have been nonreligious for a long period of time.

4. Data and Methods

4.1. Data

Data are drawn from the Nonreligious Engagement and Wellbeing Survey (NEWS). NEWS items measure variation in religious and nonreligious identities and beliefs, physical and mental health, involvement in civic life, social networks and social support. The NEWS survey was administered via the SurveyMonkey Audience service in August 2020. SurveyMonkey recruits participants through online advertising to create national online panels for survey administration. The Audience service employs sampling criteria to provide a demographically balanced sample from their larger panel of recruited respondents. We utilised the default criteria balanced to reflect United States Census benchmarks on gender, age and household income. The survey was administered to two national samples of Americans over the age of 18—one with no filtering criteria (n = 1086) and one additional pool limited to nonreligious respondents to generate a supplemental oversample (n = 240). For this analysis, these samples were combined into one larger dataset (n = 1326).

4.2. Dependent Variables

We utilise three measures of wellbeing as our primary dependent variables. The first is self‐reported life satisfaction. Respondents were asked, ‘In general, how satisfied are you with your life?’ Respondents were given the following response options: (1) very unsatisfied, (2) somewhat unsatisfied, (3) somewhat satisfied, (4) very satisfied. Self‐reported life satisfaction is a measure of subjective wellbeing that reliably and validly captures cognitive appraisal of life overall (Diener 2012). The second dependent variable is self‐reported happiness. Respondents were asked, ‘In general, how happy or unhappy do you usually feel?’ Respondents were given the following response options: (1) very unhappy, (2) somewhat unhappy, (3) somewhat happy, (4) very happy. Our third dependent variable is self‐reported health. Respondents were asked, ‘In general, would you say your health is…?’ Respondents were given the following response options: (1) poor, (2) fair, (3) good, (4) very good and (5) excellent. Despite some limitations, self‐rated health is a frequently used measure that reliably indicates the objective health status of respondents and predicts mortality (Wu et al. 2013; Layes et al. 2012). We keep all three variables as scales and utilise ordered logistic regressions.

4.3. Independent Variables

We utilise four primary independent variables to measure distinct aspects of nonreligious identity and experience. Descriptive statistics for the independent variables are shown in Table 1. We constructed these variables to mirror key variables that past survey research has found are important for mediating the relationship between religion and health, such as attendance and affiliation. However, we also created variables that enable us to test some of the qualitative findings on nonreligious health regarding how affect and change over time might matter for health outcomes among the nonreligious.

TABLE 1.

Descriptive statistics for NEWS respondents.

Demographics
Age	Mean age of respondent in years (1 = 18–29, 4 = 60+)	2.4
Female	Percent female	54%
Married	Percent married or in long‐term cohabiting relationship	50%
Parent	Percent with children	75%
Income	Mean family income in 2020 (1 ≥ $9,999, 10 = $200,000+)	3.9
Education	Mean highest level of education completed (1 = no formal education, 8 = postgraduate)	4.9
White	Percent who identify as White	62%
Black	Percent who identify as Black or African American	7%
Hispanic	Percent who identify as Hispanic	12%
Nonreligious identification
Atheist		8.6%
Agnostic		7.5%
Self‐identified nonreligious		8.1%
Nothing in particular (NIP)		5.6%
Spiritual but not religious (SBNR)		7.5%
Nonreligious group involvement
Attendance at nonreligious groups	Mean attendance (1 = never, 7 = several times a week)	1.8
Nonreligious identity duration
Recent none	Percent who switched from being religious to being nonreligious in the past 5 years	10%
Long‐time none	Percent who switched from being religious to nonreligious more than 5 years ago	17%
Always none	Percent who have always been nonreligious	19%
Affective orientations
Comfort	Percent who typically find comfort/strength in their religious/nonreligious identity	49%
Anxiety	Percent who typically experience anxiety/stress due to their religious/nonreligious identity	11%
Neutral	Percent who do not typically think much about their religious/nonreligious identity or beliefs	39%
N		1326

Our first independent variable is nonreligious identity. We asked our respondents to select a label that best represents their current religious or nonreligious preferences. We provided five different options for nonreligious people: (1) atheist, (2) agnostic, (3) nonreligious, (4) nothing in particular and (5) spiritual but not religious. While ‘atheist’ and ‘agnostic’ signal specific nonreligious identities and belief systems, the ‘nonreligious’ and ‘nothing in particular’ categories capture religiously unaffiliated people who have not taken on a specific identity in relation to religion or nonreligion. We also include the ‘spiritual but not religious’ in our nonreligious sample because they intentionally distance themselves from organised religion and our goal is to capture as much diversity in nonreligious identity as possible.

In total, 37% of our sample identify with one of these five nonreligious identity labels. Note that although we use ‘nonreligious’ as an umbrella category when talking about all of these nonreligious identity labels as a whole, ‘nonreligious’ is also a standalone identity label we offered to respondents to capture people who identify firmly as nonreligious but do not identify as atheist or agnostic. We also provided a range of options for religious respondents, including Protestant, Christian, Jewish, Muslim, Catholic, Mormon and Buddhist, but for the purposes of this analysis we collapsed all the religious respondents into our reference category. Finally, we provided an ‘other’ option for all respondents who did not identify with any of the labels we provided as options.

Our second independent variable is attendance at organised events for an explicitly nonreligious organisation. We asked, ‘How often do you attend organised events for an explicitly nonreligious/atheist/humanist organisation or group?’ We offered the following response options: (1) never, (2) once a year, (3) several times a year, (4) monthly, (5) several times a month, (6) weekly and (7) several times a week. We find that 11.5% of our respondents report attending nonreligious organisations at least once a month.

Our third independent variable is nonreligious identity duration, or how long someone has been nonreligious. We asked, ‘Which of the following describes your religious/nonreligious preferences?’ We gave the following response options: (1) I have always been religious, (2) I have always been nonreligious, (3) I have switched between religious preferences in the past 5 years, (4) I have switched between nonreligious preferences in the past 5 years, (5) I have switched from being religious to being nonreligious in the past 5 years, (6) I have switched from being nonreligious to being religious in the past 5 years, (7) I switched from being religious to nonreligious more than 5 years ago and (8) I switched from being nonreligious to religious more than 5 years ago. We allowed respondents to check all of these descriptors that applied to them. For this analysis, we collapsed these response options into four categories: (1) Always Nones who have always been nonreligious, (2) Long‐time Nones who have been nonreligious for more than 5 years, (3) Recent Nones who have been nonreligious for less than 5 years and (4) All religious response options.

Our final set of independent variables measure affective orientation. We asked our respondents, ‘Which of the following statements best represents the way you experience your religious or nonreligious identities and beliefs?’ We gave three response options: (1) I typically find comfort and strength in my religious/nonreligious identity and beliefs, (2) I typically experience anxiety and stress due to my religious/nonreligious identity and belief and (3) I do not typically think much about my religious/nonreligious identity and beliefs. For this measure, we did not separate the nonreligious from the religious, but instead we created three separate variables for use in our models: (1) comfort, (2) anxiety and (3) affective neutrality.

We tested for collinearity among our variables in order to ensure that our affective orientation variables were not collinear with our wellbeing outcome variables, some of which are also measures of affect. VIF scores are all below 1.7, which is well below the standard cut‐off of 2.5 (Johnston et al. 2017), confirming that our affective orientation variables are not highly correlated with our wellbeing outcomes.

4.4. Control Variables

Our models include a range of control variables that are detailed in Table 1. We include a scale variable for age (1–4), a dichotomous variable for gender (1 = female), a dichotomous variable for marital status (1 = married or in a long‐term cohabiting relationship) and a dichotomous variable for parental status (1 = is a parent). We include a scale variable for income (1–10) and a scale variable for education (1–8). Finally, we control for respondents' race/ethnicity.

5. Results

5.1. Bivariate Analyses

We start by presenting bivariate relationships between our three wellbeing outcomes and our nonreligious and religious identity variables in Table 2. Although we collapse the religious respondents together in our regression models to focus on variations among the nonreligious, these bivariate analyses show that the nonreligious are just as diverse as the religious when it comes to reports of wellbeing and that many nonreligious people report similar or better wellbeing as the religious. Among our nonreligious respondents, atheists report the highest levels of health and satisfaction. Although the spiritual but not religious report the highest levels of happiness, they also report the lowest levels of life satisfaction. In terms of nonreligious identity duration, recent nones report slightly higher levels of happiness and satisfaction compared to longer‐term nones, but recent nones also report the lowest levels of health. Finally, the always religious report slightly higher levels of wellbeing than the always nones across all three measures, but recent nones report similar levels of health and higher levels of happiness when compared to the recently religious.

TABLE 2.

Bivariate analyses for wellbeing outcomes, by religious identification.

	Somewhat or very happy	Good, very good or excellent health	Somewhat or very satisfied
Nonreligious identity
Atheist	78%	81%	82%
Agnostic	78%	69%	73%
Nonreligious	80%	81%	78%
SBNR	81%	72%	71%
NIP	78%	69%	76%
Recent none	86%	72%	79%
Long‐time none	80%	75%	77%
Always none	81%	77%	78%
Religious identity
Mainline protestant	69%	73%	75%
Evangelical protestant	86%	77%	91%
Catholic	88%	84%	84%
Muslim	85%	95%	81%
Jewish	81%	92%	69%
Recently religious	78%	75%	85%
Long‐time religious	88%	73%	88%
Always religious	86%	81%	85%

5.2. Predicting Life Satisfaction, Happiness and Health

In Tables 3, 4, 5 to 3, 4, 5, we use ordered logistic regressions to further test the relationships found in our bivariate analyses. We conducted Brant tests to ensure our ordinal models do not violate the parallel lines assumption. We built the models stepwise so that in Models 1–4, we examine the impacts of each variable in isolation with the dependent variable, and then in Model 5, we combine all of the variables into one model. Note that the control variables (age, gender, marital status, parental status, income, education and race) are included in all of the models, but we do not report their odds ratios.

TABLE 3.

Ordered logistic regressions of self‐reported life satisfaction.

	Model 1	Model 2	Model 3	Model 4	Model 5
Nonreligious identity
Atheist	0.64 (0.13)				0.88 (0.20)
Agnostic	0.56** (0.11)				0.77 (0.18)
Self‐identified nonreligious	0.69 (0.14)				0.96 (0.22)
SBNR	0.51*** (0.11)				0.65 (0.15)
NIP	0.72 (0.16)				1.03 (0.25)
Nonreligious identity duration
Recent none		0.50*** (0.11)			0.91 (0.17)
Long‐time none		0.56*** (0.09)			0.86 (0.15)
Always none		0.76 (0.11)			1.18 (0.19)
Nonreligious communities
Nonreligious group attendance			1.13*** (0.04)		1.11** (0.05)
Affective orientation
Comfort				1.91*** (0.22)	1.68*** (0.23)
Anxiety				0.49*** (0.09)	0.44*** (0.08)
Chi‐square	129.41***	131.02***	119.44***	178.26***	195.11***
Pseudo R 2	0.04	0.04	0.04	0.06	0.07
N	1326	1326	1326	1326	1326

Note: All models include the following controls: age, gender, marital status, parental status, income, education and race.

*p < 0.05.

^** p < 0.01.

^*** p < 0.001.

TABLE 4.

Ordered logistic regressions of self‐reported happiness.

	Model 1	Model 2	Model 3	Model 4	Model 5
Nonreligious identity
Atheist	0.69 (0.14)				0.85 (0.20)
Agnostic	0.70 (0.14)				0.85 (0.20)
Self‐identified nonreligious	0.84 (0.17)				1.08 (0.25)
SBNR	0.77 (0.16)				0.93 (0.25)
NIP	0.82 (0.19)				1.09 (0.28)
Nonreligious identity duration
Recent none		0.70 (0.15)			1.18 (0.23)
Long‐time none		0.73* (0.12)			1.00 (0.18)
Always none		0.80 (0.12)			1.17 (0.19)
Nonreligious communities
Nonreligious group attendance			1.07 (0.04)		1.06 (0.05)
Affective orientation
Comfort				1.73*** (0.21)	1.72*** (0.24)
Anxiety				0.44*** (0.08)	0.41*** (0.08)
Chi‐square	92.89***	92.92***	88.92***	150.31***	155.48***
Pseudo R 2	0.03	0.03	0.03	0.05	0.05
N	1326	1326	1326	1326	1326

Note: All models include the following controls: age, gender, marital status, parental status, income, education and race.

^* p < 0.05.

**p < 0.01.

^*** p < 0.001.

TABLE 5.

Ordered logistic regressions of self‐reported health.

	Model 1	Model 2	Model 3	Model 4	Model 5
Nonreligious identity
Atheist	1.05 (0.20)				1.05 (0.23)
Agnostic	0.68* (0.14)				0.67 (0.16)
Self‐identified nonreligious	0.99 (0.19)				0.99 (0.22)
SBNR	0.79 (0.15)				0.81 (0.17)
NIP	0.86 (0.20)				0.89 (0.22)
Nonreligious identity duration
Recent none		0.93 (0.18)			0.91 (0.16)
Long‐time none		0.94 (0.14)			1.00 (0.17)
Always none		1.17 (0.17)			1.19 (0.18)
Nonreligious communities
Nonreligious group attendance			1.06* (0.04)		1.07 (0.04)
Affective orientation
Comfort				1.08 (0.12)	1.01 (0.13)
Anxiety				0.73 (0.13)	0.69* (0.13)
Chi‐square	77.19***	73.44***	74.92***	76.62***	86.75***
Pseudo R 2	0.02	0.02	0.02	0.02	0.02
N	1326	1326	1326	1326	1326

Note: All models include the following controls: age, gender, marital status, parental status, income, education and race.

^* p < 0.05.

**p < 0.01.

^*** p < 0.001.

The odds ratios reported for Model 1 in Table 3 show that agnostics and the spiritual but not religious are significantly less likely to report being satisfied with their lives, but the other nonreligious identity categories are not significantly different from the religious reference group. In Model 2, recent nones and long‐time nones are significantly less likely to be satisfied with their lives, but always nones are not significantly different from the religious reference group. In Model 3, attendance at explicitly atheist/humanist/nonreligious organisations significantly increases reports of life satisfaction. In Model 4, our measures of both comfort and anxiety are significant, but in opposite directions. This means that people who report finding comfort in their religion/nonreligion are significantly more likely than those who are affectively neutral to report high levels of life satisfaction, whereas people who report finding anxiety in their religion/nonreligion are significantly less likely than those who are affectively neutral to report high levels of life satisfaction. Finally, in Model 5, we include all of our independent variables and find that only the measures for nonreligious organisational involvement and affective orientation remain significant. This means that attendance at organised nonreligious groups and affective orientations toward being nonreligious are likely mediating the effects of identity labels and identity duration for our nonreligious respondents, which we examine in more detail in our mediation analyses later in this section.

In Table 4, we run the same set of ordered logistic regression models but with self‐reported happiness as the dependent variable. In Model 2, long‐time nones are significantly less likely than the religious to report high levels of happiness. In Model 4, comfort predicts increased happiness, whereas anxiety predicts decreased happiness when compared to affective neutrality. Similar to our findings for life satisfaction, Model 5 shows that when we combine all of the variables into one model, nonreligious identity duration is no longer significant, but affective orientation remains significant.

Finally, in Table 5, we run the same set of ordered logistic regression models with self‐reported health as the outcome. We find that while agnostics report lower health than the religious, none of the other nonreligious identity categories or identity duration variables are significant in Models 1 and 2. In Model 3, nonreligious group attendance is significantly and positively related to self‐reported health. In Model 5, when all of the independent variables are in the same model, anxiety as an affective orientation becomes significant, meaning that those who are anxious about their religion/nonreligion are significantly less likely to report good health than the affectively neutral, but those who find comfort in their religion/nonreligion are not significantly different from the affectively neutral when it comes to self‐reports of health.

5.3. Predicting Comfort and Anxiety

Given that the affective orientation measure of whether respondents report comfort, anxiety or neutrality surrounding their nonreligion/religion was consistently significant across our models in Tables 3, 4, 5 to 3, 4, 5, we ran additional models with the comfort and anxiety affective orientation variables as the dependent variables. This helps us to further examine what might make someone more likely to experience anxiety or comfort around their nonreligion.

In Table 6, Models 1 and 2 show that all nonreligious identity categories and all nonreligious identity duration variables have a significant and negative relationship with the comfort outcome. This means that nonreligious people are less likely to report finding comfort in their nonreligion than religious people are to report finding comfort in their religion. However, in Model 3, we see that nonreligious group attendance is positively associated with the comfort outcome, meaning those who attend nonreligious groups are more likely to report comfort with their religion/nonreligion than those who do not attend such groups.

TABLE 6.

Logistic regressions of comfort with religion/nonreligion.

	Model 1	Model 2	Model 3	Model 4
Nonreligious identity
Atheist	0.11*** (0.03)			0.28*** (0.08)
Agnostic	0.10*** (0.03)			0.25*** (0.08)
Self‐identified nonreligious	0.05*** (0.02)			0.12*** (0.04)
SBNR	0.28*** (0.06)			0.58* (0.14)
NIP	0.09*** (0.03)			0.15*** (0.05)
Nonreligious identity duration
Recent none		0.14*** (0.04)		0.36*** (0.08)
Long‐time none		0.11*** (0.02)		0.31*** (0.07)
Always none		0.11*** (0.02)		0.31*** (0.06)
Nonreligious communities
Nonreligious group attendance			1.28*** (0.06)	1.10* (0.05)
Constant	0.82 (0.25)	0.60 (0.18)	0.29*** (0.08)	0.83 (0.28)
Chi‐square	347.83***	332.10***	79.34***	413.76***
Pseudo R 2	0.19	0.18	0.04	0.23
N	1326	1326	1326	1326

Note: All models include the following controls: age, gender, marital status, parental status, income, education and race.

^* p < 0.05.

**p < 0.01.

^*** p < 0.001.

In Table 7, we run the same set of models but with the anxious affective orientation variable as the outcome. In Model 1, we see no significant differences between any of the nonreligious identity labels and the religious reference groups in reports of feeling anxious about one's religion/nonreligion. This is an important finding, as even the more liminal identities of agnostic and nothing in particular, which previous scholarship has found to be correlated with lower reports of health than atheism, are not significantly different from the religious on this measure. Thus, although the nonreligious are less likely to report finding comfort in their worldviews than are the religious, this is not because they instead experience anxiety. Rather, the nonreligious in our sample are most likely to be affectively neutral.

TABLE 7.

Logistic regressions of anxiety about religion/nonreligion.

	Model 1	Model 2	Model 3	Model 4
Nonreligious identity
Atheist	0.89 (0.32)			0.78 (0.31)
Agnostic	0.75 (0.28)			0.60 (0.24)
Self‐identified nonreligious	0.70 (0.26)			0.59 (0.24)
SBNR	1.77 (0.56)			1.46 (0.51)
NIP	1.48 (0.53)			1.33 (0.50)
Nonreligious identity duration
Recent none		2.75*** (0.78)		3.25*** (0.78)
Long‐time none		1.06 (0.31)		1.56 (0.45)
Always none		1.29 (0.32)		1.63* (0.40)
Nonreligious communities
Nonreligious group attendance			1.23*** (0.07)	1.25*** (0.07)
Constant	0.93 (0.39)	0.77 (0.32)	0.56 (0.24)	0.37* (0.18)
Chi‐square	64.73***	70.18***	71.54***	104.55***
Pseudo R 2	0.07	0.08	0.08	0.11
N	1326	1326	1326	1326

Note: All models include the following controls: age, gender, marital status, parental status, income, education and race.

^* p < 0.05.

**p < 0.01.

^*** p < 0.001.

In Table 7's Model 2, we find that it is only the recent nones who are significantly more likely to report anxiety, with recent nones being over 2 times more likely than the religious reference category to report anxiety. In Model 3, we find that nonreligious group attendance significantly increases the likelihood of reporting anxiety. This is interesting, given that nonreligious group attendance also significantly increases the likelihood of reporting comfort. It could be that attending organised atheist groups gives some people anxiety. Or it could be that we are capturing people at different stages of participation. Anxiety may be driven by recent nones who are seeking out nonreligious organisations to affirm their new identities and buffer the negative effects of nonreligious stigma (e.g., Brewster et al. 2020), whereas comfort may be driven by long‐time and always nones who have sustained participation and built social networks through these organisations.

5.4. Mediation Analyses

For our final set of analyses, we ran KHB mediation analyses to further explore the effects of affective orientation and attendance at nonreligious groups, as these variables appear to be doing most of the work in our regression models. The KHB method allows researchers to decompose the total effects associated with a given relationship into direct effects and indirect effects that are confounded or mediated by other variables (E. Smith et al. 2019). We employed the khb command in Stata to calculate the significance of the total, direct and indirect effects of each key predictor. Decomposition results suggest that 52% of the relationship between being a recent none and reports of life satisfaction is mediated by the affective orientation measures, and 41% of that relationship is mediated for long‐time nones. We also find that 73% of the relationship between being a long‐time none and reports of happiness is mediated by the affective orientation measures. Finally, we find that 12% of the relationship between being a long‐time none and reports of life satisfaction is mediated by attendance at nonreligious groups.

6. Discussion and Conclusions

A large body of research in the sociology of religion has found that the existential certainty religion provides (Abu‐Raiya et al. 2015; Ellison 1991), the social support of religious communities (Lim and Putnam 2010; Shor and Roelfs 2013), and the ‘protective’ beliefs often found in religion (e.g., prohibiting drinking) (Gardner et al. 1995; George et al. 2002) are all linked to positive health outcomes. However, despite a dramatic growth in the number of nonreligious Americans since the 1990s (Pew Research Center 2024; Voas and Chaves 2016), our understanding of how being nonreligious impacts wellbeing has been slow to evolve. We know that leaving or rejecting religion can have negative effects on health (e.g., Hayward et al. 2016; Upenieks and Thomas 2021). However, the dearth of large‐scale survey research focused on nonreligious populations has hindered our ability to examine potential variation in wellbeing outcomes among the nonreligious. With this study, we join a small but growing literature that shifts the analytic focus onto the nonreligious and examines how variations in nonreligious identity and experience might impact wellbeing.

We examine how some of the key factors driving the relationship between religion and health—affiliation with an identity‐affirming group and regular attendance at a belief‐affirming organisation—predict variations in wellbeing among the nonreligious. We also examine the impact of some novel health‐promoting factors that have emerged from qualitative studies of wellbeing among the nonreligious—the duration of and affective orientation toward one's nonreligious identity. Our findings are informative for understanding how identity, social support, and affect shape wellbeing for both religious and nonreligious Americans.

We find a few significant differences across the five nonreligious identities in our sample. In our models that include only the nonreligious identity variables, agnostics and the spiritual but not religious report lower life satisfaction, and agnostics report lower self‐reported health when compared to the religious reference group. This aligns with past research that finds nonreligious people with less committed identities and beliefs, such as agnostics, often report lower wellbeing than those with more committed identities, such as atheists (Baker et al. 2018; May 2018). However, we found that our measures of identity duration, nonreligious group attendance, and affective orientation were more powerful predictors of wellbeing than identity categories alone.

By far the most significant predictor of wellbeing in our analysis was our measure of affective orientation. In response to recent research that nonreligious people's emotional responses to being nonreligious can impact their wellbeing, we developed a measure of ‘affective orientation’ to test this theory. Unsurprisingly, finding ‘strength and comfort’ in your nonreligion is better for your health and happiness than feeling ‘stressed and anxious’ about your nonreligion. However, we argue that our findings around affective orientation tell a more complicated story. Not only did we find that nonreligious people are not any more likely to experience anxiety around their beliefs than religious people, but our findings also suggest that nonreligious people's affective orientations toward being nonreligious change over their life course.

Much of the research on religion and health assumes nonreligious people are more generally prone to identity/belief anxiety because they lack the certainty and coherence offered by religion (e.g., Hayward et al. 2016). However, if that were the case, we would find anxious orientations to be common across all of our nonreligious groups. Instead, we find that anxious orientations are largely concentrated among the newly nonreligious, which suggests that this anxiety stems more from role transition than from nonreligion itself (see Ebaugh 1988). This aligns with qualitative studies of nonreligious life course narratives that find many nonreligious people become more comfortable with their nonreligion over time, even if they embrace nonreligious identities that eschew belief certainty as a central feature (Frost 2019; J. Smith 2011; Zuckerman 2011).

Relatedly, we find that the recently nonreligious (people who have been nonreligious for less than 5 years) report lower levels of life satisfaction than the religious, and we find that the long‐time nonreligious (people who have been nonreligious for five or more years) report lower levels of life satisfaction and happiness than the religious. However, our respondents who have always been nonreligious are not significantly different than the religious on any of our wellbeing measures. This finding is in line with other research that finds worse health outcomes for religious switchers than for stable affiliates or nonaffiliates (Fenelon and Danielsen 2016; May 2018).

Our findings raise important questions to be considered in future research, and they point to the need for more longitudinal studies on the relationship between religion, nonreligion, and health. Although our data in this analysis are cross‐sectional, our proxy measure for change over time suggests that initial anxieties stemming from religious disaffiliation can dissipate over time. Given that the decline in religious affiliation is such a recent event in the United States context (Voas and Chaves 2016), our cross‐sectional surveys are likely capturing a high number of people who recently left religion. This may be a major factor driving the negative relationships between nonreligion and health in prior research. As these newly nonreligious cohorts age and as more people raise their children with no religion (Voas and Chaves 2016), survey samples will include more ‘long‐time’ and ‘always’ nones—and our findings suggest that health outcomes among the nonreligious may shift as a result.

Finally, we find that participation in explicitly atheist/secular/humanist organisations is beneficial for wellbeing. In our models, increased attendance in nonreligious organisations was significantly correlated with increased life satisfaction and increased self‐reported health. NEWS is one of the first national survey studies to include a measure of ‘nonreligious attendance’, which is important given that religious attendance is often significantly correlated with increased wellbeing for religious people (e.g., Lim and Putnam 2010). However, more research is needed to examine the mechanisms through which these organisations foster wellbeing and how different types of nonreligious organisations (e.g., political vs. intellectual vs. communal) might impact wellbeing in distinct ways.

We also conducted mediation tests to further explore the strength of our predictor variables. We find that affective orientation is by far the strongest mediator of the relationship between nonreligion and wellbeing among our measures, but attendance at nonreligious organisations also mediates a substantial part of this relationship. These findings are consonant with past research that finds social support through values‐based communities is a significant factor driving health among both the religious and nonreligious (e.g., Lim and Putnam 2010; Price and Launay 2018), but they also suggest that more research needs to be done on the affective aspects of religious and nonreligious identities that drive wellbeing. In particular, we find that a disproportionate number of nonreligious people in our sample are what we label ‘affectively neutral’. Rather than feeling anxious about or comforted by their nonreligion, a majority of our nonreligious respondents reported ‘not thinking much about their nonreligion’, and these affectively neutral respondents were significantly more likely than those with anxious orientations to report positive wellbeing. The religion and health literature has largely focused on certainty and coherence, but our findings suggest that indifference and neutrality may be another pathway to wellbeing (c.f. Schnell 2010).

One important limitation of our survey is that it does not include a measure of discrimination, which is likely a key factor shaping wellbeing among the nonreligious. Nonreligious people—especially atheists—continue to be stigmatised and discriminated against in the United States (Edgell et al. 2016), and many experience stress and anxiety in their attempts to manage this stigma (Abbott and Mollen 2018; Mackey and Rios 2023). This stress may be an important factor in explaining our finding that the recently nonreligious are much more anxious about being nonreligious than are other nonreligious people. While some of this anxiety may be existential, it is likely also social (see also Edgell et al. 2017), and many of the newly nonreligious likely do not yet have the stigma management strategies that longer‐term nonreligious people have developed.

Taken together, our findings highlight the importance of taking nonreligious diversity into account in studies of religion and health. We find that many of the same factors that predict wellbeing among the religious—shared group identities, positive affect, and organisational participation—have secular correlates that can bolster wellbeing among the nonreligious. Going forward, we urge researchers to focus less on comparing the religious and the nonreligious in studies of health and wellbeing and to instead focus on the shared features of both that might promote wellbeing for everyone.

Author Contributions

Jacqui Frost: conceptualization (equal), formal analysis (lead), methodology (lead), writing – original draft (lead). Penny Edgell: conceptualization (equal), funding acquisition (lead), writing – review and editing (equal). Mahala Miller: conceptualization (equal), validation (lead), writing – review and editing (equal).

Ethics Statement

Data collection and analysis were approved by and conducted under the oversight of Rice University IRB, #IRB‐FY2021‐27.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Permissions to Reproduce Materials

The authors have nothing to report.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.