Religious Socialization and Millennial Involvement in Organized and Digital Nonbelief Activities

Organizations where nonbelieving persons gather, share and promote their views and public presence have existed since the 19th century in North America. These groups have risen in prominence and profile in recent decades. The rise of national and international groups such as the Council for Secular Humanism, the Secular Student Alliance, American Atheists, the Freedom from Religion Foundation, and the Sunday Assembly are just a few examples. The main factors driving this increased notoriety include the public popularity of New Atheist authors in the 2000s, a growing nonreligious segment of the population among whom some are seeking communities and spaces to share their nonbelief identities and views, along with a growing reaction against religiously-motivated extremism and religion in public life around the world. Nevertheless, there are also indications that only a small portion of nonbelievers take part in organized atheist, humanist and secularist activities. Even with membership to these organizations measuring in the tens of thousands, even in the hundreds of thousands by some estimates (Cragun and Manning 2017: 5; Zuckerman, Galen & Pasquale 2016: 213), this only represents a fraction of the many millions of nonbelievers surveys estimate are out there in both the United States and Canada. In 2018, an estimated 11% of the USA adult population, approximately 26.7 million people, do not believe in God or a higher power.1 In Canada in 2015, this estimate reaches 27% of the adult population (Angus Reid Institute 2015), representing approximately 7.5 million individuals. Whereas some nonbelievers are connected with atheist, humanist or secularist organizations, many more appear not to be. If we broaden our view to account for the diversity of networks that share and discuss atheist, humanist and secularist ideas and materials online, this digital creation and consumption of nonbelief content appears to be more common than individuals taking part in face-to-face activities organized by nonreligious groups. Researchers such as Addington (2017), Lundmark and LeDrew (2019), and Smith and Cimino (2012) have pointed to the importance of digital spaces for atheist activism and community building. These digital spaces seem especially crucial for a target population that is made up in large part of members of younger generations: for example, an estimated 42% of adult nonbelievers in the USA are Millennials, born between 1984 and 2000.2 There has not, however, been any research to date studying the prevalence of organized in-person and digital nonbelief activities among nonreligious populations in the USA and Canada. What proportion of nonbelievers, and of the population more broadly, take part in such activities over the course of a given year? Our study offers a first series of estimates of the presence and frequency of these activities among a representative sample of just over 2,500 young adult respondents between the ages of 18 and 35 across the USA and Canada, based on our 2019 Millennial Trends Survey data. Wilkins-Laflamme, S and Thiessen, J. 2020. Religious Socialization and Millennial Involvement in Organized and Digital Nonbelief Activities. Secularism and Nonreligion, 9: 2, pp. 1–15. DOI: https://doi.org/10.5334/snr.126

In general, individuals take part in these organized and digital atheist, humanist and secularist activities to build, reinforce and share their views and nonreligious identities with other like-minded individuals (Cimino and Smith 2014;LeDrew 2015;Smith 2013;. Researchers such as Brewster (2014), LeDrew (2015) and Smith (2013) add that nonreligious organizations can be helpful spaces for atheists to "come out." Nonbelievers, especially self-identified atheists, remain one of the least favorably viewed groups in American society, often facing distrust, negative stereotypes and discrimination (Abbott and Mollen 2018;Baker and Smith 2015;Brewster et al. 2016;Cragun et al. 2012;Edgell, Gerteis & Hartmann 2006;Gervais, Norenzayan & Shariff 2011;Hammer et al. 2012). One response to this stigmatization and discrimination may be seeking out community and like-minded individuals within organized and digital nonreligion, to cope with psychological stress, to mitigate an otherwise marginalized and stigmatized status, and to build a new positive social support system. The primary goals are to gather with others who think similarly about the world, to reinforce their shared nonbelieving worldview and, in some cases, to actively oppose the influence and spread of religion in society by creating a secular social movement (see Niose 2012).
Emerging evidence is beginning to show that nonreligious organizations are indeed found especially in areas characterized by a more normative presence of religion (see Garcia and Blankholm 2016;Wilkins-Laflamme forthcoming). Organized atheist, humanist and secularist groups emerge more often where nonbelievers are a notable minority, especially when compared to a substantial Evangelical Protestant majority in the USA. Cimino and Smith (2007) theorize that some religious nones become actively involved in nonreligious communities in reaction against their real or perceived minority status in society. Akin to Smith's (1998) use of subcultural identity theory to explain how Evangelicals leverage an embattled narrative with the rest of secular society, those without religion are similarly constructing a narrative of marginalization against Evangelicals (also see Guenther, Mulligan & Papp 2013). Cimino and Smith (2007) share interview data from nonreligious individuals whose interactions with Evangelicals contributed to a growing fear of the role that Evangelicalism plays in American society, so much so that some nonbelievers became more involved in nonreligious communities with the goal of strengthening the secular voice and impact in society.
Subcultural identity theory is a useful explanation then for nonbelievers' decisions to form or join explicitly nonreligious communities. This said, this framework has only been systematically tested so far at the macro-level of regional population and organizational characteristics (notably by Garcia and Blankholm 2016). Does it also hold at a micro-social level?
Those nonbelievers who are most likely to develop an embattled narrative are conceivably those facing regular discrimination. Research by Cragun et al. (2012) provides a descriptive picture of this discrimination. They find that, although some nonbelievers do not feel they have been personally discriminated against in the five years prior to their study, others do cite discrimination experienced in one or more of the following contexts: family, workplace, school, military, social relationships, and volunteer organizations. Hammer et al. (2012) pick up on these findings and summarize six forms of anti-atheist discrimination (presented in decreasing order of occurrence, with the exception of the final "other category"): slander; coercion; social ostracism; denial of opportunities, goods, and services; hate crimes; and other. Perceived discrimination is specifically expressed in the following ways (again, in decreasing order of occurrence): anti-atheist comments in the media; expectations to participate in religious prayers or religious services against one's will; being told that atheism is wrong; and being treated differently due to one's atheism.
Additionally, both studies by Cragun et al. (2012) and Hammer et al. (2012) find that nonbelievers who are "out" and come from more religious family backgrounds are the most likely to face discrimination and stigma from their remaining religious family, social networks and surrounding social environment: "Similar to other minority individuals with concealable stigmatized identities, atheists who more strongly identified with their atheism, who were ' out' about their atheism to more people, and who grew up with stricter familial religious expectations reported experiencing more frequent discrimination" (Hammer et al. 2012: 43). In regions where religious belief, belonging and behavior are prevalent, and where Garcia and Blankholm (2016) show that organized nonbelief is also more widespread among nonbelievers in the American context, there are also more nonbelievers coming from these more religious family backgrounds and conceivably experiencing higher levels of discrimination.
To use Bullivant's terms (2017), we wonder then whether "nonverts" are more likely to turn to organized nonbelieving groups than "cradle" religious nones? Cradle nonbelievers coming from less religious or nonreligious family backgrounds and environments do not really have a coming out experience when it comes to their nonbelief. Nonreligion and nonbelief are far more normalized for cradle nones, they often face less discrimination for it, and they often already have access to like-minded individuals within their family and surroundings. There appears less of a need then for cradle nonbelievers to seek out new safe community spaces to "come out" and discuss their worldviews with others in organized and digital nonreligious groups.
There are hints of these findings in other qualitative studies. Drescher (2016) shows for example that nonverts in the USA coming from conservative Catholic or Evangelical backgrounds were much more likely to experience a troubled passage towards disaffiliation than their counterparts from liberal Protestant families. Zuckerman (2012) similarly talks about a transformative disaffiliation experience when individuals break away from a stronger and typically more conservative religious background. Most of Schutz's (2017) interviewees involved with nonreligious organizations had some form of religious upbringing as children. Relative to those with little to no religious upbringing, are nonbelievers with stronger religious socialization experiences more likely to face greater discrimination and develop an embattled narrative, and thus pursue greater rates of participation in organized atheism, humanism and secularism once disaffiliated to cope with and counteract these negative experiences?

Vestiges of Religious Life Theory
A second, potentially complementary, mechanism could also be at play in the link between religious socialization and involvement in organized atheism, humanism and secularism among adult nonbelievers. This second mechanism has less to do with an embattled, stigmatized subcultural identity that develops out of negative social experiences when leaving religion, and more to do with seeking some of the vestiges of religious life that individuals grew up to love, minus God (Cross 2017). As Thiessen (2015) shows, religious nones who were raised in a religious tradition identify lost community as one of the main drawbacks of disaffiliation. Individuals who were religiously socialized are used to certain types of community and social interactions tied to a religious group. When they leave the faith, they still may seek out similar types of community elsewhere, including in organized and digital nonbelieving groups. Cross (2017) finds this to be a prevalent reason why many take part in the Sunday Assembly, where the familiarity of its liturgy, function, and form-all the "good" things of church minus God-is a helpful bridge for some nonreligious individuals to find and experience community with others similar to themselves. These nonverts want to break from God, but not necessarily from all the community aspects of religious life, at least not right away.

The Current Study's Hypotheses
Building on the aforementioned literature, it is conceivable that organized atheist, humanist and secularist groups along with their digital spaces may be first and foremost transitional activities, networks and environments for first generation (disaffiliated) nonreligious individuals still impacted by their religious upbringing. Although religious socialization during childhood has been shown time and again to be tied to a higher likelihood of remaining an active religious believer as an adult (see for example Bengtson, Putney & Harris 2013;Collins-Mayo and Dandelion 2010), when it is not, it may lead newly nonreligious individuals to more involved forms of organized and digital nonbelief.
For the purposes of the present study then, these subcultural identity and vestiges of religious life theories can be formulated into the following hypotheses: H 1 : Among nonbelievers, there will be a positive association between religious childhood socialization and the likelihood of being involved with organized and digital atheist, humanist and secularist groups as an adult.
H 2 (subcultural identity theory): This positive association will be mediated through having experienced discrimination, stigma or trauma due to one's nonbelief. Nonbelievers coming from a more religious background will be more likely to experience discrimination as adults, and in turn will be more likely to participate in organized and digital nonreligious activities.
H 3 (vestiges of religious life theory): Once the variable for having experienced discrimination, stigma or trauma due to nonbelief is controlled for, there will remain a distinct positive association between religious childhood socialization and involvement with organized and digital atheist, humanist and secularist groups among adult nonbelievers, attributed to seeking some vestiges of religious life among these groups.

Methodology
In order to test these hypotheses, we use data from our 2019 Millennial Trends Survey (MTS). The MTS was administered online between 4-27 March 2019 in both English and French, by Sarah Wilkins-Laflamme at the University of Waterloo, Canada. The questionnaire contains a total of 69 questions on the respondent's sociodemographic characteristics, (non)religious and (non)spiritual affiliations, beliefs and practices, friendship networks, as well as inclusivity attitudes. The complete MTS questionnaire and technical documentation can be found in the online supplementary materials: DOI: https://doi.org/10.5334/ snr.126.s1 A total of 2,514 respondents (1,508 from Canada and 1,006 from the USA) aged 18-35 completed the 15-minute web survey. Respondents were recruited through Léger's panel of members (leger360.com) to complete the survey hosted by the University of Waterloo's Survey Research Centre (uwaterloo.ca/survey-research-centre/). Potential respondents were sent an e-mail invitation to complete the web survey, and then were sent reminders up to two times, if necessary. Age, gender, regional and education level quotas were applied during the initial random selection of respondents, and later monitored as responses came in to adjust further recruitment efforts and completes. 3 Post-stratification weights were then created and applied to the statistical analyses in order to achieve greater young adult population representativeness on the variables of country of birth, household income and race/ethnicity. 4 The final response rate for the MTS was 6.5%: lower than the 10-15% response rates common for online surveys, mainly because of the additional recruitment efforts to fill some of the harder to reach quotas (notably young adult males with no university education). This survey was reviewed and received ethics clearance through the University of Waterloo's Research Ethics Committee. Table 1 contains unweighted descriptive statistics for the MTS variables used in this study, among the subsample of 982 respondents who indicate that they do not believe in God or a higher power. 5 Frequency of taking part in meetings or activities with atheist, humanist or secularist organizations; frequency of posting on social media about atheist, humanist, secularist or nonbelief values, views or practices; and frequency of reading or watching online content on atheist, humanist, secularist or nonbelief values, ideas or practices are the three outcome variables that we explore in our statistical models.
Although not mutually exclusive of one another-some organized nonbelief activities taking place online for example, and some digital content being read or watched via social media-these three variables do capture different (although not exhaustive) dimensions of organized and digital nonbelief among individuals: participation in organized groups, social media posting and digital media consumption.
In order to estimate the prevalence of organized and digital nonbelief activities among young adult USA and Canadian populations, these three outcome variables were initially analyzed using weighted descriptive univariate and bivariate Spearman correlation statistics. Then, to capture potential associations between religious socialization and frequency of participation in organized and digital nonbelief activities among young adult nonbelievers, three separate ordered logit (ologit) models controlling for sociodemographics and worldview were generated. A number of variables capturing different dimensions of religious socialization-mother's and father's religious affiliation, mixed (non)religious parents, and frequency of religious education as a child-were included as the main predictors in these ologit models. Finally, to measure the impact of nonreligious discrimination on organized and digital nonbelief activities among young adult nonbelievers, along with the direct and indirect effects of religious socialization through this mediator, a generalized structural equation (path) logit model with ordinal family links was generated. The pathways in this model are illustrated in Figure 4, controlling for age, gender, Canadian residence, level of education, rural residence and generation of immigration.

Prevalence of Organized and Digital Nonbelief Activities
We begin by exploring how widespread participation in atheist, humanist and secularist activities is in Canada and the USA, based on 18-35-year-old respondents' self-reporting. Respondents were asked: "In the past 12 months, how often on average did you practice or take part in the following activities, either in a group or on your own? Meetings or activities with an atheist, humanist or secularist organization (such as American Atheists, Atheist Alliance International, Humanist Association of Canada, Sunday Assembly, etc.)." 11% of Canadian and a quarter of USA young adult respondents said they take part in such activities at least once a month, and another 6% and 12% respectively at least once a year (see Figure 1). This can include both in-person participation and involvement with such organizations online. 67% of those who take part in organized nonbelief meetings or activities once a year or more in Canada also say they do so with a friend at least once a year as well; a rate that reaches 75% in the USA.
Higher proportions have posted on social media about atheist, humanist, secularist or non-belief values, views or practices: 17% of Canadian and 32% of American young adult respondents say they do so once a month or more, and another 16% in both countries say they do so at least once a year (see Figure 1). Even more say they read or watch online content on atheist, humanist, secularist or non-belief values, ideas or practices: 29% of Canadian and 41% of American young adult respondents say they do so once a month or more, and another 30% and 24% respectively say they do so at least once a year (see Figure 1).
Interestingly, most of these individuals who say they are involved with organized nonbelief groups and digital nonbelief activities are not necessarily nonbelievers themselves. 84% of American and 89% of Canadian young adult respondents who indicate they take part in organized nonbelief meetings or activities at least once a month, for example, say they believe in God or a higher power. These data are corroborated by another survey run in 2017 in British Columbia and the states of Washington and Oregon with adult respondents of all ages: the Pacific Northwest Social Survey. 6 In that sample, 12% of respondents who believe in God or a higher power take part in meetings or activities with atheist, humanist or secularist organizations at least once a month, compared with 2% of nonbelieving respondents. We also see this involvement of believers when it comes to digital nonbelief activities: in the 2019 MTS, 78% of those posting on social media about nonbelief at least once a month say they believe in God or a higher power, and 76% of those reading or watching online nonbelief content are believers. 7 Although existing qualitative research has focused almost solely on nonbelievers who are involved in organized atheism, humanism or secularism, there also appears to be a substantial believing population who say they come into regular contact with these groups. This could be for a number of reasons. Some may be interpreting the humanism and "Sunday Assembly" in the survey questions to include Christian humanism and Sunday church attendance, not just secular humanism and the nonreligious organization Sunday Assembly. Some might have a broader definition of which organizations fall under the "atheist, humanist or secularist" umbrella, despite the examples provided in the survey question. Others still could actually be coming into contact with nonreligious organizations to scope out the competition, or just to explore out of curiosity. In regards to the social media nonbelief postings, some may be engaging critically with atheist, humanist and secularist values, views and practices. Some believers may just be joiners in general, and be more interested in discussing and sharing worldviews and talking about their beliefs regardless of the setting. They may be brought into contact with organized nonbelief groups via joint ecumenical activities with their own religious group. They may be participating with a nonbelieving friend or family member as a show of support or as a social activity. There could also be a transition period for many young adults when they are still believing, but are thinking of leaving or have just left their religious group, and are exploring organized nonreligious group options and seeking out vestiges of religious life. More research is needed on these believers potentially involved with organized nonreligion, research that goes beyond the scope of the present study. Moving forward in this paper, we will dig deeper into the nonbelieving population that is involved with organized and digital nonbelief activities. A second trend in the data are the lower rates of participation in organized and digital nonbelief activities in Canada, compared with those in the USA. Only an estimated 3% of nonbelieving young adults in Canada take part in organized atheist, humanist or secularist meetings or activities at least once a month, compared with 12% in the USA. This finding supports subcultural identity theory, in that Canada has a much smaller Evangelical Protestant population that is much less prevalent in politics and public life (Reimer 2003;Reimer and Wilkinson 2015). Consequently, Canadian nonbelievers may feel less stigmatized and thus less of a need to participate in like-minded organizations. More generally, Canada is also a country where many citizens are less involved with organized groups when it comes to their religious or secular belief systems or worldviews. For instance, regular religious service attendance rates are much lower in Canada than in the USA (Bibby 2017;Wilkins-Laflamme 2014), as are rates of regular participation in organized nonbelief. These lower rates of participation also extend to the digital world. For example, only an estimated 17% and 13% of Canadian young adults post on social media at least once a month about religion or nonbelief respectively, compared with 32% and 27% in the USA. A third trend to explore among believers and nonbelievers alike is the link between participation in organized and digital religious activities, and organized and digital nonbelief activities. 82% of nonbelievers who take part in organized atheist, humanist and secularist meetings or activities at least once a year also say they attend religious services or meetings at least once a year. This rate rises to 97% among believers. 8 A similar link is found between digital religious and nonbelief activities: 64% of nonbeliever young adults who post about nonbelief on social media at least once a month also post about religious or spiritual beliefs, values, views or practices at the same frequency; 9 62% of nonbeliever young adults who read or watch nonbelief content online at least once a month also read or watch online content on religious or spiritual beliefs, values, views or practices at the same frequency. 10 It seems then that many who are involved with organized and digital nonbelief activities are also involved with their religious or spiritual counterparts, either out of mutual interest, due to social ties, or to engage critically with the other side.
A fourth observation that stands out to us is the overlap between different organized and digital nonbelief activities among nonbelievers: being involved with organized atheism, humanist or secularism; posting on social media about nonbelief; and consuming digital textual, visual and video nonbelief content. Table 2 contains the correlation matrix for these three variables among nonbelieving young adult respondents. All are statistically significant and positively correlated with one another. Figure 2 illustrates in turn the rates of different types of nonbelievers according to their participation at least once a year in these various activities. Overall, 51% of our young adult nonbeliever sample did not take part in any of these activities in the year prior to the survey (whom we refer to here as inactive nonbelievers); 24% only read or watched digital nonbelief content at least once in the year prior to the survey; 5% only posted on social media   about nonbelief; another 9% are social media posters and digital nonbelief content consumers; 7% took part in organized nonbelief activities, social media posting and digital nonbelief content consumption; 2% took part in organized nonbelief activities only; and another 1% are participating in organized nonbelief activities and digital nonbelief consumption without posting on social media about them.
From Figure 2, a major take-away is that just over half of young adult nonbelievers in our sample are not involved in any kind of atheist, humanist or secularist activity off-or online. Next, those who are involved in some way tend to be involved online only (38% overall among our nonbelieving sample). Last, only a small fraction of nonbelievers participate in organized atheist, humanist or secularist meetings or activities at least once a year (10% overall among our nonbelieving young adult sample). In turn, most of this latter group of organized nonbelievers, 70% of them, are also involved with their nonbelief online in some way. These findings reinforce research elsewhere that found Millennials are prone to more "passive" modes of involvement (i.e., digital consumer) versus "active" contributions to group activities, both on-and offline (Environics 2017).
Although these indicators of organized and digital nonbelief activities are limited in number in the present study, and also have their limits in terms of measurement, this research does provide a first series of representative estimates of the prevalence of these activities. Further representative surveys should be run in future to continue estimating the occurrence of nonbelief views and behavior among various populations, as well as with a larger array of indicators.

Impact of Religious Socialization on Organized and Digital Nonbelief Activities
What factors are driving some nonbelievers to be involved with organized or digital atheism, humanism or secularism? In what ways are many of those who take part in these activities distinct from many of those who do not? To what extent are organized and digital nonreligious activities more appealing to nonbelievers who come from more religious backgrounds? Are nonbelievers who experienced a religious socialization growing up more likely to face discrimination, and respond to it by taking part in atheist, humanist and secularist activities and spaces? Are they also seeking vestiges of their previous religious community life in these nonreligious groups? Table 3 contains the results (in odds ratios) of the three ordered logit models measuring the associations between sociodemographic, worldview and religious socialization predictors, and the outcomes of 1) frequency of taking part in organized nonbelief meetings or activities (collapsed into four categories: weekly, monthly, annually or not at all); 2) frequency of posting on social media about nonbelief (collapsed into five categories: daily, weekly, monthly, annually or not at all); and 3) frequency of consuming digital nonbelief content (collapsed into five categories: daily, weekly, monthly, annually or not at all).

Sociodemographics
Among the significant sociodemographic effects, we first find that older nonbelieving Millennials are less likely to consume digital nonbelief content than their younger counterparts. For example, 35-year-old nonbelievers have a 9% predicted probability of consuming digital nonbelief content at least once a month, compared with a 17% probability among 18-year-old nonbelievers. This could be due to a stage of life effect: young adults in their 30s are busier with their careers and families, whereas many teenagers and those in their early 20s may be more concerned with figuring out who they are and what they believe in (see Smith and Denton 2005;Smith and Snell 2009;Wuthnow 2007). These age variations could also be attributed to younger Millennials who grew up with technological access at an earlier age compared to older Millennials, thus consuming digital nonbelief content is more normalized and natural to them.
Existing research points to a gender divide for taking part in organized forms of atheism, humanism and secularism (Langston et al. 2017). We find evidence for a modest gender difference among young adult nonbelievers in their likelihood of participating in these activities once other sociodemographics, nonbelieving worldviews and (non) religious socialization are controlled for. Male young adult nonbelievers have a 10% predicted probability of taking part in organized nonbelief activities at least once a year, compared with a 6% probability among female or other gendered nonbelievers. 11 This gender divide may, however, be more pronounced among older generations. Future research will have to investigate this relationship further.
As initially found with the descriptive statistics in the previous section, Canadian young adult nonbelievers are significantly less likely to take part in organized nonbelief activities and to post on social media about nonbelief than their USA counterparts. This country of residence effect is not, however, statistically significant for digital nonbelief consumption once sociodemographics, nonbelieving worldview and (non)religious socialization are controlled for.
Also important to note is that level of education, rural residence and generation of immigration are not statistically significant predictors of taking part in organized or digital nonbelief activities among young adult nonbelievers. Although these factors are often tied to whether or not individuals are nonbelievers in the first place (see for example Baker and Smith 2015; Zuckerman, Galen & Pasquale 2016), they do not seem to play as large a role when we compare more or less involved forms of nonbelief.

Nonbelieving Worldviews
Those who believe in a mysterious and connected universe and natural world are more likely to take part in organized nonbelief activities (11% probability of annual or more frequent participation) than young adults who see life on Earth as purely the result of complex biological, physical and material processes (6% probability of annual or more frequent participation). This distinction is another indication that many of those involved with organized nonbelief are not necessarily staunch atheists with a purely scientific or material worldview.
In turn, young adults who say they are unsure about their beliefs are less likely to post about nonbelief online and consume nonbelief digital content than those with a more scientific and material worldview. For example, those with an agnostic worldview have a 16% predicted probability of posting about nonbelief on social media at least once a year, compared with a 24% probability among those with a scientific and material worldview. Additionally, young adults who are indifferent about beliefs are also less likely to consume digital nonbelief content than those with a more scientific and material worldview: the indifferents have a 6% predicted probability of consuming digital nonbelief content at least once a month, compared with a 16% probability among those with a more scientific and material worldview.

Religious Socialization
A number of religious socialization variables were also included in the ologit models in Table 3, in order to begin addressing H 1 while isolating any potential associations from respondents' sociodemographics and nonbelieving worldviews. Having a religiously affiliated mother shows negative associations with all three organized and digital nonbelief outcomes. However, none of these associations are statistically significant at the 95% or higher level in the models. Coming from a mixed (non)religious parental background (parents from different faith traditions, or one parent was nonreligious) also has no statistically significant effect among nonbelieving young adults. Although having mixed parents has been shown in existing research to be linked with greater chances of having no religion as an adult (Arweck and Nesbitt 2010;Crockett and Voas 2006;Zuckerman 2012), among nonbelievers it does not seem to impact levels of organized and digital nonbelief involvement.
Interestingly, and in contrast to the maternal religion predictor, having a father who is religiously affiliated shows signs of being positively linked with all three forms of organized and digital nonbelief outcomes among nonbelieving young adults. This said, only the link between father's religious affiliation and digital consumption of nonbelief content reaches statistical significance at the Table 3: Effects (in odds ratios) on frequency of taking part in organized and digital nonbelief activities in past 12 months, respondents 18-35 years old who do not believe in God or a higher power, 2019. Notes: Millennial Trends Survey 2019. Ordered logit (ologit) models. † = p ≤ .10; * = p ≤ .05; ** = p ≤ .01; *** = p ≤ .001.
Wilkins-Laflamme and Thiessen: Religious Socialization and Millennial Involvement in Organized and Digital Nonbelief Activities Art. 2, page 10 of 15 95% or higher level: nonbelievers who had a religiously affiliated father growing up have a 45% predicted probability of consuming digital nonbelief content at least once a year as adults, compared with a 36% probability among those who had a religiously unaffiliated father or no father present at all. Although literature points to the importance of a mother's level of religiosity for passing on religion to her child (Bengtson, Putney & Harris 2013;Collins-Mayo and Dandelion 2010), the results here point to a potentially more antagonistic effect of father's religiosity among children who leave their parent's faith. Having a religiously affiliated father seems to drive more digital nonbelief involvement on average among nonbelievers. Nevertheless, the socialization variable that shows the largest impact on our three organized and digital nonbelief outcomes, once other predictors are controlled for in the models, is respondents' frequency of religious or spiritual education when growing up. In the MTS, respondents were asked: "Growing up as a child between the ages of 5-12 years old, how often on average did you receive some form of religious or spiritual education at school, at home, or at a place of worship?" Answers are collapsed into five categories for the purposes of this study: at least once a day, at least once a week, at least once a month, at least once a year, or not at all. This socialization variable is significantly and positively associated with taking part in organized nonbelief activities as well as consuming digital nonbelief content among nonbelieving young adults. This confirms H 1 for these two organized and digital nonbelief outcomes: individuals who were raised with frequent religious contact and education, but who then left the faith as young adults, are more likely to be involved with organized and digital forms of nonbelief. There are also signs of a positive association with nonbelief social media posting, but this effect does not reach statistical significance at the 95% of higher level in this model. Figure 3 illustrates some predicted probabilities of these associations, along with their 95% confidence intervals. 12 Figure 4 in turn illustrates key results from the gsem path model, measuring the impact of the nonreligious discrimination mediator variable in the association between frequency of religious education while growing up and frequency of organized and digital nonbelief activities among young adult nonbelievers. 13 First, those who received a more religious or spiritual upbringing, but who have since left their faith and have become nonbelievers, are more likely to say they have experienced discrimination in the 12 months prior to the survey than the cradle nonreligious. For example, nonbelievers who received a religious or spiritual education at least once a week growing up have a 26% predicted probability of experiencing discrimination once a year or more due to their nonreligious views, compared with a 14% probability among nonbelievers who did not receive any religious or spiritual education as children. This discrimination is most likely coming from the remaining religious social networks and social environments surrounding many nonbelievers who were raised religiously.
Those nonbelievers who experience frequent discrimination due to their nonreligion are in turn more likely to take part in organized nonbelief activities, to post on social media about nonbelief, and to consume digital nonbelief content. Nonbelievers who say they experience discrimination at least once a month due to their nonreligious views have a 20% probability of being involved monthly or more with organized nonbelief activities, compared  Table 3 for complete results. with a 2% probability among nonbelievers who did not experience discrimination in the 12 months prior to the survey. Nonbelievers who say they experience monthly or more frequent discrimination have a 39% probability of posting at least once a month about nonbelief on social media, compared with a 7% probability among nonbelievers who did not experience discrimination. Monthly or more frequently discriminated nonbelievers also have a 42% probability of consuming digital nonbelief content at least once a month, compared with a 10% probability among non-discriminated nonbelievers.
These results in Figure 4 support our H 2 subcultural identity hypothesis: nonbelievers with a religious upbringing are more likely to face stigma for their nonbelief as adults, likely from within their social networks and environments, and experiencing this discrimination is in turn linked to greater involvement with organized and digital forms of nonbelief. This also means that cradle nonbelievers who experience discrimination from their surrounding environment due to their nonreligious views as young adults are more likely to take part in organized and digital nonbelief activities as well. However, as indicated in Figure 4, cradle nonbelievers are less likely to experience this discrimination in the first place.
Additional feedback loops could also be at play here: as they become involved with organized and digital forms of nonbelief, these nonbelievers may redefine some of their experiences with others as discriminatory. Their more visible "out" forms of nonbelief may also attract greater discrimination from their surrounding environment. The cross-sectional nature of the MTS data does not allow us to test which direction of causality is more at play here, but we argue that all of these mechanisms are most likely present. Future research with longitudinal data is needed to better tease out the exact nature of these relationships.
Two of the remaining three direct pathways between frequency of religious education as a child and organized and digital nonbelief activities remain statistically significant and positive, once frequency of nonreligious discrimination is controlled for in the gsem. Nonbelievers who were religiously socialized as children are still more likely to take part in organized nonbelief activities as young adults, regardless of experiencing nonreligious discrimination or not in the year prior to the survey. The same goes for the link between religious education as a child and digital consumption of nonbelief content as a young adult. Although weaker than when frequency of nonreligious discrimination is left out of the gsem, 14 these two direct associations remain present even once nonreligious discrimination is included.
This lends support to our H 3 vestiges of religious life hypothesis in these two instances: once experiencing discrimination is controlled for, religious education remains directly tied to greater involvement in organized and digital forms of nonbelief among young adult nonbelievers. These individuals are potentially seeking out similar forms of community and practice that they grew up with, despite them now being nonbelievers. Given the existing research, we argue here that the vestiges of religious life mechanism is the most likely explanation

Conclusion
This study provides a first series of representative estimates of the prevalence of organized and digital nonbelief activities among young adult populations in the USA and Canada. Although a number of nonbelievers take part in these activities, we also find that many more do not. Leading explanations as to why only a small portion of the nonreligious are involved in these organized and digital nonbelief activities may include the individualist ethos commonly found among nonreligious individuals. Being more focused on values of individualism such as independence and self-achievement drives many of the nonreligious, especially among younger generations, to be removed from nonbelief organizations and many other forms of organized life in civil society (Beyerlein and Vaisey 2013;Saroglou, Delpierre & Dernelle 2004;Woodhead 2016;Zuckerman, Galen & Pasquale 2016: 203-207). Further, atheist, humanist and secularist organizations lack many features found in religious groups, such as, "a definite and positive ideology, a centralised and formalised organisational structure, a clear system of authority, a formal procedure for resolving disputes, a gemeinschaftlich atmosphere and a permanent and loyal group of members" (Campbell 1971: 42). On the surface, nonreligious groups like the Sunday Assembly seem to embrace many of these features, yet only time and more focused research will help us to understand how effective or ineffective such organizational structures are for nonreligious groups to be sustainable and appeal to larger numbers of nonbelievers moving forward.
In line with subcultural identity theory, those in contexts where nonreligion is normalized and socially accepted do not feel as much of a need to be involved with nonbelief organizations. This may also have implications for our findings: on top of being less involved in organizations in general, our adult Millennial sample may also have some of the lowest levels of participation in organized nonbelief because nonreligion is simply less contentious among many members of this younger generation. Consequently, many nonbelieving young adults may feel less of a need to carve out and affirm a distinct nonreligious identity in light of pressures from hostile religious environments.
Digital and social media nonbelief posting, reading and watching are found in turn to be more prevalent among young adult populations than organized forms of atheism, humanism and secularism. Nevertheless, consistent with other research documenting online activities (Environics 2017: 45-46), Cimino and Smith (2014) intimate that such online (and subsequent off-line) ties among the nonreligious are fairly weak and diffuse. Although they can provide important digital spaces and networks for atheist activism and community building, Cimino and Smith also argue that the nonreligious tend selectively to drop in and out of communities as they please, with low levels of longterm commitment.
Yet for some nonbelievers, notably those coming from a more religious upbringing and who experience discrimination as young adults due to their nonreligious views, these organized and digital nonbelief activities are an important and regular part of life. One of the main contributions of this research is to show that organized and digital forms of nonbelief seem to be especially popular for those less or nonreligious individuals who had a stronger religious socialization growing up (confirming H 1 ). Religious socialization during childhood has been shown repeatedly to be tied to a higher likelihood of remaining an active religious believer as an adult (see for example Bengtson, Putney & Harris 2013;Collins-Mayo and Dandelion 2010). However, when this is not the case religious socialization may lead newly nonreligious individuals to more involved forms of organized nonbelief, in order to fight back against the perceived negative social and political impact of religion and to find a safe space to "come out" as nonreligious (confirming H 2 subcultural identity theory). These individuals may also seek to find an alternative source of community and to continue group discussions and rituals surrounding worldviews and values, even if these have changed since their childhood (confirming H 3 vestiges of religious life). The appeal of these organized and digital nonbelief activities for newly disaffiliated individuals may also be why our study observed interest for these groups among some believers, who may be considering leaving their religion. By contrast, nonreligious socialization seems instead often to lead to fewer experiences of discrimination and more disinterest in religious, atheist, humanist and secularist activities alike.
These findings have real implications for the future of organized nonreligious groups. If these groups currently appeal more to those going through the transitional phase of disaffiliation, how do they continue to flourish in contexts where nonreligion becomes less stigmatized, and growing numbers of individuals are raised without religion and do not care as much about these issues? If the shared goal of all these organizations is to normalize nonreligion in society (Cragun and Manning 2017), what happens when that goal is achieved? Do they sow the seeds of their own destruction in this sense? Canada, which we found has lower rates of participation in organized and digital nonbelief activities among its younger adult population, is an interesting example of this for nonbelief groups.
This said, the complete normalization of nonreligion in the USA, and even in parts of Canada and Europe, has not yet been achieved. Religious socialization accompanied by later adult disaffiliation is still common in North America; and there remains real or perceived external threats (especially in many non-Western nations) as well as internal threats to nonreligion's social acceptability and Church/ State separation, even in areas where religious individuals are becoming a minority (such as in Great Britain). As the (non)religious landscape evolves, notably with the sharp rise in rates of no religion in the USA since the 1990s