Chapter 14 of 16

Notes

Notes

Introduction

Richard Flory. Throughout the book we use “we,” “us,” and “our” when referencing interviews with emerging adults unless otherwise noted. We use pseudonyms for all of our interview participants.

See, for example, Pew Research Center, “Religion among the Millennials”; and  ”

See, for example, ; and

On the former, see, for example, On “spiritual but not religious,” see ; and .

. For a similarly positive view of “millennial” evangelical Christians, see Bevins, Ever Ancient, Ever New. For a positive view of religious decline among evangelical young people, see Stanton, Myth of the Dying Church.

Clydesdale and Garces-Foley, The Twentysomething Soul, pp. 11–13.

Pew Research Center, “Religion among the Millennials.”

Smith and Cooperman, “The Factors Driving the Growth of Religious ‘Nones.’ ”

See Appendix for details on each wave of data collection.

Smith and Denton, Soul Searching.

Pearce and Denton, A Faith of Their Own.

Smith and Snell, Souls in Transition.

Arnett, Emerging Adulthood.

Arnett, Emerging Adulthood.

Flory and Miller, Finding Faith. See also

Cf., Arnett and Jensen, “A Congregation of One.”

Throughout the book, we use the masculine pronoun for God for convenience of writing, but it is important to keep in mind the multiple ways that emerging adults expressed how they conceptualized God, whether as female, male, plural, or not embodied.

Chapter 1

Melinda Denton.

Smith and Snell, Souls in Transition.

Chapter 2

For a review of this literature, see Arnett, Emerging Adulthood. For a classic study of the transition of youth to adulthood in the United States, see Modell, Into One’s Own.

Arnett, Emerging Adulthood, chapter 13.

;

Smith and Snell, Souls in Transition.

The higher reported rates of giving and volunteering among our respondents may be explained by “social desirability bias,” in which respondents overestimate their volunteering or financial donations.

US Department of Labor, Bureau of Labor Statistics, “Volunteering in the United States, 2015.”

On financial giving, Smith and Davidson, The Paradox of Generosity.

It is important to note that the NSYR Wave 4 data was collected before the 2016 elections, and with the results from that election, the responses from our respondents in regard to their political perspectives could potentially be different—either more liberal or more conservative. However, based on what else we know about emerging adults through our work, we do not believe that the results would be significantly different than how they responded to our survey and interview questions. In fact, regardless where they placed themselves on the political spectrum, they likely would have responded much the same way even after the 2016 election; that politicians are mainly out for themselves and have little regard for creating actual positive social change.

See Smith and Denton, Soul Searching, chapter 4, “God, Religion, Whatever: On Moralistic Therapeutic Deism.”

; also .

On the role of inarticulate or “tacit” knowledge in cultural life and how it motivates action, see

Chapter 3

See, for example, Pew Research Center, “Nones on the Rise”; Lipka, “Millennials Increasingly Are Driving Growth of ‘Nones’ ”; Pew Research Center, “Religion among the Millennials.”

TRADREL is the variable categorizing respondents into major religious types. It is the combination of RELIG1 and RELATT, but it always prioritizes attendance. So, in essence, this variable starts as an exact replication of RELATT, but then uses RELIG1 to classify all those who could not be classified based on attendance (i.e., nonattenders and indeterminate). TRADREL is similar to the RELTRAD variable from Wave 1, with slight differences. First, TRADREL always prioritizes named attendance. Next, it splits the “no affiliation” group into two separate categories: not religious (respondents who do not attend and do not self-identify with a religion) and indeterminate (respondents who participate in or identify with a group that cannot be categorized, e.g., “just Christian”). Additionally, we were able to use prprot1 to help categorize indeterminate individuals who answered this question. The variables used to determine TRADREL included: ATTREG, CHURTYPE, BAPTIST, METHST, PRSBIAN, LUTHAN, REFMED, CONGAL, CHCHST, RELIG0, RELIG0A, RELIG0B, ATHEIST1, HALF1, OTHREL1, PRPORT, and TEENRACE.

Steensland et al., “The Measure of American Religion.”

Blumberg, “American Religion”; Lipka, “Religious ‘Nones’ Are Not Only Growing.”

Just more than half of the Indeterminate “just Christian” respondents do not regularly attend religious services, while the other half reported that they do attend services at a religious congregation.

Black Protestant loss may be due in part to sample attrition over the 10 years of the project.

Religious switching percentages for the entire sample (Waves 1–4): same religious affiliation = 48 percent; switch from one religious affiliation to another = 11 percent; switch from specific religious affiliation to Indeterminate affiliation = 12 percent; switch from religious affiliation to Not Religious = 26 percent; switch from Not Religious to Indeterminate affiliation = 1 percent; switch from Not Religious to religious affiliation = 2 percent.

Smith and Denton, Soul Searching.

Smith and Denton, Soul Searching; Pearce and Denton, A Faith of Their Own; Smith and Snell, Souls in Transition.

See in this chapter for a description of how the religious tradition variable was constructed.

We chose to exclude Wave 2 data from this analysis because we are particularly interested in understanding the changes across the 10 years of the study. Wave 2, while interesting in its own right, was collected two years after Wave 1 and thus does not provide a large number of years to add significant information to this analysis.

Specifically those who were identified with that tradition at Wave 1. Not all would have necessarily been raised in that particular tradition, but this is the case for most.

Religious service attendance is subject to known social desirability bias that leads to overreporting of attendance frequency, and these percentages should be read within that context (see Hadaway and Marler, “How Many Americans Attend Worship Each Week?”; and ). Assuming that the overreporting bias persists over time, we can learn much from the relative differences in attendance reports across the time frame of the study.

For the Wave 4 survey, the majority of respondents were given the option to complete their survey online instead of over the telephone. In the original telephone survey (Waves 1–3), the none of these views option was not read to respondents but was available if they volunteered this answer. In Wave 4, this option was visible in the online version of the survey and, for consistency across formats, was also read to respondents who completed the survey by phone. The presentation of this option likely explains much of the increase in this response category.

Pearce and Denton, A Faith of Their Own.

Smith and Denton, Soul Searching; .”

Pearce and Denton, A Faith of Their Own.

For example, Kaya Oakes, The Nones Are Alright; Linda Mercadante, Belief without Borders; Elizabeth Drescher, Choosing Our Religion.

See Flory and Miller, Finding Faith; for a personal account from a former Conservative Protestant emerging adult that also includes the appeal of more open cultural and political perspective of Mainline Protestantism, see .

See Smith and Snell, Souls in Transition, for a discussion of declines among Mainline Protestants.

For the Wave 4 survey, the majority of respondents were given the option to complete their survey online instead of over the telephone. In the original telephone survey (Waves 1–3), the I Don’t Know option was not read to respondents but was available if they volunteered this answer. In Wave 4, this option was visible in the online version of the survey. The presentation of this option likely explains some of the increase in this response category.

Smith and Snell, Souls in Transition.

Chapter 4

Pew Research Center, “ ‘Nones’ on the Rise.”

See, for example, Lipka, “A Closer Look at America’s Rapidly Growing Religious ‘Nones’ ”; ; ; .

By “congregation” we include all meeting places for religious groups such as churches, temples, synagogues, mosques, and the like.

Previous books from the National Study of Youth and Religion (NSYR) have developed different typologies of youth and emerging adult religion that describe how the young people in this study expressed how they were thinking about religion at the time of each wave of data collection. These typologies ranged from four types () to five types (), to six types (), in which each described the different levels and types of religious commitment and belief that these young people were expressing at that point in their lives. Similarly, Arnett (Emerging Adulthood) used Wave 3 NSYR data to develop a four-part typology that ranged from most religious to least religious. We took these different typologies into account in developing our typology, while also allowing our categories to emerge from the qualitative and quantitative data. The result is a similar set of categories that allow for understanding where emerging adults are now in how they express their religious identity, commitments, and involvements.

There are a handful of respondents who do not claim any religious identity, yet do attend religious services with some regularity. In these cases, respondents are coded as Marginal or Committed based on patterns of religious service attendance.

Lim, McGregor, and Putnam, “Secular and Liminal”; Pew Research Center, “ ‘Nones’ on the Rise”; Lipka, “A Closer Look at America’s Rapidly Growing Religious ‘Nones’ ”; ;

See, for example, Edgell, Hartmann, and Gerteis, “Atheists as Other”; , pp. 42–45.

Pearce and Denton, A Faith of Their Own.

, p. 28.

Lim, McGregor, and Putnam, “Secular and Liminal”; ; ; .

Burton, “Spiritual but Not Religious”; Oppenheimer, “Examining the Growth of the ‘Spiritual but Not Religious’ ”; Barrie-Anthony, “Spiritual but Not Religious.”

Chapter 5

Growth mixture modeling (GMM) is a variant of latent growth curve analysis. Instead of making the assumption that individual trajectories are deviations from one grand trajectory, GMM assumes that multiple normative trajectories better characterize the data (or, alternatively, that we are sampling from distinct subpopulations that have qualitatively distinct growth patterns). These different classes of trajectories are a latent function of the data, so no a priori assumptions concerning how individuals change religiously are necessary.

There are two common ways to present GMM results graphically: one relies on the statistical estimates based on the coefficients, and the other on the actual means for each group. We choose to present the actual group means because they are a more natural comparison to the overall means presented in . The estimated growth curves are very similar to the group means.

Because this is a combination of three measures, each measured from zero to four, we considered a shift in three to be roughly equivalent to moving a single category in each of the three measures (even though we realize this isn’t the only way to see a change of three in our composite measure). In our judgment, this is a fairly small shift in religious faith. The standard deviation for this composite measure varies by survey wave, but it is always greater than three (it varies from 3.2 to 3.9). So, even judging by the distribution of scores, these individual shifts are fairly small.

Although these figures do not include individual tests for whether one growth curve group is significantly different from every other group in that figure, this can be approximated by visually estimating whether the error bars of one group overlap the point estimate of another group. It’s important to note that error bars from two groups that “cross” one another do not necessarily mean these groups are not statistically different. Rather, it is the error bar from one group compared with the specific percentage estimate of the other group that matters.

Socioeconomic status (SES) is measured by the income and educational attainment of parent(s). High SES is defined as having at least one parent who graduated with a bachelor’s degree or higher and a household income of $80,000 or more in 2002–3.

; Verba, Schlozman, and Brady, Voice and Equality; Cnaan and DiIulio, The Invisible Caring Hand; .

Norris and Inglehart, Sacred and Secular; .

To investigate this, we ran logistic regression models predicting being in the Stable High category versus the Shallow Decline or Steep Decline categories. The odds ratio of Conservative Protestant to Mainline Protestant was 2.02 without any controls and 1.55 after controlling for religious talk in the household. The odds ratio of Conservative Protestant to Catholic was 2.13 without any controls and 1.67 after controlling for religious talk in the household. The equivalent odds ratios for Black Protestant to Mainline Protestant was 3.41 without controls and 2.43 with the additional measure of religious talk. Black Protestant to Catholic was 3.59 without controls and 2.61 with the religious talk variable. In all cases, a substantial portion of the differences between religious traditions can be accounted for by the measure of the frequency of religious talk.

The index contains measures of how often the father and mother praise and encourage, hug, tell the teen they love them, talk about personal subjects, and “hang out” with the teen. It also includes measures of how well the teen and parent get along, how close the teen feels to his or her mother and father (or similar figures), and how close the parent feels to his or her teen. Last, it includes measures of whether the parent has taken the teen to a museum, a play, or a library; worked on a project together; or played a game or sport together in the past six months. Thank you to Kyle Longest for providing the coding for this index.

This index contains measures of how often the teen reported that his or her parent figure knows what music, television, and movies are watched, whom he or she spends time with, and what he or she is actually doing when not at home. It also includes the parents’ report of how often they monitor the teen’s television and movie watching, along with the teen’s Internet use. Again, thanks to Kyle Longest for providing the coding for this index.

Wilcox and Robinson, Onward Christian Soldiers?

Chapter 6

Wilcox, Chaves, and Franz, “Focused on the Family?”;

US Census Bureau, “Median Age at First Marriage.”

.

Collapsed data from the 2012 and 2014 General Social Survey indicates that 30 percent of young people ages 25 to 28 are currently married, while 65 percent have never married. GSS 2012–14 calculated using “GSS 1972–2016 Cumulative Datafile.”

; ?”; ; Amato et al., “Precursors of Young Women’s Family Formation Pathways.”

Manning, Brown, and Payne, “Two Decades of Stability and Change in Age at First Union.”

;

In a multivariate model that controls for demographic factors, the Committed respondents reported an ideal age of marriage that is lower than each of the other groups by a statistically significant margin. Multivariate models are available from the author.

; ;

.

In multivariate regression models, differences in reports of disappointment about never getting married are statistically significant for Disaffiliated, Marginal, and Committed compared with Not Religious.

; Rostosky, Wilcox, Wright, and Randall, “The Impact of Religiosity on Adolescent Sexual Behavior”; Kirby, “Reflections on Two Decades of Research.”

In a multivariate logistic regression model, those who are not willing to cohabit in the future are 1.85 times as likely to be married at Wave 4 than those who are willing to cohabit (odds ratio = 1.85, p < 0.000).

In a multivariate logistic regression model, those who have never cohabited are 2.55 times as likely to be married at Wave 4 than those who have cohabited at some point (odds ratio = 2.55, p < 0.000).

Amato et al., “Precursors of Young Women’s Family Formation Pathways.”

The majority (72 percent) of the cohabiting and single with children respondents have never been married. The remaining 28 percent reported being previously married or currently in a civil union partnership.

Multivariate models reveal statistically significant differences in the family status of respondents in the very/extremely important category as teenagers compared with those in the not important category.

Of those Conservative Protestants who are currently unmarried without children, 3 percent have been previously married or are in a civil union, leaving just 41 percent who have not experienced any family transition.

In multivariate analyses, however, there is not a statistically significant difference between the family statuses of Mainline Protestant and Jewish respondents when compared with Conservative Protestants.

The statistics for the Black Protestants are more likely a reflection of racial patterns in marriage and childbearing than something distinct about the religious ideology of Black Protestants. In multivariate analyses that control for race, there are no statistically significant differences between Black Protestants and other religious groups.

In multivariate analysis, both Catholic and Not Religious respondents are significantly less likely to be married than their Conservative Protestant peers.

Pew Research Center, Religious Landscape Study: Marital Status.

Multivariate analysis reveals statistically significant differences in family status between Marginal and Not Religious and between Committed and Not Religious. There are no statistically significant family status differences between the Disaffiliated and Not Religious respondents.

; ; ; Amato et al., “Precursors of Young Women’s Family Formation Pathways.”

Uecker, Maryl, and Stroppe, “Family Formation and Returning to Institutional Religion in Young Adulthood”;

In a multivariate model, respondents who are cohabiting or single without children reported significantly less importance of faith compared with the married with children group.

Multinomial logistic regression was used to predict the odds of becoming more or less religious compared with staying the same religiously. Using married with children as the family status reference category, all of the other groups have significantly lower odds of saying that they became more religious over time, with those cohabiting without children having the lowest odds of becoming more religious. The odds ratios for reporting becoming less religious over time (as opposed to staying the same) were significant for the cohabiting without children group and the single without children group when each was compared with the married with children group. In other words, individuals who have not yet transitioned into a family-of-formation status were the most likely to report becoming less religious during the years of emerging adulthood. See also

Uecker, Maryl, and Stroope, “Family Formation and Returning to Institutional Religion in Young Adulthood”;

Ashley and Carla were both interviewed by Melinda Denton.

Conclusion

Mercadante, Belief without Borders; Oakes, The Nones Are Alright; Clydesdale and Garces-Foley, The Twentysomething Soul;

.

See the Pew Research Center report, “U.S. Religious Knowledge Survey”; see also .

See , pp. 198–199 on “The Default Religious Laxity Tendency.”

; See also Vaisey and Lizardo, “Cultural Worldviews.”

See See also Bellah et al., Habits of the Heart on “Sheilaism.” What we are describing among emerging adults is different from what Bellah and colleagues described as “Sheilaism,” in that Sheila was actively pursuing her own religion, while emerging adults are on average not engaged in such a pursuit and are not interested in creating their own religion so much as simply accepting, rejecting, or ignoring those elements of religion that may or may not suit their particular interests.

We do not have a survey question for belief in karma, so we don’t know how many emerging adults in our study would say that they believe in karma. However, in our in-person interviews, karma and associated similar concepts were common across the majority of emerging adults in the study.

.

; Bengtson, Putney, and Harris, Families and Faith.

.

Pearce and Denton, A Faith of Their Own.

; Arroyo, Payne, Brown, and Manning, “Median Age at First Marriage.”

.

; Uecker, Maryl, and Stroope, “Family Formation and Returning to Institutional Religion in Young Adulthood.”

For examples of “reimagining religion,” see USC Center for Religion and Civic Culture, Reimagining Religion Conference.

Christerson and Flory, The Rise of Network Christianity.

↑ Back to top

↑