Transmission of Faith in Families: The Influence of Religious Ideology

Abstract

The family and denominational factors influencing intergenerational religious transmission have been examined in a substantial body of work. Despite research identifying religious ideology as a salient aspect of American religion, however, its role in religious transmission remains unexplored. In this study, I use the National Study of Youth and Religion to test whether children’s worship attendance and centrality of faith in young adulthood differ based on whether their parents identify as religiously liberal, moderate, conservative, or none of these. I further test whether the strength of the relationship between parent and child religiosity differs between ideological groups. The primary finding is that religious transmission is stronger among children of religious conservatives than for any other group, while the other groups do not differ significantly from one another. These differences in transmission are largely explained by religious conservative parenting approaches, congregational involvement, and most importantly, more intensive religious socialization.

INTRODUCTION

A substantial body of literature confirms a strong relationship between religious beliefs and practices of parents and those of their adult children, often identified as religious transmission. The strength of this transmission varies by family factors, such as parent–child closeness, parenting style, and family structure (e.g., Myers 1996; Smith and Snell 2009). Children’s religious outcomes also differ by parental religious tradition (Vaidyanathan 2011). Families from denominations which are relatively strict, have high mean levels of worship attendance, or place a strong emphasis on close-knit community and family cohesion tend to produce more religious children (Armet 2009; Bengtson 2017).

Prior research has not examined whether the strength of transmission varies according to parental religious ideology. Religious ideology, as viewed on a liberal-to-conservative spectrum, is distinct from (though not unrelated to) other aspects of religion such as behavior, intensity, or denominational affiliation. Religious ideology has been the subject of ample and diverse theorization, with scholarly definitions focusing on such factors as beliefs about the nature of God or the bible (Davis and Robinson 1996a; Froese and Bader 2007), the appropriate role of religious influence in public life (Wuthnow 1988), or tension with the larger society (Armet 2009; Stark and Finke 2000;). Drawing from these accounts, I define religious ideology here in terms of shared sets of beliefs regarding the nature of religious authority and its appropriate role in defining morality as well as social norms and policies.

There is strong reason to believe parental religious ideology plays a role in shaping the strength of religious transmission. Influential scholarship has proposed that, in a “restructuring” of American religion, religious ideological distinctions have eclipsed denominational boundaries in social relevance (Hunter 1991; Wuthnow 1988), which, I argue here, has implications for religious identity formation and socialization of children. Parents’ religious ideology is likely to influence both the content and intensity of religious socialization, and thus may serve to either reinforce or undermine their children’s religious practice later in life.

In this study, I test this proposition by examining differences in religious transmission between four groups: children whose parents report being religious conservatives, moderates, or liberals, as well as those who do not identify any religious ideology. Using longitudinal data from the National Study of Youth and Religion (NSYR), I focus on parental influences on children’s frequency of worship attendance and self-reported importance of faith in young adulthood. I first examine whether children’s religious outcomes differ by parental religious ideology, even when controlling for other key predictors (main effects). Second, I test whether the relationship between parent religiosity and children’s religious outcomes differs according to parental religious ideology (interaction effects). Finally, I examine several possible mechanisms which might explain how differences in parent religious ideology lead to differential transmission.

As expected, I find that the strength of transmission differs by parent religious ideology. Specifically, transmission is significantly stronger for children of religious conservatives than for any other group. This effect is distinct from that of either denomination or political ideology. Differential religious transmission is largely though not entirely explained by religious conservative parenting approaches, congregational involvement, and most importantly, more intensive religious socialization within the family.

RELIGIOUS TRANSMISSION

Parental religiosity is consistently found to be the single strongest predictor of child religiosity over the life course (e.g., Myers 1996; Smith and Snell 2009). Parents who are closer to their children exert a stronger influence on their children’s religious lives. This is especially true when parents are high on multiple dimensions of religiosity, and when parents are religiously similar to one another (Bader et al. 2006; Myers 1996; Voas and Storm 2012). Religious transmission is strengthened by use of authoritative parenting styles as well as clear and frequent communication of religious beliefs to children, maintained consistently over time (Acock and Bengtson 1980; Dudley and Wisbey 2000). Families who have not experienced divorce or excessive marital conflict see higher religious continuity (Bengtson 2017). Together, these findings indicate the importance of relational cohesiveness and religious consistency within the family in promoting effective transmission.

Parental influence on children’s religiosity also varies by religious tradition. Religious continuity is especially high among groups with strong prescriptive norms and high cultural boundaries, such as Latter-Day Saints and evangelicals (Armet 2009; Smith and Snell 2009). Transmission occurs more strongly among groups with a “quasi-ethnic” aspect in which religion may be tied to other forms of identity and carry significance in multiple domains, such as Jews or Black Protestants (Bengtson 2017; Vaidyanathan 2011). In contrast, denominations with lower tension with the larger society and lower mean levels of participation see greater religious decline over time (Iannaccone 1994; Smith and Snell 2009).

The present body of research on religious transmission is largely focused on the family or denominational factors described above. This literature, however, has not examined transmission explicitly in the context of religious restructuring and polarization, in which the most salient religious boundaries are defined less by denomination than by religious ideology.

CONCEPTUALIZING RELIGIOUS IDEOLOGY

A generation of scholarship has emphasized the increasing social relevance of division between religious conservatives and liberals. In The Restructuring of American Religion (1988), Robert Wuthnow advanced the argument that distinctions between religious liberals and conservatives have largely eclipsed those between different denominations in terms of theological, social, moral, and political characteristics. This proposition has been echoed in other famous works on the American religious landscape (e.g., Hunter 1991; Putnam and Campbell 2010), and Steensland et al. (2000), in their seminal article on religious classification, called for a greater focus on differences between conservatives and liberals within denominations as well as between them. There is substantial empirical support for the existence of an ideological dimension that cuts across boundaries of religious traditions and denominations (e.g., Davis and Robinson 1996a, 1999; Starks and Robinson 2007, 2009; Wuthnow 1996), but this insight has been unevenly incorporated into the sociological study of religion.

I define religious ideology here as a shared set of interrelated beliefs, opinions, and values (Boutyline and Vaisey 2017), explicitly derived from or justified by a conception of a superhuman power (most commonly identified as [a] god) (Smith 2017), which provides an interpretive and prescriptive moral framework for understanding events, human actions, and social systems. These beliefs will range from the abstract to the concrete, but importantly, a religious ideology is in all cases embedded in a particular historical, social, and political context which shapes its development and manifestation (Wellman 2008). A religious ideology may involve elements both proximally and distally linked to a set of religious beliefs, but some religious justification must be present for a particular belief to constitute part of a religious ideology.

In the context of the United States, much has been written about the differences between religious liberals and conservatives, which can be synthesized to identify broad ideological profiles. An ideal-typical religious conservative will believe in a personally engaged God (Froese and Bader 2007) exercising ultimate and immutable moral authority (Hunter 1991; Wuthnow 1996), whose teachings can be known primarily through the medium of religious sources (e.g., religious bodies, sacred texts) (Starks and Robinson 2007, 2009). These teachings should serve as a foundation for society’s moral norms, and be reflected and acknowledged in public institutions (Hunter 1991; Wuthnow 1988). Conversely, an ideal-typical religious liberal will emphasize the personalized, pluralistic, and dynamic nature of the experience of a higher power and of religious truth (Hunter 1991), and thus disfavor the direct influence of religious authority on governing institutions (Krull 2020). Rather, religion in its organized or public form should serve as a source of inclusive community (Konieczny 2013), which prioritizes spiritual exploration, generalized compassion, and commitment to social justice over specific religious doctrines or prescriptive personal morality (Wellman 2008). These ideologies may correlate with but are not theologically bound to any given tradition or denomination, and substantial ideological variation can be found within and between Protestant denominations as well as among Catholics (e.g., Smith 1998; Starks 2009; Starks and Robinson 2009; Wuthnow 1996).

While religious and political ideologies are distinct, they are inevitably interrelated, given that religious ideologies develop within the boundaries of political systems toward which they must adopt some sort of stance, whether it be partnership, separation, or even opposition (Wellman 2008). The link between religious and political conservatism which has risen in the United States since the 1960s is well established, and to the extent that the Republican Party has adapted its platform and rhetoric to reflect religious concerns, there is some direct overlap between the religious and political ideological perspectives (Putnam and Campbell 2010). The relationship between religious and political liberalism is comparatively weaker, but the overall value orientations (e.g., focus on inclusivity and social justice) as well as some specific issue opinions (e.g., on LGBTQ rights) of religious liberals largely resemble themes found in political liberalism (Wellman 2008). Processes of social constraint and belief consolidation may further contribute to overlap of political and religious perspectives over time (Baldassarri and Gelman 2008; Boutyline and Vaisey 2017; DellaPosta 2020).

Yet this should not be taken to mean religious and political ideologies are indistinguishable, either conceptually or empirically. Robinson and colleagues (Davis and Robinson 1996a, 1996b, 1999; Starks and Robinson 2007, 2009) have shown extensively that religious ideology predicts social and political attitudes in ways that do not align with either political ideology or denomination. Specifically, the orthodox (i.e., religious conservatives) are politically conservative on cultural and social issues such as abortion or LGBTQ rights, but hold more liberal views in areas such as economics and the role of government, reflecting a communitarian orientation which does not map neatly onto the United States political spectrum; the more individualistic religious modernists (i.e., religious liberals) show the opposite pattern. Historical, theological, and empirical analyses have demonstrated ambivalence toward mainstream politics among religiously observant Protestants and Catholics (Glazier 2015; Starks 2009; Steensland and Wright 2014; Wellman 2008; Woodberry and Smith 1998). In short, overlap between religious and political ideologies is real, but partial. Even to the extent that there is overlap of opinion profiles, religious ideology influences issue salience, as religious ideologues will be more knowledgeable about and motivated by social and cultural than economic or foreign policy issues (Davis and Robinson 1996a; Guth et al. 2006). Religious ideology is most influential in domains which can be most clearly linked to religious concerns, and in ways which are not merely endogenous to political context.

Despite evidence for the distinctiveness and social relevance of religious ideology, the implications of this concept remain only partially explored. Ample scholarship has examined the link between religious ideology and political behavior and attitudes, but most quantitative research has ignored its influence in other domains of life—namely, that of the family (but see Wilcox 2004; for qualitative treatment, see Konieczny 2013; Wellman 2008). If religious ideological distinctions are as broad and as deep-rooted as prominent formulations suggest, we should be able to see them in family dynamics above and beyond the effects of denomination or religious tradition. The purpose of this study is to test this proposition, specifically by examining differential religious transmission within families based on religious ideology.

RELIGIOUS IDEOLOGY AND FAITH TRANSMISSION

Given a large body of literature demonstrating the importance of ideological differences in the modern religious landscape, there is strong reason to believe religious ideology will play a role in the strength of religious transmission. Specifically, I expect transmission to be stronger within families where parents are more religiously conservative. There are several plausible mechanisms that might produce this outcome.

Parenting Approaches

Religiosity is associated with closer family relationships (Pearce and Axinn 1998; Regnerus and Burdette 2006), which in turn predict stronger religious transmission (Bengtson 2017). Mahoney and colleagues (2003) have argued that religion is linked to a “sanctification” of family life which may lead parents to more actively invest in family relationships, and this process of sanctification is likely related to conservative religious ideology. Conservative Protestants, for instance, have been shown to exhibit high parental warmth, low verbal aggression, and authoritative parenting styles in raising their children, and this pattern is better explained by theological beliefs than denominational affiliation (Wilcox 1998; Bartkowski and Wilcox 2000). Wellman (2008) observes an intense focus among evangelicals on nurturing “strong, functional family life,” and Konieczny (2013) notes a similar prioritization of family life and protectiveness toward children among traditional compared to liberal Catholics. Differences in family relationships by parental religious ideology may contribute to differential religious transmission.

Family Religious Socialization

Relatedly, religious conservatives and liberals are likely to differ in their goals and strategies for religious socialization of their children. Among conservative parents, religious engagement and belief are normative moral commitments which pertain directly to ultimate concerns of eternal salvation or damnation (Bengtson 2017; Wellman 2008); thus, religious socialization will be a high priority, manifested in family religious practice as well as explicit messaging on the part of parents. For instance, religious conservative parents may be more likely to say grace over meals, engage in family prayer or devotions, or discuss religious considerations when making decisions (Wuthnow 1996). In contrast, religious liberal parents will view religious faith and practice more in terms of a personal journey than a normative commitment of ultimate moral consequence. Efforts at transmission of religious belief and practice will thus be treated as a lesser priority, or even a violation of children’s autonomy (Wellman 2008; see also Starks and Robinson 2007). Differences in religious socialization should therefore produce different religious outcomes in children based on parental religious ideology, even between parents with similar levels of religious commitment.

Congregational Embeddedness

In addition to direct religious socialization, religious conservative and liberal parents may differ in the cultural environments to which they expose their children. Parents with more conservative religious beliefs may invest more time and money into congregational life due to a perception of greater religious rewards (Iannaccone 1994), and experience a sense of tension with the larger society which leads to greater embeddedness in religious social networks (Smith 1998; Stark and Finke 2000). Among religious liberal parents, valuation of autonomy and inclusivity will work to undermine a sense of personal obligation toward congregational investment, as well as any preference for religious over nonreligious social networks (Wellman 2008). These factors may contribute to stronger religious transmission in children of conservatives compared to liberals.

THE PRESENT STUDY

In this study, I test both additive and interactive hypotheses regarding the effect of parental religious ideology on intergenerational religious transmission against a null hypothesis that transmission does not differ between groups (see figure 1). The additive hypothesis is simply that parental religious ideology is an independent predictor of children’s religious outcomes. So, holding other key factors constant, the children of more religiously conservative parents will be more religious than will the children of more liberal parents (i.e., there will be a significant main effect). The interactive hypothesis is that the relationship between parent and child religiosity will be stronger when parents are more religiously conservative, indicative of a difference in the strength of religious transmission (i.e., there will be a significant interaction effect). In other words, religiously conservative parents will be more influential on the later religious practices of their children than will religiously liberal parents. I further test the extent to which family relationships, religious socialization, and congregational embeddedness might serve as mechanisms through which parental religious ideology influences children’s religious outcomes.

FIGURE 1.

Intergenerational Religious Transmission by Parent Religious Ideology.

DATA

The data used for this study come from Waves 1 and 4 the NSYR. This is a nationally representative panel study following respondents from adolescence to adulthood. At Wave 1 (collected 2002–2003), 3,370 youth (ages 13–17) and one resident parent figure, mother if available and father or other household adult if not, were surveyed using random-digit-dial telephone interviews. Wave 4 data were collected in 2013, when respondents were between 23 and 28 years old, using internet survey for some respondents and telephone interview for others. Due to attrition, the sample size for the final wave was 2,144. This study employs three subsamples. After excluding cases without a valid longitudinal weight (n = 73) or which were missing data on any key variables or controls (n = 51), descriptive findings include 2,020 observations. The first set of regression analyses further excludes cases where parents identified as nonreligious at Wave 1 (n = 121), as this group has a total lack of variation on key independent variables. Specifically, nonreligious parents are defined by having no worship attendance or religious affiliation, both of which are included in multiple regression models. This leaves a sample of 1,899 cases. Finally, analyses examining mechanisms further exclude cases with missing data on any intervening variables (n = 20), leaving a sample of 1,879 observations for the final set of models.

Dependent Variables

I examine two outcomes for children at Wave 4 of data collection. Frequency of religious attendance is measured in seven categories, ranging from (0) Never to (6) More than once a week. Respondents were further asked, “How important or unimportant is religious faith in shaping how you live your daily life?” with available responses ranging from (1) Not at all important to (5) Extremely important. These items capture both behavioral and subjective aspects of religious commitment.

Independent Variables

The key predictors for religious transmission are parental worship attendance and importance of faith at Wave 1. The parental worship attendance item is identical to that described above for children at Wave 4. The Wave 1 item measuring parental importance of faith is worded similarly to that used for their adult children, though response categories are slightly different and range from 1 to 6.

Parent religious ideology is measured categorically based on the following question: “When it comes to your religious beliefs, compared to other religious Americans, do you usually think of yourself as…” with response options including (1) Very liberal, (2) Liberal, (3) Moderate, (4) Conservative, (5) Very conservative, and (6) Haven’t thought much about this. I coded these responses into four categories: Liberal, Moderate, Conservative, and No religious ideology (with the latter also including a handful of parents who responded “Don’t know” or “Refused”). As this item presupposes at least a minimal level of religiosity, parents who reported no worship attendance and no religious affiliation were not asked this question. For the sake of comparison in descriptive statistics, however, I included these parents as an additional group, Not religious, for a total of five categories (although the Not religious group is excluded from regression analyses, as described above).

I control for a variety of family religious and sociodemographic characteristics. A categorical variable of parental religious tradition, using Lehman and Sherkat’s (2018) religious typology, is included in the analysis to control for the effects of denomination on Wave 4 outcomes. Family structure is measured with a dichotomous variable indicating whether the respondent’s biological parents were married at Wave 1. To measure family religious agreement, I include a dichotomous variable indicating whether or not the respondent’s resident parents practice the same religion. Background SES is measured as highest level of parental education at Wave 1 (in four categories from (1) High school or less to (4) Graduate degree). Income is measured based on 11 responses reflecting $10,000 increments, and those who responded “Refused” or “I don’t know” (n = 106) coded to the mean response (“$50–$60k”). Sex, age at Wave 4, race, and Southern residence at Wave 1 are also included in the models.

To account for the possible confounding effects of political ideology, I include an item in which parents were asked how they think of themselves politically. Possible responses to this item are nearly identical to those from the question about religious ideology, and similarly coded into four groups of Conservative, Moderate, Liberal, and No ideology.

Intervening Variables

Finally, I examine a variety of possible intervening variables which may explain differences in children’s religious outcomes by parent religious ideology. I focus specifically on family relationships, religious socialization, and congregational embeddedness at Wave 1.

The NSYR contains a variety of questions about parent–child relationships. I include three mean scales based on these measures. The first is drawn from five items regarding the frequency of positive interaction between children and the parent who completed the Wave 1 survey. Youth were asked how often their parent hugs them, offers encouragement, and tells them they love them, as well as how often they talk about personal things or “just have fun hanging out and doing things together” (α = .77). The second scale is drawn from four items in which parents are asked about their monitoring of children’s activities (location, peer group, and TV and internet usage) (α = .61). The third is focused on parental control of children, and developed from four items concerning children’s perceptions of how likely they are to be disciplined if they have “done something wrong” and how upset they expect parents would be if they learned their children were skipping school, using illegal drugs, or having sex (α = .57).

I include two items which reflect family processes of religious socialization. The first is a dichotomous item indicating whether or not youth report having prayed with their parents in the past year outside of mealtimes or religious services. In the second, youth were asked how often their family discusses “God, the Scriptures, prayer, or other religious or spiritual things together,” with responses ranging from (1) Never to (6) Every day.

Finally, two measures are included to capture family congregational embeddedness. First, parents were asked how often they participate in religious activities in their congregations outside of worship services, “such as potlucks, scripture studies, choir practice, small groups, etc.” At minimum, higher parent religious involvement will indicate modeling religious commitment for their children, and it will likely reflect greater family congregational ties as well. Secondly, parents were asked how often they have encouraged their teenage child to participate in religious youth groups, with responses from (1) None to (4) A lot.

Survey item wording, response categories, and coding used to create all variables are presented in Supplementary Appendix B.

ANALYTIC STRATEGY

I first present descriptive statistics for each religious ideological group (as well as the nonreligious), and the full sample, to examine differences in composition between groups. Second, I present multiple regression analyses for the Wave 4 child religious outcomes of frequency of worship attendance and importance of faith. As each of these outcomes are ordinal, I apply ordered logistic regression techniques to examine differences in children’s outcomes based on parent religious ideology. The first set of analyses (table 2) tests the additive and interactive hypotheses for both outcomes. In a second set of analyses (table 3), I examine the extent to which differences in outcomes by parent religious ideology are explained by intervening variables. I further report the results of decomposition analyses identifying the explanatory power of different mediating variables. A longitudinal weight is applied to all regression analyses to address sample attrition across waves.

TABLE 2.

Ordinal Logistic Regression Models for Children’s Wave 4 Religious Attendance and Faith

Predictors	Worship Attendance			Importance of Faith
	M1	M2	M3	M4	M5	M6
Parent religious ideologya
Liberal	−0.84***	−0.42*	0.26	−0.83***	−0.62**	1.33
Moderate	−0.62***	−0.28†	0.55	−0.70***	−0.49**	1.79*
No ideology	−0.48**	0.31†	0.50	−0.41**	−0.46**	1.64*
Parent religious attendance	0.29***	0.26***	0.40***
Parent importance of faith				0.50***	0.38***	0.70***
Parent religious traditionb
Liberal Protestant		−0.03	−0.04		−0.01	−0.04
Sectarian/ Baptist		0.37*	0.35*		0.39**	0.39**
Catholic		−0.26†	−0.26†		−0.23	−0.23
Latter-Day Saints		0.75**	0.72**		0.74*	0.72*
Other		−0.25	−0.26		−0.27	−0.26
Parents married		0.05	0.05		0.07	0.07
Parents are same religion		0.14	0.15		0.47***	0.47***
Parents’ highest educationc
Some college		0.03	0.02		−0.27†	−0.28†
Bachelor’s degree		−0.12	−0.15		−0.20	−0.22
Advanced degree		−0.25	−0.28		−0.20	−0.23
Family income		0.05*	0.05*		−0.03	−0.03
Female		0.33**	0.33**		0.56***	0.55***
Age (W4)		0.02	0.03		0.04	0.04
Raced
Black		0.46**	0.53**		1.00***	1.04***
Hispanic		0.47*	0.49*		0.51**	0.55**
Other race		0.03	0.04		0.09	0.10
South		0.15	0.17		0.21†	0.21†
Parent political ideologya
Liberal		−0.62**	−0.61**		−0.40*	−0.41*
Moderate		−0.51**	−0.50**		−0.35*	−0.36*
No ideology		−0.18	−0.18		−0.15	−0.14
Interactionsa
Parent attendance × Liberal			−0.15†
Parent attendance × Moderate			−0.19*
Parent attendance × No ideology			−0.20*
Parent faith × Liberal						−0.35*
Parent faith × Moderate						−0.42*
Parent faith × No ideology						−0.38**

Notes: N = 1,899. Longitudinal weight applied to all analyses.

^a Reference: conservative.

^bReference: Moderate/other protestant.

^cReference: high school or less.

^dReference: White.

^† p < .10; *p < .05; **p < .01; ***p < .001.

TABLE 3.

Ordinal Logistic Regression Models of Intervening Variables Predicting Children’s Wave 4 Religious Outcomes

Predictors	Worship Attendance					Importance of Faith
	M1	M2	M3	M4	M5	M6	M7	M8	M9	M10
Parent religious ideologya
Liberal	−0.41*	−0.30†	−0.44*	−0.31†	0.18	−0.57**	−0.43*	−0.56**	−0.42*	0.62
Moderate	−0.22	−0.19	−0.24	−0.14	0.47	−0.39*	−0.33*	−0.38*	−0.26	1.34
No ideology	−0.32†	−0.26	−0.33*	−0.26	0.48	−0.42**	−0.31†	−0.39*	−0.29†	1.22†
Parent religious attendance	0.25***	0.19***	0.18***	0.12**	0.24***
Parent importance of faith						0.35***	0.24***	0.28***	0.19***	0.40**
Interactionsa
Parent attendance × Liberal					−0.11
Parent attendance × Moderate					−0.14
Parent attendance × No ideology					−0.18*
Parent faith × Liberal										−0.18
Parent faith × Moderate										−0.29†
Parent faith × No ideology										−0.28*
Parenting approaches
Positive parental interaction	0.06			−0.08	−0.08	0.24**			0.08	0.08
Parental monitoring	0.18†			0.12	0.12	0.19*			0.11	0.11
Parental control	0.33**			0.29**	0.30**	0.33**			0.25*	0.25*
Religious socialization
Parent/child pray together		0.31**		0.33**	0.33**		0.34**		0.30*	0.29*
Family religious discussion		0.21***		0.19***	0.19***		0.32***		0.28***	0.27***
Congregational embeddedness
Parent religious activity			0.13**	0.11**	0.10*			0.17***	0.11**	0.10**
Youth group encouragement			0.04	0.00	0.01			0.04	−0.01	−0.01

Notes: N = 1,879. Models include all covariates from table 2. Longitudinal weight applied to all analyses.

^aReference: conservative.

^† p < .10; *p < .05; **p < .01; ***p < .001.

RESULTS

Descriptive Statistics and Group Differences

Table 1 shows means and proportions for each variable across parental ideological groups as well as for the total descriptive sample. (See Supplementary Appendix A for additional descriptive statistics.) These results show low mean religious attendance among Wave 4 respondents. However, there is substantial variation based on parental religious ideology, and largely in the expected ways: children of religious conservatives have the highest religious attendance, followed by religious moderates or those with no ideology, and with the lowest attendance among children of religious liberals and, especially, nonreligious parents. Children’s reported importance of faith is relatively higher than their attendance, but distributed similarly between groups.

TABLE 1.

Mean/Proportion by Parent Religious Ideology

Variable	Parent Religious Ideology					Total	Range	Significance
	Conservative	Moderate	Liberal	No Ideology	Not Religiousa
W4 child outcomes
Religious attendance	2.51	1.47	1.11	1.63	0.78	1.69	0–6	***
Importance of faith	3.57	2.95	2.65	3.08	2.12	3.05	1–5	***
W1 predictors
Parent religious attendance	4.60	3.62	2.73	3.06	0.00	3.34	0–6	***
Parent importance of faith	5.64	5.13	4.57	4.84	3.09	4.96	1–6	***
Parent religious tradition
Liberal Protestant	0.04	0.05	0.11	0.03	0.00	0.05		***
Mod/other Protestant	0.26	0.28	0.29	0.24	0.00	0.25
Sectarian/Baptist	0.40	0.31	0.21	0.38	0.00	0.32
Catholic	0.20	0.30	0.24	0.29	0.00	0.24
Latter-Day Saints	0.08	0.01	0.01	0.02	0.00	0.03
Other	0.03	0.05	0.13	0.04	0.00	0.05
None	0.00	0.01	0.00	0.00	1.00	0.06
Parents’ married	0.72	0.61	0.55	0.54	0.54	0.60	0–1	***
Parents are same religion	0.74	0.61	0.45	0.55		0.56	0–1	***
Parents’ highest education								***
High school or less	0.12	0.19	0.16	0.31	0.35	0.21
Some college	0.31	0.30	0.33	0.41	0.30	0.34
Bachelor’s degree	0.33	0.24	0.25	0.19	0.15	0.25
Advanced degree	0.24	0.26	0.26	0.08	0.21	0.20
Family income	6.79	6.86	6.39	5.50	6.43	6.32	1–11	***
Female	0.54	0.51	0.57	0.53	0.41	0.53	0–1
Age (W4)	25.41	25.38	25.56	25.50	25.75	25.48	23–28
Race								***
White	0.79	0.74	0.72	0.65	0.79	0.72
Black	0.09	0.11	0.12	0.18	0.07	0.12
Hispanic	0.08	0.10	0.08	0.10	0.07	0.09
Other race	0.04	0.05	0.08	0.06	0.07	0.06
South	0.41	0.38	0.38	0.40	0.24	0.39	0–1
Parent political ideology								***
Conservative	0.77	0.30	0.11	0.17	0.23	0.35
Moderate	0.09	0.42	0.24	0.24	0.20	0.23
Liberal	0.04	0.16	0.51	0.09	0.31	0.18
No ideology	0.10	0.12	0.14	0.49	0.26	0.24
N (proportion)	560 (0.28)	352 (0.17)	395 (0.20)	592 (0.29)	121 (0.06)	2,020
Intervening variables
Parenting approaches
Positive parental interaction	4.08	4.02	3.99	3.93		4.01	1–5	*
Parental monitoring	4.29	4.13	4.06	4.10		4.15	1–5	***
Parental control	4.50	4.38	4.25	4.34		4.37	1–5	***
Religious socialization
Parent/child pray together	0.58	0.40	0.33	0.36		0.43	0–1	***
Family religious discussion	4.11	3.21	2.73	3.11		3.35	1–6	***
Congregational embeddedness
Parent religious activity	3.72	2.72	2.52	2.57		2.93	1–7	***
Youth group encouragement	3.39	3.05	2.65	2.81		2.99	1–4	***
N (proportion)	557 (0.30)	349 (0.19)	392 (0.21)	581 (0.31)		1,879

Notes: All means/proportions are unweighted. Significance calculated using one-way ANOVA analyses for continuous or dichotomous items, and Pearson’s χ ² tests for categorical items.

^a“Not religious” group excluded from significance tests.

*p < .05; ***p < .001.

Parents across groups report higher religiosity at Wave 1 than their children at Wave 4. Between groups, however, the differences are similar to those seen for children. Specifically, religious liberals have the lowest levels of worship attendance (other than nonreligious parents, who by definition never attend) and conservatives the highest. The same general pattern is observed for parental importance of faith. Many differences in denominational composition by religious ideological group reflect what would be expected based on past literature. A plurality of religious conservative parents is Sectarian/Baptist (though Sectarians/Baptists are as likely to report no religious ideology as they are to be conservative). Liberal Protestants are, appropriately, concentrated in the religiously liberal group. Catholics are spread out across religious ideological categories, reflecting significant diversity within this tradition.

As expected, the overlap between political and religious ideology is substantial. This overlap is strongest for conservatives—77% of religious conservatives also identify as political conservatives. This is unsurprising given the modern conservative political movement’s explicit incorporation of causes relevant to religious conservatives, such as abortion and church-state relations (Putnam and Campbell 2010; Wuthnow 1988). This overlap is somewhat weaker for the other groups but still substantial (42%, 51%, and 49% for religious moderates, liberals, and those who report no ideology, respectively). (The theoretical and empirical implications of the high correlation between religious and political ideology are explored more fully in Supplementary Appendix C. The question of whether parental religious ideology, political ideology, or religious tradition best explain children’s Wave 4 outcomes is examined in Supplementary Appendix E.)

It is worthwhile here to examine the No ideology group more closely. Parents in this group report lower levels of worship attendance and importance of faith than either conservatives or moderates, but somewhat higher than liberals. Perhaps the most notable differences between the No ideology group and the others are sociodemographic: they have the lowest average levels of education and income and are the most likely to belong to racial minority groups. While it is speculative, it may be that a lack of identification of a religious ideology may reflect barriers to the awareness, engagement, or inclusion which contribute to development of an ideological identity. This is consistent with research which finds greater ideological coherence (and polarization) among elite compared to marginalized populations (Baldassarri and Gelman 2008).

Group differences in intervening variables look essentially similar to those seen for parent and child religiosity variables. Across items, religious conservative parents report the highest levels of involved and authoritative parenting, religious socialization, and congregational embeddedness, while religious liberal parents report the lowest levels, and those with no ideology most closely resemble religious liberals.

Analysis of variance (ANOVA) and chi-squared tests show that religious ideological group differences are highly significant for nearly every item. The primary exceptions are gender, age, and Southern residence at Wave 1.

Parent Religious Ideology and Transmission of Worship Attendance and Importance of Faith

Ordered logistic regression analyses in table 2 demonstrate differences in children’s Wave 4 religious outcomes of worship attendance and importance of faith by parent religious ideology, as well as the relationship between parent and child religious characteristics. In each model, conservatives serve as the reference group. Model 1 shows that children of religious conservatives have significantly higher levels of religious attendance than those of any other group, and that this effect is independent of parents’ worship attendance. The magnitude of this difference is strongest between children of conservatives and liberals, followed by moderates and then those with no ideology. In Model 2, after controlling for parent religious denomination and political ideology as well as family and sociodemographic covariates, the differences between children of religious conservatives and either moderates or those with no ideology are reduced to marginal significance, but that between children of liberals and conservatives remains significant at the p <.05 level. These results lend support to the additive hypothesis.

Interactions between parental religious attendance and ideology, indicative of differences in the strength of religious transmission, are shown in Model 3. These results suggest that religious transmission is stronger for religious conservatives than the other groups; however, the pattern of differences runs somewhat contrary to expectations. While transmission of religious attendance is significantly stronger for conservatives than moderates or those with no ideology at the p <.05 level, this difference is only marginally significant for children of conservatives compared to liberals. In other words, while children of religious conservatives attend worship more often than those of liberals in absolute terms, the relationship between parent and child worship attendance differs only weakly between these two groups.

Results for children’s importance of faith are broadly similar to those of attendance, though here differences based on parent religious ideology are somewhat more consistent. Model 4 shows that when controlling for parents’ importance of faith, children of religious conservatives report significantly higher levels of faith than for any other group, and the largest difference is found between liberals and conservatives. When including controls in Model 5, these differences are reduced somewhat but remain highly significant. Model 6 shows significant interactions between parent religious ideology and importance of faith—transmission of faith is stronger for religious conservatives than for any other group. Thus, for children’s importance of faith, both the additive and interactive hypotheses are supported.

As logistic regression coefficients can be non-intuitive to interpret and, in the case of interactions, potentially misleading (Mize 2019), the relationships found in Models 3 and 6 are plotted in figures 2 and 3. These figures show the cumulative predicted probabilities of children’s outcomes based on parent reports for frequency of worship attendance and importance of faith, respectively. Probabilities are calculated using average marginal effects. In each figure, the left panel shows predicted probabilities for children of religious conservatives, while the right panel shows the average of the predicted probabilities for all other groups. I take this approach, first, for ease of comparison, and second, because the differences between the other groups are barely discernible when plotted and do not reach statistical significance in any model. (Separate plots for each parent ideological group, and with all seven categories of attendance shown, can be viewed in Supplementary Appendix D.) The key takeaway for figures 2 and 3 is the same—the steeper slopes in the left panels indicates a stronger link between parent and child religious characteristics for religious conservatives than the other groups. Additionally, at the higher levels of parent religiosity, children of religious conservatives are more likely to report higher levels of religiosity themselves than are children from the other groups.

FIGURE 2.

Cumulative Predicted Probabilities for Children’s Wave 4 Religious Attendance.

FIGURE 3.

Cumulative Predicted Probabilities for Children’s Wave 4 Importance of Faith.

Potential Mechanisms between Parent Religious Ideology and Children’s Religious Outcomes

In a second set of analyses, I examine possible mechanisms explaining differences in children’s religious outcomes by parent religious ideology, focusing first on worship attendance and then importance of faith (table 3). Model 1 shows that when including family relationship items, parental control is a significant predictor of children’s worship attendance; however, it does not explain the difference between children of religious conservatives and liberals. In Model 2, both parent–child joint prayer and family religious discussion have highly significant effects on children’s attendance, while differences by parent religious ideology are no longer fully statistically significant. Model 3 shows that congregational embeddedness measures do not explain the difference between children of religious conservatives and liberals, and for those with no ideology this difference actually becomes slightly stronger. Thus, I find strong support for parent religious socialization, but not family relationships or congregational embeddedness, as a primary mediating pathway explaining differences in children’s worship attendance by parent religious ideology. Finally, in Model 5, we see that with all mediators included, transmission of religious attendance is significantly stronger for religious conservatives than those with no ideology, but not liberals or moderates. These results suggest that this set of explanatory mechanisms explains additive effects almost entirely, and interactive effects partially.

Differences in children’s importance of faith by parent religious ideology are less sensitive to the inclusion of intervening variables than for religious attendance, but the pattern is broadly similar. While most of the individual intervening variables in Models 6–8 independently predict higher importance of faith, religious socialization items lead to the greatest reduction in the differences in outcome by parent religious ideology, though these differences remain statistically significant for children of conservatives compared to liberals and moderates. In Model 9, when all mediators are included, the difference in importance of faith between children of religious conservatives and moderates is no longer significant and for those with no ideology becomes marginal, while the coefficient for liberals remains statistically significant. Model 10 suggests that the full set of mediators partially explains differences in the relationship between parents’ and children’s importance of faith (i.e., interactive effects), as only the difference between children of conservatives and those with no ideology remains statistically significant.

To more closely examine the explanatory effects of intervening variables, I apply multiple mediator, or KHB, analyses (Breen et al. 2013) to Models 4 and 9 of table 3 (results not shown). This technique can be applied to nonlinear regression models to decompose the total effect of a predictor variable on an outcome to separate the direct effect from the indirect effects of one or more mediating variables. This approach further allows for assessment of the proportion of indirect effects which are explained by each mediating variable. Results show that in every case, family religious discussion is by far the most powerful explanatory variable, accounting for between 44% and 62% of each mediated difference in children’s religious outcomes by parent religious ideology. The remaining effects are best explained by parental control, parent–child joint prayer, and parent religious activity, though none of these accounts for more than 29% of any mediated effect. These findings underscore the importance of micro-level family interaction as a primary mode of religious transmission.

DISCUSSION

Past scholarship has clearly established a strong relationship between parent and child religiosity, and highlighted both the family-level and denominational factors which may influence the strength of this transmission. However, research has not directly examined the role of parental religious ideology in shaping religious transmission, despite a large literature which has emphasized the relevance of ideological division in the American religious landscape (e.g., Hunter 1991; Wuthnow 1988). In this study, I have sought to address this gap by examining differences in transmission based on parental religious ideology. Results show both an additive and interactive relationship between ideology and children’s religious outcomes of worship attendance and importance of faith. The general pattern is that religious transmission occurs more strongly for children of religious conservatives than for any other group. This effect is distinct from differences based on either religious tradition or political ideology, and is largely but not fully explained by differences in religious socialization between groups.

These findings lend support to Wuthnow’s (1988) account of religious restructuring, and suggest that it has application not only for the social and political domains (which have been the focus of most research in this area), but also for understanding family-level religious dynamics. As conceptualized here, religious ideology has direct implications for the centrality of religion in family life, and hence, for the nature and intensity of religious socialization of children. A more conservative religious ideology entails a component of religious normativity and belief in divine reward/punishment which promotes religious transmission, while a liberal ideology involves an emphasis on personal autonomy and religious subjectivity which undermines it, even when parents have similar levels of religious commitment. These differences transcend denominational as well as political boundaries.

While the overall argument that religious ideology shapes religious transmission is supported here, some findings run contrary to expectations. Based on the hypotheses displayed in figure 1, religious transmission should be strongest for religious conservatives, then moderates, then liberals. This prospect is only partially supported. The magnitude of differences in children’s religious outcomes is strongest between religious conservatives and liberals, as expected, but differences in main effects between liberals, moderates, and parents with no ideology are not significant in any model. Thus, the key distinction seems to be between children of religious conservatives and everyone else. This may indicate a threshold effect of ideology, wherein wider cultural norms militate against religious commitment in young adulthood, and only parents with the strongest ideological commitments are sufficiently able or motivated to instill countercultural religious norms in their children which are maintained later in life. This is only speculative but may be a fruitful area for future research.

A still more counterintuitive finding regarding group differences pertains to interactive effects. As expected, the relationship between parent and child religiosity is strongest for religious conservative parents. Unexpectedly, the effect for conservative parents is most similar to that of liberals. It is the moderate and non-ideological groups who exhibit the weakest link between parent and child religiosity. One potential explanation for this finding is that more ideologically extreme parents will exercise greater coherence and intentionality in their religious lives which strengthens their influence on their children. In contrast, because moderate and non-ideological parents may hold belief systems which are less distinct from those found in the wider culture, their influence on their children’s religious lives is weaker relative to non-familial influences. It should be noted, however, that while the pattern of differences in interaction effects between the liberal, moderate, and non-ideological groups is consistent, these differences are not statistically significant, and so any proposed implications are only suggestive.

This study is subject to some limitations. First, it should be noted that I have examined the strength of the transmission of religious commitment, as moderated by parental religious ideology. This is not the same as examining the transmission of religious ideology itself. I have neither shown nor argued that religiously liberal parents are less successful at transmitting their beliefs or values to their children—only that they see weaker transmission of conventional markers of religious commitment. The transmission of religious ideologies is beyond the scope of both this study and of the NSYR, which does not contain direct measures of children’s religious ideologies or parents’ theological beliefs. This relates to a second limitation: I have defined religious ideology in terms of shared sets of religious beliefs, a conception which receives strong support in past theory and literature (e.g., Davis and Robinson 1996a; Hunter 1991; Wuthnow 1996). As I lack measures of the beliefs themselves, however, the survey item on parental religious ideology should be regarded as a minimal indicator of the underlying concept. Future research should examine whether there are particular beliefs, or combinations of beliefs, which are most instrumental in shaping the strength of religious transmission. Third, the NSYR sample is basically homogenous with respect to age/period/cohort. It is unclear whether the differences I find between groups will hold across the life course or decline over time as those with low religiosity at the time of NSYR data collection resume religious practice later in life. However, given an overall decline in normative pressure toward religious commitment (Hout and Fischer 2002) and the relatively minor influence of family formation on returns to religion (Uecker et al. 2016), there is little reason to expect that the overall pattern of religious differences found in this study is specific to young adulthood. Finally, the analyses here are limited to the primarily Christian United States context. As parental influences on children’s religious outcomes have been found to vary widely by national context (Kelley and De Graaf 1997), and the form a religious ideology takes is itself shaped by particular political and historical factors (Wellman 2008), the extent to which these findings might be applied to other countries or religious traditions is an open question.

I have shown that the macro-level account of religious restructuring in the United States, in which ideological distinctions have overtaken denominational ones, can improve our understanding of the micro-level intrafamilial processes by which religious commitment is transmitted from parents to young adult children. Religious transmission is strongest among religious conservatives, and this is largely explained by their more frequent efforts at religious socialization in day-to-day family life when children are growing up. I have left unexplored the intervening and contextual factors which may shape children’s religious trajectories in between their teenage years and young adulthood. Youth are likely to face increased exposure to secularizing influences in emerging adulthood (Smith and Snell 2009), and it is unclear how parental religious ideology may shape how children respond to these influences. Future research should explore the ways in which religious ideological differences, both in individuals and families, unfold over the life course.

FUNDING

This research was supported by funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) to the Population Research Institute at The Pennsylvania State University for Population Research Infrastructure (P2C HD041025) and Family Demography Training (T-32HD007514). The findings and conclusions do not necessarily reflect the views of the funding agency.