Parental Influence and Intergenerational Transmission of Religious Belief, Attitudes, and Practices: Recent Evidence from the United States

Abstract

A traditionally salient topic of empirical investigation in the sociology of religion, this paper seeks to offer a recent investigation into the intergenerational transmission of religion and the parental forms of religious engagement that predict adult engagement with religion. The study of this paper explores the intergenerational transmission of religion, focusing on the parental forms of religious identity and engagement that influence religious identity, beliefs, and practices in adulthood. By analyzing the 2018 GSS dataset in the United States with multiple regression analyses, I found strong parental and childhood influences on adult religiosity, religious service attendance, and belief in God. Indeed, this engagement often mirrors parental engagement for these variables. However, while paternal religious identity often predicts these religious variables, I found that the religious identity and engagement of parents generally do not predict religious identity in adulthood. Ultimately, while these results generally show strong predictive mechanisms of intergenerational transmission, they also illustrate that these relationships are variably dependent on the form of parental and adult religious engagement, and which parent participates or is associated with that engagement.

1. Introduction

In sociological investigations of religion, one prominent subject of extensive debate and research is the factors that influence religious identity, religiosity, and religious engagement (e.g., religious service attendance). Within this previous research, one central question is whether such influence comes, and to what degree, from someone’s background, especially their upbringing and parental influence. A related question is whether the influences exerted in one’s social background via these familial mechanisms continue their influence into one’s current situation (e.g., Ploch and Hastings 1998). Likewise, if so, which factors exactly have this kind of staying power of influence? Understanding the magnitude and dynamics of intergenerational religious influence and transmission has been a salient concern of social scientists in recent decades (e.g., Hoge and Petrillo 1978; Willits and Crider 1989; Ploch and Hastings 1998; Myers 1996; Voas and Crockett 2005). Therefore, because religious transmission is a dynamic, and necessarily a generationally sensitive, phenomenon, there is a need for constant empirical observation and update. I thus seek with this paper to provide a recent empirical update to this corpus of work to understand which results can still be confirmed and whether there are any new observations in understanding the intergenerational transmission of religion.

In answer to these foundational research objectives, some have found the potent influence of upbringing. For instance, Voas and Crockett (2005) argue that how someone is raised is the single best predictor of their current religious involvement, even as this relationship is not one-to-one, with parental transmission of religion less potent than parental transmission of religious absence. Additionally, even if people may be slightly less religious than their parents, and few are more religious, the phenomenon of parental influence upon religious belief and engagement is relatively stable across global societies (Storm and Voas 2012).

I set out in this paper to critically compare and contrast variables associated with the religious dimensions of social origin and assess whether those influences continue into people’s current religious life. I do so by assessing which of these influences are more predictive of personal religious beliefs and engagement in adulthood. Toward this research aim, I ask the following guiding question: What is the nature and strength of the relationship between parental religious identity and practice and adult religious practice in the United States today, if any? The United States provides a particularly fertile research environment to provide updated answers to this question because of the high-quality social survey data on religion, its high prevalence of religion, and the comparatively high social influence that religion exerts in this national context (Bengston et al. 2009).

2. Parents and Intergenerational Religious Transmission

The influence of religious social origins in shaping religious behavior in both childhood and adulthood is a consistent theme in the broader scholarly discourse on religious continuity and evolution. When considering social origin in this paper, I focus on the influence of parents primarily, and childhood engagement secondarily, with both being nearly inextricably linked to each other. Despite profound societal changes across the world over the last half-century, including both secularization and diversification of religious and spiritual beliefs in much of the world, the family remains a primary source and incubator of dispositions toward religious belief that influence both of these broader trends. Indeed, the family is a primary site of religious reproduction, perhaps especially in irreligious households (e.g., Pusztai and Demeter-Karászi 2019).

It is in this context that Hervieu-Léger (2001) argues for a ‘chain of memory’ as the mechanism of intergenerational religious transmission, whereby transmission is sustained through collective rituals and teachings within a community, be they religious or familial, in place of a more universal collectivity. These collective rituals may include prominent variables explored in this study, such as religious service attendance and the religious belief of parents, which in turn may contribute to the type of familial teachings that can form a chain of memory. However, the universality of Hervieu-Léger’s work across the broad spectrum of religious belief has also been questioned (e.g., Greaves 2009). Greaves (2009) argues that orientations to institutions, traditions, and thus the role of iconoclasm differs widely among ‘religions’, and thus the experience and trends of religion generally, and Christianity particularly, in traditionally Christian contexts are not able to be universally applied. Indeed, even among countries of different Christian denominational entrenchment, differences have been found regarding changes in intergenerational religious orientation (e.g., Molteni and Biolcati 2018). This may also be the case within many of the same societies that today experience a substantial measure of religious pluralism, such as the United States.

Much prior research has shown that religious adherence, attitudes, and behaviors of parents contribute to the intergenerational transmission of religious belief, attitudes toward religion and religious concepts, and behaviors such as prayer or church attendance (Bao et al. 1999; Bader and Desmond 2006; Baker-Sperry 2001; Bengston et al. 2009; Francis 1993; Güngör et al. 2011; Hardie et al. 2016; Jennings et al. 2009; Myers 1996; Patacchini and Zenou 2016; Ploch and Hastings 1998; Stolzenberg et al. 1995; Storm and Voas 2012; Thornton et al. 1992; Voas and Crockett 2005; Voas and Storm 2012). This is true even as a minority of studies investigating such dynamics find that these influences do not hold into adulthood (e.g., Willits and Crider 1989), that the strength of this transmission mechanism is relatively weak (Hoge and Petrillo 1978; Hoge et al. 1982), or that the effects are highly variable (Wilson and Sandomirsky 1991), which is perhaps a product of other, potentially less overtly religious influences of past or present environment, such as social and cultural networks or participation (e.g., Gemar 2020, 2022, 2023; McAndrew and Richards 2020; Patacchini and Zenou 2016; Ploch and Hastings 1998). The highly variable, and sometimes incomplete, lessened, or absent transmission of religious belief and engagement is ultimately a source of decline in affiliations with many religions of the world and attendant secularization (Chaves 1989; Voas and Crockett 2005; Voas and Storm 2012).

For those studies that have found parental (and in some cases also grandparental) influence as a significant mechanism in the intergenerational transmission of religious belief, attitudes, and engagement, there is variability in the precise elements of religion and modes of transmission found. One prominent topic of study in this regard is the intergenerational transmission of religious service attendance, often specifically (Christian) churchgoing (e.g., Bao et al. 1999; Bader and Desmond 2006; Hardie et al. 2016; Ploch and Hastings 1998; Voas and Storm 2012). For instance, Ploch and Hastings (1998) find small but statistically significant relationships between parental religious service attendance and the attendance of their adult children, finding that while also small (and statistically significant), it is rather marital status and having children currently that is more predictive of attendance. From a slightly different perspective, this type of finding is echoed more recently by Schleifer and Chaves (2017), who find that adding a school-aged child to the household directly leads to an increase in the family’s religious service attendance, something which, in turn, may have a circular effect upon transmission to the child. However, Stolzenberg et al. (1995) argue that the effect of children on church membership is highly variable.

Among other more recent studies, Bader and Desmond (2006) found that high frequency and consistency of parental attendance at religious services were an important predictor and influence of intergenerational religious transmission. They also found, however, that when parental attendance is inconsistent, then it is rather the importance that parents place upon religion that is more important than attendance as an intergenerational transmission mechanism, even as any incongruence between these two tends to diminish the effect of each (Bader and Desmond 2006). Both parental religious affiliation and service attendance have also been found to be a mitigating influence on the adolescent decline in religious service attendance as they age into important life course events (Hardie et al. 2016). Voas and Storm (2012) find that this parental influence is most potent and predictive of the intergenerational transmission of churchgoing when both parents attend church when the child is young, rather than only one parent.

With respect to other forms of parental and socialized origin influence upon the intergenerational transmission of religion, there are varied findings of both mechanisms (e.g., religiosity, identity) and the magnitude of the relationship. One foundational study by Myers (1996) focuses on the important influence of parental religiosity. Myers (1996) finds that the religiosity of adult children is highly predicted by the religiosity of one’s parents.

One way that variability in this relational influence is found is by considering which type of religion is being passed down, and in which context. For instance, Jacob and Kalter (2013) find that religiosity among Muslim immigrant families is highly stable, or even increases among generations, while for Christian immigrant families, inter-generational religiosity tends to decline. This again highlights the differences that might exist across religious groups in the same context, especially if religion becomes tied to other forms of identity. Similarly, Güngör et al. (2011) find that intergenerational transmission into adult religiosity (including identity, beliefs, and practices) among Muslims within historically Christian and secularized contexts is remarkably strong. The reason for this increased religious transmission was argued to be the effect of a continued emphasis on the broader cultural heritage among the families and broader community of these groups (Güngör et al. 2011).

Another way that the variable impacts of inter-generational religious transmissions are seen is via the influence of different parents upon different types of children. For instance, while Francis and Brown (1991) find that the effect of parental religiosity decreases for children as they age into adolescence, Francis (1993) later finds that parental influence actually increases as children grow through adolescence. Francis (1993) also finds that mothers’ religious practice was a more powerful predictor of both son and daughter practice than fathers’ influence. Gutierrez et al. (2014) finds that mothers had the strongest socializing role within the familial structure and the most predictive influence on intergenerational religious transmission to children. However, men may also be disproportionately influenced by their father’s religious behavior in childhood because of sons often emulating their fathers (Gutierrez et al. 2014). Hayes and Pittelkow (1993) argue that gendered differences in religious supervision may affect the nature of religious transmission between parents, although the general parental influence is the primary predictor of transmission. Given this prior literature, I also focus in this research on understanding the potential differential influence of maternal and paternal religious identity and engagement on the religious identity, beliefs, and practices of adult offspring.

The next sections on the data, methods, and findings of the research in this paper give more detailed information as to the variables used for going about finding both the aggregate and differential predictive capacity of parental and childhood variables on adult engagement with religion. From this prior literature and empirical research on the intergenerational transmission of religion, I approach our guiding research question moving forward with the following hypotheses:

H1. 

There will be a noticeable relationship between childhood and parental religious orientation and the current beliefs and practices of the respondents in this research.

H2. 

This relationship will be variable in presence or degree among different religious groups and among different parents.

3. Data and Methods

The research for this paper utilizes data from the General Social Survey (GSS), which is a leading national survey research program of the United States, with the aim of gauging current public views, attitudes, and social behaviors. Managed by the National Opinion Research Center (NORC) at the University of Chicago, the GSS has been regularly conducted since 1972, comprising 34 survey cycles from 1972 to 2022, with 2018, 2021, and 2022 having the most iterations. Because the specific questions often shift between survey cycles to reflect evolving societal concerns and prevalent areas of interest for researchers, I use here the 2018 iteration of GSS (n = 2348). The demographic information of this sample largely tracks the US Census Bureau’s 2018 estimates for the core demographic variables of age, sex, and race in the population that is 18 or older (see

Table A1. Basic demographic information from the 2018 GSS compared to the US Census Bureau’s 2018 estimates for basic demographic information.

	US Census	Relative Frequency
	(2018 Estimates 18+)	(GSS)
Age
18–24	13.5%	11.8%
25–34	18.0%	19.0%
35–49	24.6%	25.8%
50–64	25.2%	24.9%
65+	15.8%	18.3%
Sex
Male	48.4%	45.5%
Female	51.6%	54.5%
Race
Other	24.6%	12.8%
Black	12.2%	14.8%
White	63.2%	72.4%

). However, while the categories for age and sex more closely track the census data, as does the Black proportion of the sample, there is an overrepresentation of White Americans and an underrepresentation of other races.

This is the most recent cycle of the GSS that includes all of the necessary information for the research of this article, specifically information about parental religious engagement. Questions about parental religious service attendance were previously asked in 2008, 1998, and 1991, which were also the same previous GSS years that asked about childhood religious service attendance, mother’s religious identification when the respondent was a child, and father’s religious identification when they were a child.

The statistical methods used for the analysis of this paper are multinomial regression analyses. This type of regression analysis allowed us to efficaciously capture predictive effects of prior and parental religious influence variables on current variables of adult respondent religious identification and engagement. The specific variables used to capture prior and parental religious influence were the religious service attendance of each of the respondent’s mother and father when the respondent was a child, the childhood religious service attendance of the respondent (at the age of 12), the respondent’s mother’s religion when the respondent was a child, and the respondent’s father’s religion when they were a child. These five variables represent the independent variables for the regression analyses of this paper, and the categories that I use for each variable can be viewed in

Table 1. Relative frequencies of the ten religious variables used in the regression analyses of this paper.

	Relative Frequency
	(GSS)
Dependent variables
Religiosity
Very religious	15.5%
Moderately religious	37.7%
Slightly religious	25.2%
Not religious	21.6%
Religious service attendance
Every week or more	22.9%
Almost every week	12.4%
Several times/yr. to 1/month	16.4%
Once a year or less	18.5%
Never	29.5%
Strength of religious affiliation
Strong affiliation	23.4%
Low to medium affiliation	42.2%
No religion	34.4%
Change in belief in God
Believe now and always have	73.6%
Believe now but not before	11.0%
Not now but did	9.7%
Not now and not ever	5.8%
Religious identity
Protestant	47.6%
Catholic	23.5%
Jewish	1.6%
Other	1.5%
Buddhism	0.8%
Hinduism	0.3%
Islam	0.8%
Orthodox Christian (Greek, Russian, etc.)	0.3%
No religion
Independent variables
Childhood rel. service attendance b
Every week or more	54.5%
2–3/mo. to nearly 1/week	15.6%
Several times/yr. to 1/mo.	10.5%
Once a year or less	9.0%
Never	10.4%
Mother’s attendance
Every week or more	49.9%
2–3/mo. to nearly 1/week	12.0%
Several times/yr. to 1/mo.	11.4%
1–2/yr. or less	12.1%
No mother present	0.9%
Never	13.7%
Father’s attendance
Every week or more	35.2%
2–3/mo. to nearly 1/week	10.0%
Several times/yr. to 1/mo.	12.4%
1–2/yr. or less	14.6%
No fatherpresent	5.4%
Never	22.5%
Father’s religion
Protestant	43.0%
Catholic	36.1%
Jewish	1.8%
Orthodox Christian	0.4%
Muslim	1.1%
Other	1.1%
No religion	16.5%
Mother’s religion
Protestant	50.1%
Catholic	36.7%
Jewish	1.7%
Orthodox Christian	0.5%
Muslim	0.9%
Other	1.3%
No religion	8.6%

^b Childhood attendance at age 12.

. The relative frequencies of each of these categories within the GSS sample also appear in Table 1.

I likewise used five different dependent variables in the regression analyses so as to be able to compare if parental and childhood influences are stronger on different aspects of religious engagement and identification. These five variables are religious service attendance, religious preference (i.e., identity), strength of affiliation with that religious identity, self-rated religiosity, and the change in the respondent’s belief in God over the course of their life. The specific categories and relative frequencies for each of these variables can again be viewed in Table 1. In general, the relative frequencies of these variables illustrate the continued importance of religion within contemporary American society. By using this high-quality dataset and these multiple regression analyses, I am able to understand if any of these prior and parental influences have predictive power on any of the current religious identities and activities of their adult children, which parental characteristics are most influential, and which current forms of current religiosity and religious engagement are most impacted.

4. Findings

Our regression analyses consist of seven separate multinomial logistic regression models. The results of these analyses can be seen in

Table 2. Log odds ratios from regression analysis of current religiosity and past influence variables. ^a

		Very Religious	Moderately Religious	Slightly Religious
Childhood rel. service attendance b
Every week or more		1.383 **	1.439 ***	1.129 **
2–3/mo. to nearly 1/week		0.378	1.027 *	1.382 **
Several times/yr. to 1/mo.		0.437	0.996 *	0.698
Once a year or less		−0.920	0.263	0.749
Never		---	---	---
Mother’s attendance
Every week or more		0.665	0.321	−0.170
2–3/mo. to nearly 1/week		1.408 *	0.864	0.574
Several times/yr. to 1/mo.		1.782 **	0.578	0.708
1–2/yr. or less		0.503	0.545	0.761
No mother present		−0.154	16.255	15.045
Never		---	---	---
Father’s attendance
Every week or more		−0.082	−0.162	0.321
2–3/mo. to nearly 1/week		−0.726	0.052	0.089
Several times/yr. to 1/mo.		0.297	0.854	1.242 **
1–2/yr. or less		0.048	0.260	0.132
No father present		−0.816	−0.027	0.009
Never		---	---	---
Father’s religion
Protestant		2.544 ***	0.976 **	0.755 *
Catholic		2.366 ***	1.002 **	0.980 **
Jewish		1.151	−0.502	1.180
Orthodox Christian		−0.369	18.908	1.473
Muslim		0.359	0.261	18.230
Other		1.686	1.563	0.248
No religion		---	---	---
Mother’s religion
Protestant		−0.708	0.487	−0.091
Catholic		−1.159	0.528	0.203
Jewish		−1.117	−0.213	−1.443
Orthodox Christian		17.405	−0.903	−0.934
Muslim		0.219	1.324	−18.805
Other		0.260	2.059	−0.414
No religion		---	---	---
Nagelkerke R2	0.261

^a Reference group = ’not religious’; ^b Childhood attendance at age 12; * p < 0.05; ** p < 0.01; *** p < 0.001.

,

Table 3. Log odds ratios from regression analysis of current religious service attendance and past influence variables. ^a

		Every Week	Almost	Several x/yr.	Once a Year
		or More	Every Week	to 1/month	or Less
Childhood rel. service attendance b
Every week or more		0.952 *	0.919	1.853 ***	0.893
2–3/mo. to nearly 1/week		0.635	1.368 *	1.524 *	1.753 ***
Several times/yr. to 1/mo.		0.781	1.428 *	1.729 **	1.469 **
Once a year or less		−0.855	−1.765	0.921	0.791
Never		---	---	---	---
Mother’s attendance
Every week or more		0.939 *	−0.685	0.226	−0.689
2–3/mo. to nearly 1/week		0.108	−0.412	0.706	−0.524
Several times/yr. to 1/mo.		1.883 ***	0.245	1.571 **	0.434
1–2/yr. or less		0.654	−1.225	0.487	−0.012
No mother present		−17.404	−18.484	1.158	−17.904
Never		---	---	---	---
Father’s attendance
Every week or more		0.017	−1.351 **	0.015	0.617
2–3/mo. to nearly 1/week		−0.433	0.912	0.07	−0.007
Several times/yr. to 1/mo.		−0.814	0.696	−0.172	0.057
1–2/yr. or less		−0.036	0.956	−0.244	0.700 *
No father present		0.208	0.116	0.892	−0.209
Never		---	---	---	---
Father’s religion
Protestant		1.304 ***	0.578	0.697	0.47
Catholic		1.155 **	1.059 *	0.712	0.503
Jewish		0.625	−0.098	−0.121	−0.962
Orthodox Christian		−18.614	−15.349	−15.468	−1.162
Muslim		15.941	−0.41	16.192	−0.231
Other		0.415	0.121	−13.529	−0.345
No religion		---	---	---
Mother’s religion
Protestant		0.502	0.819	−0.405	0.281
Catholic		0.157	0.134	−0.112	0.533
Jewish		−0.744	−0.465	0.256	2.215 *
Orthodox Christian		18.439	0.816	−0.364	0.973
Muslim		−15.6	2.017	−15.381	−0.482
Other		1.208	1.163	−13.822	0.552
No religion		---	---	---
Nagelkerke R2	0.288

^a Reference group = ’never’; ^b Childhood attendance at age 12; * p < 0.05; ** p < 0.01; *** p < 0.001.

,

Table 4. Log odds ratios from regression analysis of change in the belief of God and past influence variables. ^a

		Not Now,	Not Now,	Believe Now,
		Not Ever	But Did	Not Before
Childhood rel. service attendance b
Every week or more		−2.476 ***	−1.458 **	−2.535 ***
2–3/mo. to nearly 1/week		−3.976 ***	−0.841	−1.516 **
Several times/yr. to 1/mo.		−2.301 ***	−0.937	−1.501**
Once a year or less		−1.633 **	−1.646 *	−1.429 **
Never		---	---	---
Mother’s attendance
Every week or more		−0.361	0.536	−0.236
2–3/mo. to nearly 1/week		0.666	0.431	−0.834
Several times/yr. to 1/mo.		−0.079	−0.079	−1.617 *
1–2/yr. or less		−1.299	0.966	−0.328
No mother present		−13.173	−13.373	0.579
Never		---	---	---
Father’s attendance
Every week or more		−2.070 **	0.352	0.137
2–3/mo. to nearly 1/week		−0.951	−0.148	−0.029
Several times/yr. to 1/mo.		−1.355	0.385	0.380
1–2/yr. or less		−0.803	−0.930	−0.185
No father present		−1.015	−0.751	−1.437
Never		---	---	---
Father’s religion		−0.518	−1.335 **	0.595
Protestant		−0.214	−0.521	0.127
Catholic
Jewish		−10.975	−25.648	−39.883
Orthodox Christian		−0.742	15.281	0.324
Muslim		−16.744	−17.068 ***	4.168
Other		−0.726	−19.269	−1.563
No religion		---	---	---
Mother’s religion
Protestant		−0.618	−0.240	−0.753
Catholic		−1.789 *	−0.892	−0.755
Jewish		12.546	25.832	25.963
Orthodox Christian		−13.997	−14.753	−15.523
Muslim		3.479	15.471	−5.162 *
Other		2.521	2.364	1.929
No religion		---	---	---
Nagelkerke R2	0.311

^a Reference group = ’believe now, always have’; ^b Childhood attendance at age 12; * p < 0.05; ** p < 0.01; *** p < 0.001.

,

Table 5. Log odds ratios from regression analysis of strength of religious affiliation and past influence variables. ^a

		No Religion	Somewhat/
			Not Very Strong
Childhood rel. service attendance b
Every week or more		−1.374 ***	0.055
2–3/mo. to nearly 1/week		−0.899	0.631
Several times/yr. to 1/mo.		−0.292	1.113 *
Once a year or less		0.052	1.361 **
Never		---	---
Mother’s attendance
Every week or more		−0.101	−0.206
2–3/mo. to nearly 1/week		−0.428	−0.029
Several times/yr. to 1/mo.		−0.992 *	−0.756
1–2/yr. or less		−0.629	−0.154
No mother present		−14.789	1.944
Never		---	---
Father’s attendance
Every week or more		0.133	−0.240
2–3/mo. to nearly 1/week		0.230	−0.094
Several times/yr. to 1/mo.		0.451	−0.043
1–2/yr. or less		−0.751	−0.377
No father present		−0.345	−0.549
Never		---	---
Father’s religion
Protestant		−1.846 ***	−0.844 **
Catholic		−1.360 ***	−0.318
Jewish		−1.508	0.035
Orthodox Christian		−0.791	15.116
Muslim		−0.153	17.878
Other		−0.756	−2.267 *
No religion		---	---
Mother’s religion
Protestant		−0.719	0.467
Catholic		−0.707	0.984 *
Jewish		−0.659	0.359
Orthodox Christian		−18.314	−16.945
Muslim		−1.945	−17.937
Other		−1.727	1.005
No religion		---	---
Nagelkerke R2	0.257

^a Reference group = very strong religious affiliation; ^b Childhood attendance at age 12; * p < 0.05; ** p < 0.01; *** p < 0.001.

,

Table 6. Log odds ratios from regression analysis of reading religious scriptures and past influence variables. ^a

		Yes
Childhood rel. service attendance b
Every week or more		0.817 **
2–3/mo. to nearly 1/week		0.303
Several times/yr. to 1/mo.		0.584
Once a year or less		−0.255
Never		---
Mother’s attendance
Every week or more		0.323
2–3/mo. to nearly 1/week		0.069
Several times/yr. to 1/mo.		0.194
1–2/yr. or less		0.338
No mother present		0.063
Never		---
Father’s attendance
Every week or more		−0.155
2–3/mo. to nearly 1/week		−0.336
Several times/yr. to 1/mo.		−0.156
1–2/yr. or less		−0.028
No father present		−0.091
Never		---
Father’s religion
Protestant		0.797 **
Catholic		0.479
Jewish		0.257
Orthodox Christian		1.368
Muslim		19.020
Other		1.160
No religion		---
Mother’s religion
Protestant		0.190
Catholic		−0.266
Jewish		−0.415
Orthodox Christian		18.360
Muslim		−16.427 ***
Other		−0.945
No religion		---
Nagelkerke R2	0.132

^a Reference group = No; ^b Childhood attendance at age 12; * p < 0.05; ** p < 0.01; *** p < 0.001.

and

Table 7. Log odds ratios from regression analysis of frequency of prayer and past influence variables. ^a

		Frequently	Regularly	Infrequently
		(Several/wk)	(Approx. 1/wk)	(1/yr.–1/month)
Childhood rel. service attendance b
Every week or more		1.461 ***	2.253 *	1.615 **
2–3/mo. to nearly 1/week		0.454	1.396	1.216 *
Several times/yr. to 1/mo.		1.094 *	2.936 **	1.163
Once a year or less		−0.061	0.827	0.858
Never		---	---	---
Mother’s attendance
Every week or more		−0.240	−0.170	−0.774
2–3/mo. to nearly 1/week		0.313	0.308	−0.081
Several times/yr. to 1/mo.		0.098	−0.103	0.036
1–2/yr. or less		−0.117	0.735	0.163
No mother present		1.388	−0.851	−1.075
Never		---	---	---
Father’s attendance
Every week or more		0.355	0.034	0.315
2–3/mo. to nearly 1/week		0.091	−0.607	0.028
Several times/yr. to 1/mo.		0.019	−0.139	0.416
1–2/yr. or less		0.388	−0.603	0.277
No father present		0.620	0.316	−0.761
Never		---	---	---
Father’s religion
Protestant		1.217 ***	1.367 *	1.194 *
Catholic		0.299	0.348	0.601
Jewish		1.247	−1.363	1.561
Orthodox Christian		−1.899	2.483	−3.113
Muslim		25.724 ***	7.946	0.169
Other		0.133	−3.218	−1.663
No religion		---	---	---
Mother’s religion
Protestant		0.176	−0.408	0.297
Catholic		0.888	1.409	1.061
Jewish		−2.192	2.228	−0.571
Orthodox Christian		2.152	−0.154	−0.215
Muslim		−24.367	−8.300	0.914
Other		1.577	5.659 *	4.446
No religion		---	---	---
Nagelkerke R2	0.213

^a Reference group = Never; ^b Childhood attendance at age 12; * p < 0.05; ** p < 0.01; *** p < 0.001.

. While there are only six tables presenting the results of each regression analysis, there were seven regressions performed. The final regression is not presented as a table because there was only one statistically significant result when religious identity was the dependent variable. This result was that those who had Jewish mothers or fathers were statistically much more likely to report being Jewish than having no religion compared to those whose mothers or fathers had no religion.

Table 2 displays the results for the regression model with our five independent variables and self-rated religiosity as the dependent variable. As the results of this table show, those who attended religious services in their childhood were much more likely to be ‘very religious’ than non-religious at the time of the survey compared to those who reported never attending religious services at the age of twelve. Similarly, higher (although not the highest) religious service attendance of the respondent’s mother predicted the same. However, while it was mother’s religious service attendance, it was father’s religious identity that proved predictive, where those whose fathers were Protestant or Catholic were much more likely to report being very religious later in life than not religious, compared to those with no religion. Mother’s religion showed no statistically significant predictive result for respondent’s religiosity. Additionally, while only the highest level of religious service attendance was predictive of being ‘very religious’, the top three levels of attendance frequency were predictive of moderate religiosity, and the top two levels also of slight religiosity. More moderate mother’s attendance and father’s religious identity (as Protestants or Catholics) were predictive of the respondent being moderately religious, while moderate father’s religious service attendance (along with the same results for father’s identity) were predictive of the respondent being slightly religious.

Table 3 displays the results for religious service attendance and the five independent influence variables. With ‘never’ attending religious services as the reference category, the results show that the highest levels of religious service attendance in childhood are most predictive of moderate attendance later in life, although higher levels in childhood are associated with higher levels of attendance later in life generally. The two variables that are most predictive of the highest levels of attendance among survey respondents were mother’s moderate to high religious service attendance and father’s religious identity as Protestant or Catholic (compared to no religion). Father’s religious service attendance and Catholic identity are also predictive of respondents attending religious services ‘almost every week’. Finally, those with Jewish mothers are most likely to attend services once a year or less.

Table 4 displays the results for the regression analysis of the predictive capacity of our five past influence variables on the change in the respondent’s belief in God over time. Compared to those who have never and continue to not believe in God, those who report childhood religious service attendance are much more likely to report always believing in God and continuing to do so, with high levels of father’s religious service attendance and mother’s Catholic identity also predictive of this relationship to statistically significant levels. Elevated levels of childhood attendance at religious services, along with father’s religious identification with either Protestantism, or especially with Islam, are much more likely to report believing both now and always than to have lost their belief in God (i.e., reporting that they do not now believe in God but used to do so). The strength of childhood religious service attendance again shows up as those who attended services as a child being much more likely to report believing now and always than believing now and not before. Moderate mother’s attendance predicts a similar relationship, while mother’s religion being Islam also predicts believing now and always over believing now and not before, compared to no attendance or religion, respectively.

Table 5 reports the results of the multinomial regression analysis of the influence variables and the strength of religious affiliation that the respondent has with their religious identity. With those respondents who report as a very strong X (religious identity) as the reference category, the results show that those who attended religious services as a child every week or more are significantly much more likely to report very strong religious affiliation contemporarily, as well as those who had mothers with moderate religious service attendance and fathers who were Protestants or Catholics. Compared to those with very strong current religious affiliation, those who had moderate religious service attendance as a child, and those whose fathers were either Protestants or had an ‘other’ religion were more likely to have a somewhat or not very strong religious affiliation compared to no childhood attendance or paternal religious identity, respectively. However, individuals with Catholic mothers were more likely to report somewhat or not very strong current religious affiliation, as opposed to a very strong one, compared to those whose mothers had no religion.

Table 6 and Table 7 show regression analyses with the private religious practices of prayer and scripture reading (outside of religious services) as dependent variables. As the results in these tables show, higher childhood attendance at religious services and father’s (Protestant) religious identification are the primary predictors of religious scripture reading and higher rates of prayer. The final statistically significant result shows that those whose mother had a Muslim religious identity are much less likely to read religious scriptures outside of religious services in adulthood, but much more likely to pray frequently if their father had a Muslim religious identity.

Finally, because of the empirical importance displayed in these results regarding parental Protestant identification, we continue with a further series of regression analyses to disaggregate different types of Protestant respondents. To achieve this, I rely on Steensland et al.’s (2000) schema for recoding GSS denominations into mainline and more conservative, or evangelical, Protestant denominations.

Table A2. Regression analyses comparing Evangelical Protestant upbringing to Mainline Protestant upbringing and the religious variables of this paper.

Log odds ratios from regression analysis of current religiosity and Protestant upbringing.
		Very Religious	Moderately Religious	Slightly Religious
Evangelical Protestant		0.985 ***	0.449 *	0.235
Mainline Protestant R2	--- (0.021)
Reference group = ‘not religious’
Log odds ratios from regression analysis of current religious service attendance and Protestant upbringing.
		Every week	Almost	Several x/yr.	Once a year
		or more	every week	to 1/month	or less
Evangelical Protestant		0.710 **	0.055	−0.277	−0.052
Mainline Protestant R2	--- (0.028)
Reference group = ‘never’
Log odds ratios from regression analysis of change in the belief of God and Protestant upbringing.
		Not now,	Not now,	Believe now,
		not ever	but did	not before
Evangelical Protestant		−1.356	−1.234 **	−0.631
Mainline Protestant R2	--- (0.056)
Reference group = ‘believe now, always have’
Log odds ratios from regression analysis of strength of religious affiliation and past influence variables.
		No religion		Somewhat/
				not very strong
Evangelical Protestant		−0.573 **		−0.607 ***
Mainline Protestant R2	--- (0.022)
Reference group = very strong religious affiliation
Log odds ratios from regression analysis of reading religious scriptures and Protestant upbringing.
		Yes
Evangelical Protestant		0.905 ***
Mainline Protestant R2	--- (0.061)
Reference group = No
Log odds ratios from regression analysis of frequency of prayer and Protestant upbringing.
		Frequently	Regularly	Infrequently
		(several/wk)	(approx. 1/wk)	(1/yr.–1/month)
Evangelical Protestant		1.284 ***	0.897	0.526
Mainline Protestant R2	--- (0.051)
Reference group = Never

* p < 0.05; ** p < 0.01; *** p < 0.001.

displays the results of these regression analyses and shows that those raised in evangelical Protestant denominations are significantly more likely to be more religiously identifying, believing, and practicing across all the six measures.

5. Discussion and Conclusions

I set out with the research of this paper to try and update understandings of the relationship, extent, and variability of parental and childhood religious influences on religious identity, beliefs, and practices in adulthood. For this, I used a prevalent and widely respected dataset for social science research, one with a strong record of producing high-quality research on religion in the United States. Using the only iteration of this dataset in a decade that reported variables of parental religious engagement, I hope to contribute these updated results to augment the corpus of studies into these relationships to add to the existing literature on the intergenerational transmission of religion in contemporary societies. From the results of our analyses, I can generally answer both hypotheses of this paper regarding transmission in the affirmative and can see results and broad themes that confirm previous studies, problematize others, and add new knowledge to the existing empirical literature on religion and society. The results highlight broad themes that validate previous research and produce new insights into the ongoing scholarly discourse on religion and its intergenerational transmission in contemporary societies.

First, while variable across our five aspects of religion, I found strong parental and childhood predictive influence on adult religiosity, religious service attendance, the trajectory of their belief in God, and the strength of their current religious affiliation. However, there was weak evidence for the intergenerational transmission of religious identity, except for those with Jewish parents. The three variables that show the most consistent predictive impact as to the future engagement with religion generally are the religious service attendance of the respondent in childhood, the religious service attendance of the respondent’s mother, and the religious identity of the respondent’s father. In many ways, this echoes the study of Francis (1993), which found that it was mothers’ religious practice that was a more powerful predictor for the practices of their children. However, the important impact of a father’s religious identity (especially Protestant identity, but also others, such as Muslim in the case of prayer) may play an important role in supporting the religious service attendance and private religious practice of family members, perhaps even regardless of how frequently they attend religious services or attend with other family members. This may also echo the findings of Voas and Storm (2012), who emphasize the importance of intergenerational religious transmission of two parents attending church when their children are young. McPhail (2019) likewise has argued for the decreased transmission effect of religious heterogamy among parents. Alternatively, it may be that differential parental effects occur, with sons being more influenced by the religiosity of their fathers (e.g., Gutierrez et al. 2014).

A second broad theme is that in line with prior research, I do not see any strong evidence of a type of ‘backlash’ to parental or childhood religious engagement, with a notable exception of those whose mother was Muslim being significantly less likely to read scriptures in adulthood, perhaps exemplary of differential effects among both religions and parental identity, although more consistent results to this effect would be needed. This is generally to say that there are few findings from the analysis of this paper that show a negative correlation between parental religious influence and religious engagement later in adulthood. Indeed, if anything, I see religious behavior in adulthood that very closely mirrors parental religious behavior. Therefore, I can confidently place the results of this research within the prior literature that emphasizes an intergenerational effect of religion, rather than in those studies that find little evidence of this. For instance, the results of our analysis suggest that while personal religious service attendance in childhood is predictive of future attendance, that future attendance is less than in childhood and instead more closely mirrors parental attendance, especially mother’s religious service attendance patterns from childhood.

Therefore, even if the adult religious behavior (e.g., religious service attendance) is sometimes not as strong as it was when they were twelve, it is not necessarily significantly weaker than the prior generation. While it could be argued that this shows the incremental intergenerational decline of engagement with religion (e.g., Storm and Voas 2012), especially around father’s religious service attendance and adult attendance, it also may also be more an effect of life course influence by which religion takes a more ‘adult’ place within the life of the child as they grow into adulthood and major life events happen (Hardie et al. 2016). These findings also suggest the reflection of a type of ‘religious residue’ in debates of religious decline in traditionally Christian countries, whereby those who were raised in a religious household, but have since left that religion, still exhibit greater measures of religiosity than those who grew up with no religion (Beider 2023). This effect is also particularly strong for those Protestant denominations that are more conservative, or ‘evangelical’. Such findings corroborate previous research regarding this group of more conservative Protestant denominations as distinctly different in cultural disposition and practice, both within and outside the family structure, in a way that facilitates a stronger residue and more consistent transmission across generations (see also Bartkowski 2001; Bartkowski and Ellison 1995; Ellison and Musick 1995; Gay et al. 1996; Sherkat and Ellison 1991).

A final broad theme of the results of this paper is that parental and childhood engagement with religion is much more predictive of religious engagement and attitudes later in life than they are of religious identification later in life, echoing the strong conclusions to this effect by Clements and Bullivant (2022), thereby underscoring the presence of a sort of religious residue (Beider 2023) or potentially a type of ‘chain of memory’ (Hervieu-Léger). Notably, parental religious identification did not serve as a predictor for religious identification in adulthood, except for Jewish respondents. It is also true that even among the independent variables of our analysis, while father’s religious identity showed predictive results, parental religious identification broadly was less predictive than religious service attendance in understanding adult religious belief and engagement. However, there were notable instances of strong predictive results for individuals with Muslim or Jewish parentage. Ultimately, however, it is also the case that these highlight one of the primary limitations of this paper. The subcategories for religious identification were not populated enough to expect robust results from a regression analysis of these categories and may therefore be more of a function of this limitation in the data for religions that are not Protestants, Catholics, or those with no religion than it is indicative of anything else. This is especially a consideration given prior findings of strong intergenerational transmission among religious minority groups and immigrant populations (e.g., Jacob and Kalter 2013).

In conclusion, the objective of this study was to contribute to the ongoing research on the intergenerational transmission of religion by exploring how childhood and parental religious influences shape the religious identity, beliefs, and practices of adult children, as reflected in their responses to the 2018 General Social Survey (GSS). Through multiple regression analyses of this quality dataset of religion in the United States, the findings of this research confirm and highlight the significant influential relationship between childhood and parental religious identity and the engagement of their adult children. Among the most powerful predictors of this transmission are the religious identity of the respondent’s father, the religious service attendance of their mother, and their own religious service attendance as a child. While there was some lessened intensity of religious engagement in adulthood, the respondent’s religious behaviors largely mirrored parental engagement. However, among the variables studied, parental religious identity was the least predictive. These results show that while highly predictive, the predictive mechanisms of intergenerational religious transmission are variable and depend on the type of religious engagement, and which parent is associated with that engagement.