Religion, Convention, and Paternal Involvement

Journal of Marriage and Family 64 (August 2002): 780–792780

W. BRADFORD WILCOX University of Virginia

l

Religion, Convention, and Paternal Involvement

Family scholarship has generally overlooked theinfluence that religion may have on paternal in-volvement. Accordingly, using longitudinal datataken from the National Survey of Families andHouseholds, I examined the influence of religiousaffiliation and attendance on the involvement ofresidential fathers in one-on-one activities, dinnerwith their families, and youth activities and foundreligious effects for each of these three measures.Virtually no evidence was found for a competinghypothesis that these effects are artifacts of a con-ventional habitus such that the type of men whoare more conventional in their patterns of civicengagement are both more religious and more in-volved with their children. However, civic en-gagement is positively related to paternal involve-ment.

In recent years, scholars of family life have in-creasingly turned their attention to fatherhood inan effort to determine how men’s involvement inthe lives of their children bears on child well-be-ing, gender equality, and marital quality (Booth &Crouter, 1998; Coltrane, 1996; Lamb, 1997). Re-search focusing on one critical dimension of thefather role—paternal involvement by residentialfathers—has focused on the ways in which socio-economic status, gender attitudes, and the child’sgender, age, and race influence the level and typeof paternal involvement among residential fathers(Cooksey & Fondell, 1996; Harris & Morgan,

Department of Sociology, University of Virginia, P.O. Box400766, Charlottesville, VA 22904-4766 ([email protected]).

Key Words: civic engagement, convention, paternal in-volvement, religion.

1991; Marsiglio, 1991; Pleck, 1997). But this re-search has largely passed over the ways in whichthe institutional contexts of men’s lives influencepaternal involvement (Doherty, Kouneski, & Er-ickson, 1998). This study focuses on the voluntaryinstitution with which Americans are the most ac-tively affiliated, religion (Putnam, 2000), testingwhether men’s religious culture and participationare related to the extent and type of their paternalinvolvement.

Extrafamilial institutions have long beenknown to shape parental values and behavior. Re-ligious institutions have been particularly influ-ential carriers of family-related culture over thecourse of American history. Their influence con-tinues to this day, even though some of the dis-tinctive parental value orientations that once di-vided Protestants and Catholics have disappeared(Alwin, 1986). Religious institutions have alsobeen powerful sources of family-related social in-tegration, offering family-related activities as wellas informal networks that provide social supportand control for family-oriented behavior (Ellison,1994; Stolzenberg, Blair-Loy, & Waite, 1995).However, there has been little quantitative re-search using nationally representative data that fo-cuses on the influence that religion may have oncontemporary paternal involvement (but see Bart-kowski & Xu, 2000). Using longitudinal datafrom two waves of the National Survey of Fam-ilies and Households (NSFH), this study addressesthis research gap by examining the associationsbetween religion and three areas of paternal in-volvement: one-on-one activities, dinner withone’s family, and youth activities.

Given the longstanding association between re-ligion and convention in American life, it is also

781Religion and Paternal Involvement

possible that any relationships found between fa-therhood and religion are an artifact of an under-lying exogenous conventional effect (Stolzenberget al., 1995). In other words, the kind of men whoare actively involved with their churches and theirchildren may be the kind of men who are moreconventional in general. This conventional effectis explored by testing whether civic engagementaccounts for any of the relationships found be-tween religion and paternal involvement.

Exploring the links between religion, conven-tion, and paternal involvement is particularly im-portant in light of recent research that suggestspaternal involvement is related to a range of pos-itive child outcomes. Paternal involvement is pos-itively associated with the educational attainment,economic achievement, emotional well-being, andthe social competence of children (for a review,see Amato, 1998). In fact, taking into account thejoint contributions of fathers and mothers to theirchildren, Amato concluded, somewhat surprising-ly, ‘‘that fathers are about as important as mothersin predicting children’s long-term outcomes’’ (p.268). By exploring the connections between reli-gion, convention, and paternal involvement usingdata from the 1987–1988 and 1992–1994 wavesof the NSFH, this study should also provide someinsight into the possible relationships between re-ligion, civic engagement, and child well-being.

RELIGION, CONVENTION, AND FATHERHOOD

Religious Participation and Culture: Seedbeds ofPaternal Involvement?

Because little research has focused on religion andfatherhood (Bartkowski & Xu, 2000), I rely onthe literature on religion and parenting to framemy hypotheses regarding religion and fatherhood.Recent research on the relationship between reli-gion and parenting suggests that there are two re-ligious factors that influence parenting: religiousparticipation and the distinctive cultures of reli-gious traditions (Alwin, 1986; Clydesdale, 1997;Ellison, Bartkowski, & Segal, 1996; Pearce & Ax-inn, 1998; Wilcox, 1998).

Religious participation can be taken as an in-dication of the extent to which a person is inte-grated into the institutional life of a particular re-ligious institution (Durkheim, 1897/1951; Pearce& Axinn, 1998). Such integration could influenceparenting values and behavior in a number ofways. First, the rituals and discourse persons en-counter in religious institutions—from baptisms to

Father’s Day sermons—dramatize the moral rela-tions that bind parents to their children, often en-dowing them with a transcendent character. Sec-ond, by offering worship services, educationalprograms, and social activities for families, reli-gious institutions offer opportunities for parents tospend time with their children. Third, the family-oriented informal networks that parents encounterin religious institutions may provide social affir-mation of parenting, guidance on questions relatedto child-rearing concerns, and social sanctions forparents who depart from child-rearing norms(Pearce & Axinn; Stolzenberg et al., 1995). Fi-nally, by imposing a meaningful order upon ev-eryday life and more liminal periods (illness, un-employment, and so on), religious participationmay help parents deal with stresses that can oth-erwise harm parent-child interactions (Ellison,1994).

A number of recent studies have found thatreligious participation is linked to parental behav-iors and values (Alwin, 1986; Bartkowski & Xu,2000; Clydesdale, 1997; Pearce & Axinn, 1998;Wilcox, 1998). Alwin found that parents who par-ticipate in church activities are more likely to val-ue obedience in their children than other parents.Parents who attend church are more likely to beinvolved with their children’s education (Clydes-dale). Maternal church attendance, as well as theimportance of religion to the mother, is associatedwith improvements in the mother’s perception ofthe quality of the mother-child relationship (Pearce& Axinn). Church attendance has also been linkedto a warm, expressive style of parenting (Wilcox).Fathers who attend church frequently are morelikely to monitor their children, to praise and hugtheir children, and to spend time with their chil-dren (Bartkowski & Xu). Thus, religious partici-pation, which is understood here as an indicatorof religiosity, would seem to foster an authorita-tive, active, and expressive style of parenting.

Although religious institutions may uniformlyfoster a family-focused ethos that is associatedwith higher levels of parental involvement, it isalso possible that the culture (ideology, rituals,and norms) of particular religious traditions mayincorporate distinctive practices and ends thatguide parental behavior. First, the worship styleand larger religious ethos associated with differentreligious traditions may shape the style of parentalbehavior by providing models for particular fa-milial practices. Second, religious discourse maystress particular religious or family-related ends

782 Journal of Marriage and Family

that, in turn, guide the ways that parents spendtime and interact with their children.

The empirical research regarding the influenceof distinctive religious cultures on parental behav-iors and values is mixed. Although Alwin (1986),Clydesdale (1997), and Pearce and Axinn (1998)found evidence that religious participation shapesparenting behavior and values, they found no ev-idence that religious culture per se shapes parentalvalues and practices. Alwin found that the differ-ences that once marked Catholic and Protestantchild rearing orientations have disappeared, large-ly as a consequence of marked acculturation andsocioeconomic mobility among Catholics sincethe 1960s. Alwin concluded, ‘‘Lenski’s (1963) ‘re-ligious factor’ of the 1950s has all but vanishedin American society, at least with respect to pa-rental socialization values’’ (p. 423). Similarly,Pearce and Axinn found that religious affiliationdid not influence the mother’s or child’s percep-tion of the quality of the mother-child relationship.Moreover, in looking at a range of family behav-iors, Clydesdale found that theological conserva-tism was not associated with parental involvementin children’s education. This literature would sug-gest that religiosity, rather than a commitment toa distinctive religious culture, is associated withgreater investments in fathering.

However, a growing body of literature on re-ligion and parenthood suggests a competing hy-pothesis about the influence of religious cultureon parental involvement (Ellison et al., 1996; Wil-cox, 1998). This literature has focused on conser-vative Protestantism, which has devoted signifi-cant pastoral attention to family life since the1970s. Conservative Protestant churches and par-achurch ministries have stressed values such astraditional gender attitudes, strict discipline, ex-pressive parenting, and parental involvement.Moreover, because of their pietist tradition of wor-ship and an increasingly therapeutic approach torelationships, conservative Protestant churcheshave an expressive ethos that may carry over intofamily life. Indeed, this research literature sug-gests that the family culture produced by this sub-culture is associated with higher rates of corporalpunishment (Ellison et al.) and a warm, expressivestyle of parent-child interaction in nondisciplinarysituations (Wilcox).

Although this research has generally not fo-cused on men, much of the discourse produced inthis conservative Protestant subculture is aimedspecifically at fathers. Although conservativeProtestant groups generally stress men’s tradition-

al role as the primary breadwinner and head ofhousehold, they also place a great deal of empha-sis on men’s roles as husbands and fathers. Theconservative Protestant subculture orients fathersto their children in at least two ways. First, fathersare exhorted to model for their children the lovethat God has for persons by being an active, ex-pressive, and strict parent (Bartkowski & Xu,2000; Wilcox, 1998). Second, the expressive ethosthat now characterizes large sectors of conserva-tive Protestantism is associated with a focus onrelationships that may translate into greater pater-nal involvement (Wilcox). Indeed, one study pub-lished on religion and fatherhood finds that con-servative Protestant fathers combine a strictapproach to discipline with a warm, affective ap-proach to nondisciplinary interactions (Bartkows-ki & Xu). Accordingly, a conservative Protestantaffiliation should be associated with increased pa-ternal involvement.

There is considerably less research on family-related culture in mainline Protestant and Catholicchurches. Since the 1970s, mainline and Catholicchurches have generally not devoted as muchsymbolic and pastoral attention to family life asconservative Protestant churches (Browning,1995). Mainline churches have focused more onpublic issues like social justice, peace, and envi-ronmentalism than on family life (Browning). Inthis period, American Catholic churches have de-voted less attention to family life, largely as a con-sequence of disagreement within their church overfamily and sexual matters since Vatican II (Gree-ley, 1990).

Of course, many mainline and Catholic church-es continue to foster a Golden Rule Christianitycentered on an ethic of care for family and friends(Ammerman, 1997). This brand of Golden RuleChristianity sidesteps contested moral and sexualmatters but stresses a generic ethic of love thatmay promote greater paternal involvement. More-over, these churches tend to be child-centered,with vibrant Sunday schools, children’s choirs,and other family activities. Thus, the Golden RuleChristianity found in many Catholic and mainlineProtestant churches may be associated with great-er paternal involvement. Nonetheless, I predictthat mainline Protestant and Catholic men will notbe as consistently involved with their children asconservative Protestant men, who are more likelyto be exposed to a sustained and specific ideologyemphasizing family involvement. Thus, this liter-ature review suggests two competing hypotheses,which run as follows:


Hypothesis 1. Because of the importance of reli-gious culture in shaping paternal behavior, con-servative Protestant men display greater levels ofpaternal involvement than unaffiliated men andthey are more consistently involved with theirchildren than mainline Protestant and Catholicmen. Nonetheless, men affiliated with Catholicand mainline churches are more involved withtheir children than unaffiliated men.

Hypothesis 2. Because religion promotes family-focused behavior uniformly among parents,church attendance—an important indicator of re-ligiosity—is associated with greater paternal in-volvement. Moreover, any association between re-ligious culture and paternal involvement is aconsequence of an underlying association betweenreligiosity and paternal involvement. Thus, con-trols for church attendance will eliminate any di-rect effects of religious culture on paternal in-volvement.

Some scholars who have turned their attentionto the relationship between religion and family lifehave perceptively argued that this reciprocal re-lationship may ‘‘result from common antecedentsrather than from a direct causal link’’ (Clydesdale,1997; Stolzenberg et al., 1995, p. 85). The mostlikely suspect is a kind of conventional effect thatincludes a range of prosocial behavior and ties,where religiosity is just one indicator of a person’sintegration into the broader social order (Durk-heim, 1897/1951). In the American context, thisconventional effect may be rooted in a kind ofmiddle-class convention that reached its zenith inthe 1950s. This conventional style was made upof an assemblage of values, norms, and practicesthat included church-going, civic engagement, anda familial orientation. With respect to religion andthe family, the paradigmatic formulation of thismiddle-class convention was ‘‘the family thatprays together stays together.’’ Indeed, the 1950sexperienced a dramatic increase in three indicatorsof this conventional style: family-focused behav-ior, civic engagement, and religious practice (Mar-ler, 1995; Putnam, 2000).

Thus, the reciprocal relationships found be-tween religion and a family-centered lifestyle maybe an artifact of a conventional 1950s-style hab-itus (Bourdieu, 1977, p. 82). Following Bourdieu,habitus is used to refer to a ‘‘system of lasting,transposable dispositions’’ that are structured bypast experiences and, in turn, structure ongoingperceptions, values, and practices (Bourdieu, pp.82–83). In the case at hand, men exposed to a

1950s-style conventional habitus may be living inways that reproduce the cultural patterns they en-countered while growing up. Moreover, the pullof such a habitus would be strongest when menand women are in that stage of the life course thatcoincides with their own childhood; that is, whenthey have children of their own. Thus, any linkbetween religion and family life may in fact be aproduct of an underlying conventional habitus thatstructures involvement in the family, religion, andcivil society.

But how relevant is this 1950s-style habitus tocontemporary fathers? Despite marked socialchange since the 1950s, there is evidence to sug-gest that the kind of conventional habitus outlinedabove continues to shape the contemporary socialpractice of American fathers. Recent work byNock (1998) suggests that, for men, this type ofprosocial habitus is tied to the institution of mar-riage and childrearing. Nock shows how marriageand childrearing continue to be associated withheightened commitment on the part of men to civ-ic engagement. Thus, the following is hypothe-sized:

Hypothesis 3. A conventional habitus marked byreligious attendance, civic engagement, and fam-ily-centered activity accounts for any link betweenreligion and paternal involvement. This possibilityis particularly important in light of Bartkowskiand Xu’s (2000) finding that religious participa-tion is linked to paternal involvement. If religiousinvolvement simply amounts to a form of conven-tional behavior, civic engagement can be expectedto eliminate the net effect of religion on paternalinvolvement. However, if religion has a uniqueeffect on paternal involvement, then religious ef-fects should remain robust to models includingcivic engagement.

METHOD

I relied on The 1987–1988 NSFH-1, a nationallyrepresentative survey of 13,017 adults age 19 andover (Sweet, Bumpass, & Call, 1988), and its fol-low-up survey in 1992–1994 (NSFH2; Bumpass& Sweet, 1995). The response rate for NSFH2primary respondents was 82%. Using data fromNSFH1 and NSFH2, the analyses focused on onesubset of the data: 1,019 primary respondents whowere residential fathers in both waves of NSFHand residential fathers of school-age children(ages 5–18) during NSFH2. Specifically, residen-tial fathers are defined as biological, adoptive, or


step-fathers that are living with their children atthe time of the survey.

I relied on respondent reports of activity inthree areas of paternal involvement at NSFH2(1992–1994) for dependent variables: one-on-oneinteraction, dinner together, and youth-related ac-tivities. One-on-one interaction, was measured byreliance on respondents’ reports of involvement inthe following four activities: ‘‘leisure activitiesaway from home,’’ ‘‘at home working on a projector playing together,’’ ‘‘private talks,’’ and ‘‘help-ing with reading or homework.’’ Responses rang-ing from 1 5 never or rarely to 6 5 almost everyday were summed to create a scale based on themean response to each of the items (Cronbach’salpha 5 .78).

For dinner together, the respondent’s answersto the following question were relied on: ‘‘Howmany evenings last week did your whole familyliving here eat dinner together?’’ Responses werecoded from 0 to 7. Fathers were also asked howmany hours per week they spent in an averageweek as a participant, advisor, coach, or leader inthe following youth-related activities: school ac-tivities, community youth groups (e.g., Scouts),sports activities, and religious youth groups. Be-cause of rightward skewedness, I relied on the nat-ural log of the sum total of hours devoted to thesefour activities to measure weekly paternal in-volvement in youth-related activities. Values of 0(i.e., no time devoted to youth-related activities),were recoded to .01 before transformation. TheNSFH does not specifically ask parents if theyouth activities they participated in involved theirown children. However, the questions about youthactivities come in the middle of a battery of par-enting questions. For this reason, as well as thefact that the sample is made up of residential fa-thers, it is safe to assume that most fathers arereporting activities that involved their own chil-dren. The NSFH also does not ask parents to in-dicate if youth activities that are not overtly reli-gious, such as Boy Scouts, are sponsored by areligious organization. Thus, this measure ofyouth-related activities may incorporate a greaterdegree of religious influence than is readily ap-parent in the data.

The sample was divided using data fromNSFH1 (1987–1988) into the following religiouscategories: conservative Protestant, mainlineProtestant, Catholic, and unaffiliated. More spe-cifically, respondents who identified with South-ern Baptist, Assembly of God, Pentecostal, Mis-sionary Alliance, Christian Reformed, and a

number of other fundamentalist and evangelicalchurches were coded as conservative Protestant.Respondents who indicated an Episcopal, Luther-an, Presbyterian, Methodist, or Congregational af-filiation were coded as mainline Protestants. Re-spondents who indicated no religious affiliationwere coded as unaffiliated, and they serve as thecomparison category in my analyses. Jewish andOther (Mormon, Unitarian, Muslim, and so on)respondents were excluded from the analyses be-cause of the small number of respondents in bothof these categories. Respondents were also asked‘‘How often do you attend religious services?’’This question measured frequency of church at-tendance, which ranged from 0 5 never to 8 5several times a week.

A measure of civic engagement from NSFH1(1987–1988) tested whether any effects of religionon paternal involvement might be an artifact ofmen’s conventionality. Civic engagement wasmeasured by participation, from 1 5 never to 55 several times a week, in civic groups (e.g., fra-ternal or political organizations), work-relatedgroups (e.g., professional associations), and ser-vice organizations. The mean responses to elevenmeasures of nonreligious civic activities wereused to create a scale coded from 1 to 5 (Cron-bach’s alpha 5 .58). Four types of religious andyouth-related civic participation were excludedbecause they may tap the same behavior examinedin the measures of church attendance and youth-related activity.

The following NSFH1 variables, which areknown to be associated with paternal involvement(Marsiglio, 1991) and might otherwise confoundany relationships between religion and paternal in-volvement were controlled for: respondent’s edu-cation (from high school to graduate school, cod-ed from 1 to 6); age (in years); respondent’shousehold income (logged); race–ethnicity of therespondent (Black; Hispanic; reference category5 White–Anglo); region (South; Northeast; NorthCentral; reference category 5 West). Also contro-led for was gender traditionalism at NSFH1,which was measured by respondent’s attitudes to-ward the following items: ‘‘It is much better foreveryone if the man earns the main living and thewoman takes care of the family and home,’’ ‘‘pre-school children are likely to suffer if their motheris employed,’’ ‘‘if a husband and wife both workfull-time, they should share housework tasksequally,’’ and (dis)approving of ‘‘mothers whowork full-time when their youngest child is underage 5.’’ The scale is based on the mean response


TABLE 1. DESCRIPTIVE STATISTICS OF ALL VARIABLES

M SD

Control variables, NSFH1BlackHispanicEducationAgeHousehold income (logged)SouthNortheastNorth CentralWestGender traditionalism

0.0970.1182.639

36.42710.0730.3540.1830.2680.1963.028

0.2970.3231.3217.0361.8010.4780.3870.4430.3970.714

Control variables, NSFH2MarriedAge of youngest childPreschool childrenSchool-age childrenAll step childrenBlendAll biological childrenAll male childrenMixed genderAll female childrenEmployedShift workWeekly hours of employment

0.9439.2000.2491.8230.1380.0540.8180.3040.4570.2390.8820.412

41.758

0.2324.9180.5410.8160.3350.3650.4010.4600.4980.4260.3220.493

18.430

Religious factors, NSFH1Conservative ProtestantCatholicMainline ProtestantNo affiliationChurch attendance

0.2490.3000.3260.1294.011

0.4330.4580.4690.3352.872

Convention, NSFH1Civic engagement 1.156 0.226

Dependent variables, NSFH2One-on-one interaction (1 to 6)Dinner together (0 to 7)Youth-related activities

(hours, logged)

3.6654.445

21.302

1.0572.246

3.045

Note: NSFH1 5 National Survey of Family Households,Wave 1. NSFH2 5 National Survey of Family Households,Wave 2. N 5 1,019.

to these four items and was coded from 1 to 5(Cronbach’s alpha 5 .61).

The following family characteristics at NSFH2were controlled for because they are also knownto be related to paternal involvement (Cooksey &Fondell, 1996; Harris & Morgan, 1991): biologi-cal composition of the family (a blend of biolog-ical and/or adopted children and step-children; allbiological and/or adopted children; reference cat-egory 5 all step-children); gender of children (allmale children; a mix of male and female children;reference category 5 all female children); andmarital status of respondent (married; referencecategory 5 single father). Finally, the followingemployment characteristics at NSFH2 were con-trolled for because employment status and sched-ules are likely to influence paternal involvement:a dummy variable measuring employment, a dum-my variable tapping shift work, and average week-ly hours of employment.

For each of the dependent variables, a seriesof hierarchical ordinary least squares (OLS) re-gression models were estimated to determine theeffects of religious factors on residential paternalinvolvement even after controlling for a range ofpotential covariates—including the measure ofconvention. The first model, which includes mea-sures of religious affiliation, tests the effect of re-ligious culture on paternal involvement. The sec-ond model, which includes a measure of churchattendance, tests the independent effect of reli-gious participation on paternal involvement andexplores the extent to which the effects of reli-gious culture are mediated by religious participa-tion. The third model, which includes a measureof civic engagement, tests whether religious ef-fects on paternal involvement are artifacts of aconventional habitus.

RESULTS

Table 1 presents means and standard deviationsfor residential paternal involvement, religious af-filiation, church attendance, convention, fathercharacteristics, and family characteristics. Thedata are weighted to adjust for sampling proba-bilities.

Table 2 reports regression results for modelsexploring the links between religion, convention,and residential paternal involvement in one-on-one activities. The hypotheses was evaluated byrunning three models estimating the effects of re-ligious variables measured in 1987–1988(NSFH1) on paternal involvement 5 years later

(NSFH2). Models 1–3 indicate that, consistentwith Hypothesis 1, men with a conservative Prot-estant affiliation at NSFH1 are consistently morelikely than unaffiliated men to engage in one-on-one interaction with their school-age children atNSFH2. In other words, the positive effect of con-servative Protestant affiliation persists even aftercontrolling for attendance and civic engagement.These findings are also inconsistent with Hypoth-eses 2 and 3, which assumed, respectively, thatmeasures of attendance and convention wouldeliminate the effect of religious culture.

Moreover, Models 1–3 indicate that conserva-


TABLE 2. COEFFICIENTS FROM OLS REGRESSION MODELS: PATERNAL INVOLVEMENT IN ONE-ON-ONE ACTIVITIES WITH

SCHOOL-AGE CHILDREN (NSFH2)

Model 1

b SE

Model 2

b SE

Model 3

b SE

Control variables, NSFH1BlackHispanicEducationAgeHousehold income (logged)SouthNortheastNorth CentralGender traditionalism

20.0250.224*0.170***

20.013*0.027

20.12520.03820.10320.072

0.1160.1100.0260.0050.0180.0870.1020.0910.044

20.0110.252*0.182***

20.012*0.027

20.10720.03620.08620.062

0.1160.1110.0260.0050.0180.0870.1020.0910.044

20.0770.248*0.174***

20.014**0.023

20.10520.07120.12320.057

0.1150.1100.0260.0050.0180.0860.1010.0910.043

Control variables, NSFH2MarriedAge of youngest childPreschool childrenSchool-age childrenBlendAll biological childrenAll male childrenMixed genderEmployedShift workWeekly hours of employment

20.511***20.104***20.510***20.179***

0.331*0.402***0.301***0.1300.1580.065

20.010**

0.1460.0100.0820.0460.1640.1010.0840.0840.1680.0660.003

20.501**20.104***20.503***20.169***

0.326*0.416***0.302***0.1260.1520.073

20.009**

0.1460.0100.0810.0460.1640.1010.0840.0840.1680.0660.003

20.518***20.106***20.498***20.145**

0.0950.359*0.300***0.1180.1560.069

20.010**

0.1450.0100.0810.0450.1620.1480.0830.0830.1660.0650.003

Religious factors, NSFH1a

Conservative ProtestantCatholicMainline ProtestantChurch attendance

0.252*b

0.0780.033c

0.1050.1010.098

0.342**b

0.1610.103c

20.027*

0.1120.1070.1020.012

0.305**b

0.1650.108c

20.028*

0.1120.1060.1010.017

Convention, NSFH1Civic engagement 0.623*** 0.135

InterceptAdjusted R2

5.269***0.220

0.358 5.146***0.223

0.361 4.643***0.235

0.375

Note: NSFH1 5 National Survey of Family Households, Wave 1. NSFH2 5 National Survey of Family Households,Wave 2. N 5 982.

aThe comparison category is unaffiliated. b,c Coefficients with different superscripts are significantly different (p , .05).*p , .05. **p , .01. ***p , .001.

tive Protestant fathers are more involved in one-on-one interaction than mainline Protestant fathersbut not Catholic fathers. However, Table 2 pro-vides no evidence that mainline Protestant andCatholic affiliations in 1987–1988 are associatedwith greater levels of paternal involvement in1992–1994. Thus, Table 2 suggests that the dis-tinctive stress on familial involvement found inthe conservative Protestant subculture leads tohigher levels of paternal investment in one-on-oneactivities.

Models 2 through 3 suggest that church atten-dance is not linked to paternal involvement inone-on-one activities. In fact, contrary to Hypoth-esis 2, church attendance in 1987–1988 is nega-tively associated with paternal involvement in

1992–1994. This finding is particularly surprisingbecause it runs counter to the cross-sectional studydone by Bartkowski and Xu (2000), which founda positive relationship between church attendanceand one-on-one interaction using cross-sectionaldata from NSFH1.

But Model 3 does show that the measure of aconventional habitus is positively related to one-on-one involvement for residential fathers. Spe-cifically, civic engagement in 1987–1988 is a sig-nificant and very powerful predictor of paternalinvolvement in 1992–1994. This means that resi-dential fathers who are actively involved in thecivic life of their communities are also more likelyto engage in one-on-one activities such as home-work help with their own children. Nevertheless,


TABLE 3. COEFFICIENTS FROM OLS REGRESSION MODELS OF FAMILY DINNER TOGETHER FOR FATHERS WITH

SCHOOL-AGE CHILDREN (NSFH2)

Model 1

b SE

Model 2

b SE

Model 3

b SE


21.224***0.311

20.0490.0100.0680.1950.2190.081

20.113

0.2570.2460.0580.0120.0410.1940.2290.2050.099

21.246***0.270

20.0670.0080.0680.1630.2160.055

20.129

0.2570.2470.0590.0120.0410.1950.2280.2060.099

21.293***0.253

20.0770.0080.0670.1600.1690.012

20.118

0.2570.2460.0590.0120.0410.1950.2290.2060.099


0.20520.132***20.31020.191

0.2770.732**0.3170.190

20.06220.594***20.002

0.3260.0230.1860.1040.3670.2260.1880.1880.3740.1480.007

0.18220.133***20.31920.206*

0.2830.708**0.3120.195

20.04620.607***20.002

0.3260.0230.1850.1040.3670.2260.1880.1880.3740.1480.007

0.15620.137***20.30520.16720.615

0.1640.3150.198

20.03120.618***20.003

0.3260.0230.1850.1020.3710.3400.1870.1860.3730.1470.007



0.528*0.3800.253

0.2390.2280.222

0.3810.2410.1410.044

0.2550.2430.2320.027

0.3330.2630.1590.044

0.2550.2420.2310.027

Convention, NSFH1Civic engagement 0.549 0.303

InterceptAdjusted R-squared

4.447***0.097

0.804 4.662***0.098

0.814 4.635***0.104

0.872

Note: NSFH1 5 National Survey of Family Households, Wave 1. NSFH2 5 National Survey of Family Households,Wave 2. N 5 1,019.

aThe comparison category is unaffiliated.*p , .05. **p , .01. ***p , .001.

as indicated above, Model 3 provides very littleevidence that the link between religion and fatherinvolvement in one-on-one activities is an artifactof convention, because the conservative Protestantcoefficient declines by only 11% from Model 2 toModel 3.

Table 2 also indicates that education, havingbiological children, adopted children, or both,having all male children, and being Hispanic arepositively associated with one-on-one paternal in-teraction. Father’s age, having older children or alarger family, being married, and spending longhours in paid work are negatively related to thistype of interaction.

Table 3, which reports regression results forfathers’ dinner with their families, lends some

support to Hypothesis 1. Model 1 indicates that aconservative Protestant affiliation in 1987–1988predicts higher levels of dinner together in 1992–1994. In fact, over an average year, conservativeProtestant fathers will have roughly 27 more din-ners with their families than unaffiliated fathers.(Contrary to Hypothesis 1, mainline Protestantand Catholic affiliations at NSFH1 are not relatedto dinner together at NSFH2; however, mainlineProtestant and Catholic fathers are not signifi-cantly different from conservative Protestant fa-thers in their propensity to have dinner with theirentire family.) Accordingly, Table 3 provides ad-ditional evidence that the family-oriented cultureof conservative Protestantism promotes paternalinvolvement.


However, once attendance is included in Model2, this affiliation effect is completely attenuated,lending some support to Hypothesis 2. Thus,Models 1–3 suggest that the positive effect of con-servative Protestant affiliation on dinner partici-pation is mediated, at least in part, through thedirect effect of church attendance.

However, contrary to Hypothesis 2, Models 2–3 indicate that church attendance is not signifi-cantly related to dinner together. Also, contrary toHypothesis 3, Model 3 provides no indication thata conventional habitus is associated with dinnertogether. Civic engagement in 1987–1988 is notrelated to dinner together 5 years later for resi-dential fathers. Turning to the control variables,Table 3 indicates that African-American fathers,fathers employed in shift work, and fathers witholder children are consistently less likely to havedinner with their entire family. Fathers with allbiological children, adopted children, or both, aremore likely to have dinner together with theirfamily.

Table 4 presents the results from regressionsestimating the effects of religion on fathers’ in-volvement in youth-related activities. Model 1 in-dicates that conservative Protestant and Catholicfathers, but not mainline Protestant ones, are sig-nificantly more involved in youth-related activitiesthan unaffiliated fathers. Moreover, Catholic fa-thers are significantly more involved in youth-re-lated activities than mainline Protestant fathersand remain so even after controlling for churchattendance and civic engagement.

Thus Model 1 provides partial support for Hy-pothesis 1, which suggested that the family focusof conservative Protestant culture translates intogreater paternal involvement. Here again, how-ever, conservative Protestant fathers are not sig-nificantly more involved than Catholic fathers (seealso Table 2). Indeed, Catholic fathers are moreinvolved than conservative Protestants fathers inthis domain of fathering, though the difference isnot statistically significant.

However, consistent with Hypothesis 2, the ef-fects of religious affiliation are mitigated by theinclusion of church attendance. Models 2 and 3indicate that church attendance, which is positive-ly related to paternal involvement in youth-relatedactivities, accounts for about half of the positiveeffects of conservative Protestant and Catholic af-filiation on such involvement. The addition ofchurch attendance also renders the Catholic andconservative Protestant coefficients nonsignifi-cant. These findings indicate that the affiliation ef-

fects are mediated by the positive direct effect ofchurch attendance. Thus, consistent with Hypoth-esis 2, religious participation is linked to greaterpaternal involvement in youth-related activities.

Here, the institutional effect of church atten-dance is, in all likelihood, not simply a conse-quence of family-centered discourse and socialnetworks found in churches. Many churches spon-sor religious youth groups, Scout groups, andsports teams. Thus, fathers who are actively affil-iated with churches may have greater opportuni-ties and incentives to get involved in youth-relatedactivities in their own churches.

Model 3 displays the effect of the conventionalhabitus, measured by civic engagement. Civic en-gagement is positively related to paternal involve-ment in youth-related activities. In other words,residential fathers who are active in the civic lifeof their communities are also more likely to beinvolved in youth-related activities. But Model 3only provides limited support for Hypothesis 3 in-sofar as the measures of convention reduce theeffect of church attendance by less than 10%. Thissuggests that the effect of church attendance onpaternal involvement in youth-related activities islargely robust to conventional effects. In otherwords, the link between religiosity and this typeof paternal involvement is not simply a conse-quence of an underlying link between the conven-tional habitus and paternal involvement.

Table 4 also indicates that education, havingmale children, and being African-American arepositively related to paternal involvement inyouth-related activities. However, fathers witholder children, fathers with more traditional gen-der attitudes, and fathers with more preschoolchildren are less likely to be involved in such ac-tivities.

DISCUSSION

This study poses two central questions: (a) are re-ligious culture and participation associated withresidential paternal involvement? and (b) can theassociations between religion and fatherhood beattributed to an underlying conventionality effect?This study indicates that religion is related to pa-ternal involvement in all three areas that were ex-amined: one-on-one engagement, dinner withone’s family, and volunteering for youth-relatedactivities. The findings presented here also suggestthat these religious effects are, for the most part,not artifacts of what I call a conventional habitus.In particular, religious predictors of paternal in-


TABLE 4. COEFFICIENTS FROM OLS REGRESSION MODELS ON PATERNAL INVOLVEMENT IN (LOGGED) YOUTH-RELATED

ACTIVITIES (NSFH2)

Model 1

b SE

Model 2

b SE

Model 3

b SE


0.981**20.010

0.382***0.008

20.0690.0270.160

20.20620.287*

0.3720.3700.0810.0170.0590.2810.3350.2920.138

0.914*20.142

0.314***0.005

20.07120.087

0.13020.31320.331*

0.3690.3680.0820.0170.0590.2800.3320.2910.137

0.70120.171

0.277**0.001

20.08020.098

0.04320.44420.304*

0.3650.3630.0810.0170.0580.2750.3290.2880.135


0.18820.120***21.017***

0.1620.4300.5641.415***1.032***0.7580.0440.001

0.4760.0330.2660.1470.5260.3250.2700.2670.5460.2100.010

0.13820.125***21.053***

0.1120.4590.4971.408***1.051***0.794

20.01720.001

0.4720.0330.2640.1460.5220.3230.2680.2650.5420.2090.010

0.09220.127***21.034***

0.1600.5370.8571.403***1.040***0.827

20.04220.002

0.4660.0330.2590.1420.5090.4630.2620.2600.5340.2060.009



0.976**1.132**b

0.503c

0.3310.3210.306

0.4530.661b

0.099c

0.150***

0.3540.3400.3210.038

0.2940.648b

0.044c

0.140***

0.3500.3340.3160.038

Convention, NSFH1Civic engagement 2.205*** 0.419

InterceptAdjusted R-squared

23.148**0.107

1.133 22.467*0.121

1.138 24.933***0.148

1.211

Note: NSFH1 5 National Survey of Family Households, Wave 1. NSFH2 5 National Survey of Family Households,Wave 2. N 5 912.

aThe comparison category is unaffiliated. b,c Coefficients with different superscripts are significantly different (p , .05).*p , .05. **p , .01. ***p , .001.

volvement were robust to the specification of con-trols for civic engagement in two out of the threemeasures that were examined. This means that re-ligion appears to make a unique contribution topaternal involvement above and beyond its statusas a conventional activity. In all likelihood, thespecific attention that religious institutions dedi-cate to family life accounts for the religious ef-fects found in this study.

Religious culture—in this case, conservativeProtestant affiliation—has an independent effecton residential paternal participation in one-on-oneactivities, even after controlling for generic reli-giosity and a conventional habitus. ConservativeProtestant fathers are more likely to be involvedwith their children in personal activities such as

personal talks than unaffiliated and mainline Prot-estant men. Moreover, conservative Protestantmen are also more likely than unaffiliated men tohave dinner with their children and to participatein youth-related activities. These findings are par-ticularly striking in light of the gender tradition-alism championed by conservative Protestantchurches and embraced to a large degree amongconservative Protestant rank-and-file. Neverthe-less, these findings are consistent with resultsfound in earlier research on conservative Protes-tantism (Bartkowski & Xu, 2000; Wilcox 1998).

However, the models for paternal participationin youth-related activities suggest that religiosityis a more salient predictor than religious cultureof some forms of paternal involvement. Specifi-


cally, I found that the positive effect of conser-vative Protestant and Catholic affiliations onyouth-related activities dramatically declined and,in most cases, disappeared when controlling forchurch attendance. These results lend credence tothe stress that Alwin (1986), Clydesdale (1997),and Pearce and Axinn (1998) place on the powerof religiosity as an influence on family life. In alllikelihood, the normative discourse, family-fo-cused social networks of support and control, andpsychological support associated with religious in-stitutions help explain the link between religiousparticipation and paternal involvement in youth-related activities (Christiano, 2000).

An alternative possibility is that paternal reli-gious participation is only associated with youth-related activities because religious youth groupsare included in the measures of such activities.Youth activities were separated into religious andnonreligious activities and additional models wererun to test this possibility. But logistic models in-dicate that paternal church attendance is associ-ated with higher rates of involvement in both re-ligious and nonreligious youth activities (analysisnot shown). Of course, it is possible that somenonreligious activities, such as Scouts, are spon-sored by religious organizations, because theNSFH did not ask parents if such activities weresponsored by a religious organization. Thus, thisstudy may overestimate the contribution of reli-gious participation to involvement in nonreligiousyouth activities.

Surprisingly, evidence was also found that re-ligious participation is negatively associated withpaternal involvement in one-on-one interaction.As noted earlier, this finding runs contrary to thecross-sectional research conducted by Bartkowskiand Xu (2000). Perhaps exposure to the more tra-ditional messages and social networks associatedwith religious institutions is negatively related tospecific types of paternal involvement that areseen as more progressive: in this case, one-on-oneinteraction. Other types of interaction—dinnerwith one’s family and youth-related activities, forinstance—may be more in keeping with tradition-al expressions of paternal care.

Nonetheless, the argument about the power ofreligiosity to foster family-focused behavior in auniform fashion does not account for the findingsthat only conservative Protestant and Catholic af-filiations, and not mainline Protestant affiliations,are associated with higher levels of paternal in-volvement in family dinners and youth-related ac-tivities. I suspect that two factors account for the

distinctive affiliation effects found in this article.First, ancillary regression analyses reveal thatconservative Protestant and Catholic residentialfathers are significantly more likely at NSFH1 toattend church (analysis not shown). Thus, thesetwo religious groups appear to enjoy greater reli-gious strength than mainline Protestantism. Thisvitality may translate into more opportunities andincentives for residential fathers to devote them-selves to their children.

Second, it may be that high-attending Catholicand conservative Protestant fathers are both moreinvolved but have distinctive culturally based mo-tivations for their involvement. In an effort to testthis, I ran further regression analyses to see ifCatholic and conservative Protestant men focusedtheir paternal involvement in youth activities indifferent areas and found that conservative Prot-estant fathers were disproportionately involved inreligious youth groups, whereas Catholic fatherswere disproportionately involved in secular youth-related activities (analysis not shown). The con-servative Protestant commitment to religiousyouth groups is in keeping with research that sug-gests members of this subculture focus their timeand attention on religiously centered activities af-fecting their own families and churches at the ex-pense of nonreligious civic activities (Putnam,2000). By contrast, the Catholic commitment tononreligious youth activities appears to be inkeeping with the more worldly, communitarianethos found in Roman Catholicism (Greeley,1990). Thus, it appears that distinctive dimensionsof religious culture do have indirect and direct ef-fects on patterns of paternal participation in youth-related activities.

This study also provides some evidence thatthe relationship between religion and paternal in-volvement is not an artifact of convention, be-cause the indicator of a conventional habitus gen-erally did not eliminate the net effects of religiousvariables when it was added to the statistical mod-els. This is a particularly important finding be-cause Bartkowski and Xu (2000) have recentlyfound positive relationships between religious par-ticipation, as well as conservative Protestant affil-iation, and paternal involvement. This suggeststhat religion exercises a unique role in influencingpaternal involvement above and beyond its statusas a conventional activity.

Given the longitudinal design of this study andthe fact that religious effects vary by denomina-tion, I am fairly confident that my findings re-garding the influence of religion on paternal in-


volvement do not incorrectly specify the directionof causality. Of course, I do not have any infor-mation on religious identity or practice for theseresidential fathers prior to the birth of their chil-dren. Thus, it could be that birth of a child gen-erates a strong family orientation among somemen, which in turn fuels religious commitmentand paternal involvement (Dollahite & Hawkins,1998). Still, the fact that most of the religious ef-fects vary by denomination, which is not likely tobe related to such a family-focused factor, stronglysuggests that religion has an independent effect onpaternal involvement.

Although the primary focus of this study is notthe effect of a conventional habitus on paternalinvolvement, civic engagement is strongly asso-ciated with paternal involvement in a wide rangeof activities. Accordingly, future research shouldexplore in greater detail the mechanisms that pushsome men into a conventional habitus marked byhigh levels of civic engagement and paternal in-volvement.

Finally, this study may have important impli-cations for research on child well-being. The lit-erature on paternal involvement indicates thatsuch involvement is positively associated with arange of beneficial child outcomes (Amato, 1998;Lamb, 1997). This study suggests that religion—measured by affiliation and church attendance—may be indirectly linked to child outcomesthrough its association with paternal involvement.For similar reasons, paternal civic engagementmay also exert indirect positive effects on childrenthrough its association with increased paternal in-volvement. Thus, future research should examinepossible links between child well-being and reli-gion, as well as child well-being and paternal civicengagement.

CONCLUSION

This study’s findings must be placed in perspec-tive. The effects of religious culture and partici-pation are relatively small, especially in compar-ison to demographic and life course factors suchas the age, gender, and number of children. More-over, this study may overestimate the religious ef-fects on paternal involvement if religious menhave inflated their estimates of paternal involve-ment because of the normative messages they en-counter in their churches.

Nevertheless, this study suggests that religiondoes play a role in shaping men’s commitments totheir children. This finding is particularly striking

because this study and previous studies have notdocumented a consistent effect on fatherhood in-volvement for gender ideology (see Marsiglio,1991), another potential source of cultural influ-ence over paternal involvement. In all likelihood,the institutional power that religious congrega-tions can bring to bear on their members is onereason why religion appears to be a more pow-erful predictor of paternal involvement than gen-der ideology. The irony is that religious institu-tions, generally taken to be carriers of moretraditional mores, seem to be showing some suc-cess in closing the gap between new fatherhoodculture and conduct observed by LaRossa (1988).In other words, at least in terms of the quantity oftheir involvement, conservative Protestant andCatholic residential fathers are more likely to em-body the new fatherhood style Furstenberg (1988)attributes to Good Dads.

NOTE

This article was supported in part by a grant from theLouisville Institute. I gratefully acknowledge helpfuladvice from Marcia Carlson, Paul DiMaggio, Christo-pher Ellison, Sara McLanahan, Mark Regnerus, LynnRobinson, Darren Sherkat, Brian Steensland, BruceWestern, and Robert Wuthnow.

REFERENCES

Alwin, D. F. (1986). Religion and parental child-rearingorientations: Evidence of a Catholic-Protestant con-vergence. American Journal of Sociology, 92, 412–440.

Amato, P. R. (1998). More than money? Men’s contri-butions to their children’s lives. In A. Booth & A. C.Crouter (Eds.), Men in families: When do they getinvolved? What difference does it make? (pp. 241–278). Mahwah, NJ: Erlbaum.

Ammerman, N. T. (1997). Golden rule Christianity:Lived religion in the American mainstream. In D.Hall (Ed.), Lived religion in America: Toward a his-tory of practice (pp. 196–216). Princeton, NJ: Prince-ton University Press.

Bartkowski, J. P., & Xu, X. (2000). Distant patriarchsor expressive dads? The discourse and practice of fa-thering in conservative Protestant families. The So-ciological Quarterly, 41, 465–485.

Booth, A. & Crouter, A. C. (1998). Men in families:When do they get involved? What difference does itmake? Mahwah, NJ: Erlbaum.

Bourdieu, P. (1977). Outline of a theory of practice.New York: Cambridge University Press.

Browning, D. S. (1995). Religion and family ethics: Anew strategy for the church. In N. T. Ammerman &W. C. Roof (Eds.), Work, family, and religion in con-temporary society (pp. 157–176). New York: Rou-tledge.

Bumpass, L. L., & Sweet J. A. (1995). Cohabitation,marriage, and union stability: Preliminary findings


from the NSFH2 (NSFH Working Paper No. 65).Madison, WI: Center for Demography and Ecology,University of Wisconsin—Madison.

Christiano, K. J. (2000). Religion and the family inmodern American culture. In S. K. Houseknecht & J.G. Pankhurst (Eds.), Family, religion, and socialchange in diverse societies (pp. 43–78). Oxford, UK:Oxford University Press.

Clydesdale, T. T. (1997). Family behaviors among earlyU.S. baby boomers: Exploring the effects of religionand income change, 1965–1982. Social Forces, 76,605–635.

Coltrane, S. (1996). Family man. New York: OxfordUniversity Press.

Cooksey, E. C., & Fondell M. M. (1996). Spending timewith his kids: Effects of family structure on fathers’and children’s lives. Journal of Marriage and theFamily, 58, 693–707.

Doherty, W. J., Kouneski E. F., & Erickson M. F. (1998).Responsible fathering: An overview and conceptualframework. Journal of Marriage and the Family, 60,277–292.

Dollahite, D. C., & Hawkins A. J. (1998). A conceptualethic of generative fathering. The Journal of Men’sStudies, 7, 109 –132.

Durkheim, E. (1951). Suicide. New York: Free Press.(Original work published 1897)

Ellison, C. G. (1994). Religion, the life stress paradigm,and the study of depression. In J. S. Levin (Ed.), Re-ligion in aging and health: Theoretical foundationsand methodological frontiers (pp. 78–121). NewburyPark, CA: Sage.

Ellison, C. G., Bartkowski, J. P., & Segal, M. L. (1996).Conservative Protestantism and the parental use ofcorporal punishment. Social Forces, 74, 1003–1029.

Furstenberg, F. F. (1988). Good dads-bad dads: Two fac-es of fatherhood. In A. Cherlin (Ed.), The changingAmerican family and public policy (pp. 193–218).Washington, DC: Urban Institute Press.

Greeley, A. M. (1990). The Catholic myth: The behavior

and beliefs of American Catholics. New York: Scrib-ner.

Harris, K. M., & Morgan, S. P. (1991). Fathers, sons,and daughters: Differential paternal involvement inparenting. Journal of Marriage and the Family, 53,531–544.

Lamb, M. E. (1997). The role of the father in childdevelopment. New York: Wiley.

LaRossa, R. (1988). Fatherhood and social change.Family Relations, 37, 451–457.

Marler, P. L. (1995). Lost in the fifties: The changingfamily and the nostalgic church. In N. T. Ammerman& W. C. Roof (Eds.), Work, Family, and Religion inContemporary Society (pp. 23–60). New York: Rou-tledge.

Marsiglio, W. (1991). Paternal engagement activitieswith minor children. Journal of Marriage and theFamily, 53, 973–986.

Nock, S. L. (1998). Marriage in men’s lives. New York:Oxford University Press.

Pearce, L. D., & Axinn, W. G. (1998). The impact offamily religious life on the quality of mother-childrelations. American Sociological Review, 63, 810–828.

Pleck, J. H. (1997). Paternal involvement: Levels,sources, and consequences. In M. E. Lamb (Ed.), Therole of the father in child development. New York:Wiley.

Putnam, R. (2000). Bowling alone: The collapse andrevival of American community. New York: Simonand Schuster.

Stolzenberg, R. M., Blair-Loy, M., & Waite, L. J.(1995). Religious participation in early adulthood:Age and family life cycle effects on church member-ship. American Sociological Review, 60, 84–103.

Sweet, J. A., Bumpass, L. L., & Call, V. (1988). Thedesign and content of the National Survey of Familiesand Households (NSFH Working Paper No. 1). Mad-ison, WI: Center for Demography and Ecology, Uni-versity of Wisconsin-Madison.

Wilcox, W. B. (1998). Conservative Protestant child-rearing: Authoritarian or authoritative? American So-ciological Review, 63, 796–809.

Documents

Religion, Convention, and Paternal Involvement