socquart religion

20
The Sociological Quarterly 47 (2006) 175–194 © 2006 Midwest Sociological Society 175 The Sociological Quarterly ISSN 0038-0253 Blackwell Publishing Ltd.Oxford, UK and Malden, USATSQThe Sociological Quarterly0038-02532006 Blackwell Publishing Ltd.2006471175194ADOLESCENCE, FAMILY STRUCTURE, RACE, AND RELIGION Religious Change and Adoles- cent Family DynamicsMark D. Regnerus and Amy Burdette Opinions reflect those of the author and do not necessarily reflect those of the granting agency. *Direct correspondence to Mark D. Regnerus, Department of Sociology, University of Texas at Austin, 1 University Station A1700, Austin, TX 78712-0118; e-mail: [email protected] RELIGIOUS CHANGE AND ADOLESCENT FAMILY DYNAMICS Mark D. Regnerus* University of Texas at Austin Amy Burdette University of Texas at Austin Much more has been documented about the influence of religion on parenting practices than on how the former may shape family life from the perspective of adolescents. Building a conceptual model of religion and changing family relations, we assessed the particular influence of adolescent religious change on the dynamics of their relationships with their parents, and overall satisfaction with their families. Employing data from two waves of the National Longitudinal Study of Adoles- cent Health (Add Health), results suggested that growth in personal religious salience—how important religion is in adolescents’ lives—is uniquely and consistently related to better family rela- tions, even after accounting for behavioral changes—such as excessive drinking and drug abuse— that are detrimental to both religiosity and family relations. Positive family functioning and good parent–child relationships during adolescence elicit ideal outcomes well into adulthood, including the psychological well-being and life satis- faction of both parents and children (Amato 1994; Knoester 2003). Children who feel emotionally close to their parents are also more apt to exhibit “collectivist value orienta- tions,” in contrast to more individualist orientations (Bengtson, Biblarz, and Roberts 2002:28). Research on families and religion have often noted links between religious involvement, particular religious affiliations, and a variety of family outcomes, including marital duration and satisfaction, authoritative (and in some cases, authoritarian) parenting practices, and parenting consistency (Thornton 1985; Bahr and Chadwick 1988; Brody et al. 1994; Wilcox 1998). In general, the tenor of religious effects on families is typically positive. Yet there has been considerably more research interest in the influ- ence of religion on child rearing and parenting practices (e.g., Ellison and Sherkat 1993a,b; Wilcox 1998) than in looking at how religion may shape family life from the per- spective of the child. Indeed, the perspective that is typically featured in such studies is nearly exclusively from that of parents. Employing data from two waves of the National Longitudinal Study of Adolescent Health (Add Health), we avoid this common focus on parents and instead evaluate how adolescents perceive the quality of their interpersonal relationships with family members. Given that adolescence is the peak period of religious instability across the life course, we

Upload: cinderella1977

Post on 30-Sep-2015

5 views

Category:

Documents


3 download

DESCRIPTION

paper on religion

TRANSCRIPT

  • The Sociological Quarterly

    47

    (2006) 175194 2006 Midwest Sociological Society

    175

    The Sociological Quarterly ISSN 0038-0253

    Blackwell Publishing Ltd.Oxford, UK and Malden, USATSQThe Sociological Quarterly0038-02532006 Blackwell Publishing Ltd.2006471175194ADOLESCENCE, FAMILY STRUCTURE, RACE, AND RELIGION

    Religious Change and Adoles-

    cent Family DynamicsMark D. Regnerus and Amy Burdette

    Opinions reflect those of the author and do not necessarily reflect those of the granting agency.

    *Direct correspondence to Mark D. Regnerus, Department of Sociology, University of Texas at Austin, 1

    University Station A1700, Austin, TX 78712-0118; e-mail: [email protected]

    RELIGIOUS CHANGE AND ADOLESCENT FAMILY DYNAMICS

    Mark D. Regnerus*

    University of Texas at Austin

    Amy Burdette

    University of Texas at Austin

    Much more has been documented about the influence of religion on parenting practices than on

    how the former may shape family life from the perspective of adolescents. Building a conceptual

    model of religion and changing family relations, we assessed the particular influence of adolescent

    religious change on the dynamics of their relationships with their parents, and overall satisfaction

    with their families. Employing data from two waves of the National Longitudinal Study of Adoles-

    cent Health (Add Health), results suggested that growth in personal religious saliencehow

    important religion is in adolescents livesis uniquely and consistently related to better family rela-

    tions, even after accounting for behavioral changessuch as excessive drinking and drug abuse

    that are detrimental to both religiosity and family relations.

    Positive family functioning and good parentchild relationships during adolescence elicitideal outcomes well into adulthood, including the psychological well-being and life satis-faction of both parents and children (Amato 1994; Knoester 2003). Children who feelemotionally close to their parents are also more apt to exhibit collectivist value orienta-tions, in contrast to more individualist orientations (Bengtson, Biblarz, and Roberts2002:28). Research on families and religion have often noted links between religiousinvolvement, particular religious affiliations, and a variety of family outcomes, includingmarital duration and satisfaction, authoritative (and in some cases, authoritarian)parenting practices, and parenting consistency (Thornton 1985; Bahr and Chadwick1988; Brody et al. 1994; Wilcox 1998). In general, the tenor of religious effects on familiesis typically positive. Yet there has been considerably more research interest in the influ-ence of religion on child rearing and parenting practices (e.g., Ellison and Sherkat1993a,b; Wilcox 1998) than in looking at how religion may shape family life from the per-spective of the child. Indeed, the perspective that is typically featured in such studies isnearly exclusively from that of parents.

    Employing data from two waves of the National Longitudinal Study of AdolescentHealth (Add Health), we avoid this common focus on parents and instead evaluate howadolescents perceive the quality of their interpersonal relationships with family members.Given that adolescence is the peak period of religious instability across the life course, we

  • 176

    The Sociological Quarterly

    47

    (2006) 175194 2006 Midwest Sociological Society

    Religious Change and Adolescent Family Dynamics

    Mark D. Regnerus and Amy Burdette

    are particularly interested on how change in adolescent childrens religiosity affects howthey view or experience their parents and family. Before we examine these relationships,we highlight several key recent studies in the field of religion and family before outlininga few conceptual frameworks by which adolescent religious change on parentchild rela-tionships might be understood.

    ADOLESCENT RELIGIOUS DEVELOPMENT

    Survey research on American adolescents continues to note a developmental componentto religious involvement. Each additional year of adolescence brings with it significantphysical, psychological, and social changes that are thought to influence their religiousparticipation (Smith et al. 2002: 597). For example, the frequency of regular religious ser-vice attendance tends to decline between 8th and 12th grades. In 1997, approximately 44percent of 8th graders reported attending services weekly compared to 38 percent of 10thgraders, and 31 percent of 12th graders (Johnston, Bachman, and OMalley 1999). Datafrom the Survey of Parents and Youth note that while 50 percent of 13-year-olds reportparticipating in a religious youth group in the previous seven days, this figure declines inlinear fashion, bottoming out at 29 percent for 18-year-olds, constituting a 43 percentdecline over five years (Smith et al. 2002). Age differences in religious salience are notedmuch less commonly. A study of adolescents in Iowa found that while the frequency ofreligious service attendance dipped during the high school years, levels of participation inother religious activities tended to rise over the same period (King, Elder, and Whitbeck1997). Yet, adolescence and emerging adulthood is also the life stage when religious con-version is most likely to take place. Toward this end, Ozorak (1989) suggests that polariza-tion in religiosity occurs during adolescence; the decreases (in religiosity) of somewhat ormoderately religious youth mask the increases of the (fewer) very religious.

    RELIGION AND FAMILY RELATIONS

    How religious involvement and tradition have shaped family formation, relations, andparenting practices has been documented in a variety of studies (for a summary, seeMahoney et al. 2001). The messages, doctrines, social networks, and family activity set-tings that JudeoChristian religious institutions provide are powerful forces for encour-aging and reinforcing family solidarity, role definitions, and childrens obedience (Ellisonand Sherkat 1993b; Curtis and Ellison 2002). By way of enhancing public and private reli-giosity, family members may cultivate values emphasizing love, care, and forgiveness(Curtis and Ellison 2002). At the same time, however, the evolution of accepted familyforms (e.g., the prevalence of stepfamilies, growing religious heterogeneity within fami-lies, singlehood, childlessness, etc.) has also served to modifypractically if not theolog-icallyhow congregations, denominations, and individuals understand family relations(Thornton 1985).

    Yet, whether and how religion actually affects interpersonal relationships betweenfamily members remain poorly understood (Holden 2001). In this study, we consider this

  • Mark D. Regnerus and Amy Burdette

    Religious Change and Adolescent Family Dynamics

    The Sociological Quarterly

    47

    (2006) 175194 2006 Midwest Sociological Society

    177

    question, and in particular focus our attention on adolescents affective relationshipswith parents and familythat is, their feelings of closeness, being understood, having funtogether, and being satisfied with parental and family relationships (Bengtson et al.2002).

    In one of the most comprehensive and long-term studies on the subject, Pearce andAxinn (1998) found evidence connecting maternal religiosity with young adult childrensreports of higher-quality motherchild relationships. In particular, the importance ofreligion in the mothers life was more tightly linked with motherchild closeness than waschurch attendance or affiliation. Additionally, when the mother and child matched in theimportance that each accorded religion, both maternal and child reports of relationshipquality were more positive. When 18-year-olds attended religious services at levels com-parable to their mothers, the latter tended to report significantly better relationship qual-ity a full five years later. In studies of observed family interactions, greater generalreligiousness of family members appeared to be linked to more cohesiveness (Brody,Stoneman, and Flor 1996) and self-reported cohesion (Abbott, Berry, and Meredith1990) within families.

    In spite of scholarly and lay interest in the effect of religious involvement on familylifeparticularly with respect to parenting practicesprevious research have been lim-ited in at least three key respects. First, few studies have incorporated adolescent perspec-tives on family relations. Researchers have typically relied on the perceptions of parents inassessing the quality and stability of family life and parentchild relationships. This lackof method variance is noteworthy in light of research, which suggests that parents andadolescents often exhibit varying perspectives on conflict within the family (Collins 1990;Aquilino 1999). Second, previous research relies heavily on cross-sectional data. As a con-sequence, scholars are unable to explore the changes in family relationships over time.Third, developmental concerns about adolescent religious change or transformation(i.e., becoming more religious or losing religion altogether) have not been given ampleattention, orwhere they have been notedare not evaluated adequately because ofdata limitations or the narrow focus of particular research interests. Although our datawill not document the exact manner in which religious change influences parentchildand family relations, there are several possible pathways of influence. We briefly outlinethree in particular, namely: (1) increased opportunities for parentchild interaction, (2)immersion into family-centered belief systems, and, alternately, (3) religious exitprocesses.

    ParentChild Interaction

    Through increased participation in religious services and activities, parents andchildren spend time together in a family-affirming environment. Churches and otherreligious organizations often plan activities (e.g., camps, retreats, fatherdaughterbreakfasts) to promote positive family relationships (Abbott et al. 1990; Pearce andAxinn 1998). Further increasing opportunities for interaction (again, conceived to be apositive thing) are the above-average number of social ties and more cohesive networksamong persons and generations within congregationsan example of intergenerational

  • 178

    The Sociological Quarterly

    47

    (2006) 175194 2006 Midwest Sociological Society

    Religious Change and Adolescent Family Dynamics

    Mark D. Regnerus and Amy Burdette

    closure that is thought to encourage stronger parentchild relationships (Coleman1988; Smith 2003).

    Family-Centered Belief Systems

    Religious institutions not only foster social support and provide opportunities for familyinteraction, but also encourage parents to be highly committed to both their children andtheir spouse. More subtly, children are urged to honor and obey their parents (Ellison andSherkat 1993b). In this way, among others, family members can learn what is expected ofthem as well as what to expect from each other. Such specific moral directives concerningfamily relations can constitute a key form of religious influence among youth (Smith2003). Because religion can sanctify (i.e., lend religious importance to) family life, ittypically offers purposes and processes that have no direct equivalent within secular sys-tems of meaning and motivation. This sanctification presents family members not onlywith new sources of conflict, but also new resources for resolving conflict (Mahoney2005). For example, apart from fee-based marital and family counseling services, thereare fewer resources for irreligious Americans to draw upon for family relations helpthan for religious Americans (e.g., courses on parenting, occasional free child care, con-gregational interaction opportunities, social control of deviant behavior of both parentsand youth, etc.). Thus, increasing religiosity may prompt more consistent agreementbetween parents and children about the boundaries of acceptable childand for thatmatter, parent and familybehavior (Bartkowski and Ellison 1995). The opposite maybe true when religiosity diminishes. On the other hand, when incompatible or dissimilarreligious perspectives exist within families, religious change could exacerbate conflicts.

    The Religious Exit Explanation

    Unlike the two pathways previously noted, which focused on positive religious change,another plausible account about the association between religion and adolescent familyrelations concerns those who leave organized religion. This approach emphasizes thepossibility that some youth select themselves out of religious involvement, suggesting thatreverse causation is at work in apparent religious effects. From this perspective, religiondoes not influence family relations, despite initial appearances. It is thought that someadolescents, for whatever reasons, develop certain attitudes and behaviors (e.g., becomedepressed, lose their virginity, commit crimes, drink excessively, etc.) that are at odds withtheir religious beliefs and practices, and thus subsequently reduce their religious involve-ment, alter their religious beliefs, or drop out of organized religion altogether. This couldfunction to create observed (but not real) associations between religion or religiosity andparticular outcomes, since the religious sample comprises those that did not drop out ordecline in religiosity. Inverse associations between religiosity and deviant outcomes (e.g.,poor family relations) may be because the respondents with the negative outcomesreported diminishing religiosity. In this way, religion itself does not influence peopletoward better outcomes, but rather is avoided by people with worse outcomes. In general,there is considerable evidence across a variety of outcomes to suggest that certain behav-iors (e.g., delinquency, cohabitation, sex, etc.) may also result in diminished religiosity or

  • Mark D. Regnerus and Amy Burdette

    Religious Change and Adolescent Family Dynamics

    The Sociological Quarterly

    47

    (2006) 175194 2006 Midwest Sociological Society

    179

    extent of religious involvement of the actors (Thornton and Camburn 1989; Thornton,Axinn, and Hill 1992; Benda and Corwyn 1997; Benda and Toombs 2002).

    HYPOTHESES

    Social scientists often refer to religious effects or influences of religiosity, but religiosityis a complex, multidimensional concept with private and public aspects that defy simpleexplanation and that can yield very different results in studies of adolescent outcomes(Regnerus 2003). For our purposes, then, different forms of religious change may lendthemselves more or less to directly affecting adolescents perception of family relations.Increasing attendance or importance of religion in adolescent childrens lives will likelycontribute to a number of processes related to better family relations. With respect tochurch or religious service attendance, their opportunities for focused interaction withparents and exposure to messages concerning ideal parentchild relations and responsi-bilities will increase. Increased interaction between family members may not result fromincreasing religious salience, but the moral directives that characterize the family unit ascontaining religious significanceand that parents are to be obeyed and familiesenjoyedmay be more readily ascertained and adhered to as religion becomes moreimportant in adolescents lives.

    The research conclusions and theoretical pathways previously noted collectively sug-gest the outlines of a model (Figure 1) of the dynamics between developing religiosity andchanging family relations. In particular, some religious changessuch as the upheavaland potential religious discord that accompanies shifting or differing affiliations (Sherkat

    FIGURE 1.

    A Conceptual Model of Religion and Changing Family Relations.

    Heightened substance useand delinquency

    Shift in religious affiliation (Religious instability, discontinuity

    in exposure to family-centered messages and opportunities)

    +

    Change infamily relations

    Baseline level and growthin religious resources

    (Access to parentchild interaction opportunities, exposure to family-centered

    moral directives, intensity ofcommitment to family)

  • 180

    The Sociological Quarterly

    47

    (2006) 175194 2006 Midwest Sociological Society

    Religious Change and Adolescent Family Dynamics

    Mark D. Regnerus and Amy Burdette

    and Darnell 1999; Myers 2004)are thought to be aggravating. Studies of married cou-ples have detected more significant conflict and disruption in marriages involving one (ora more) conservative Protestant partner (Ellison, Bartkowski, and Anderson 1999; Curtisand Ellison 2002). Discordant attendance among spouses appears detrimental to maritalhappiness and stability (Heaton and Pratt, 1990). Thus, the lack of religious congruencemay correspond with a variety of poor relationship outcomes. Additionally, parents andchildren who draw upon discordant religious foundations may be more apt to displayrelationship problems.

    Other religious changes are thought to benefit family relations, including the height-ening of religiosity and its concomitant exposure to religious messages about family rela-tions, opportunities to positively interact with family members, familiarity with otherfamily role models, and so on. The evidence concerning religious exits, however, suggeststhat parallel (and popular) processes during adolescence that heighten youth delin-quency and substance use may both directly diminish the quality of family relations andmitigate religious effects by undermining religious development.

    Nevertheless, Wimberleys (1989) work on a hierarchy of salience among personsvarious roles (e.g., religious believer, son, employee, etc.) suggests that when ones rolescall for incompatible actions (e.g., to save money or to send children to private religiousschooling, etc.), a hierarchy of salience influences the choice of action. That is, the discreteidentity that is most important, or more important relative to the other role identities,will motivate action. Thus, religious salience is thought to directly shape religious normadherence, perhaps acting as a stimulant in obeying religious moral directives concerningthe family. This emphasis on role importance, together with previous research on familyrelations, suggests that subjective religiosity (i.e., personal religious salience) as well aspublic religiosity (i.e., attendance) will be associated with family relationship outcomes(Pearce and Axinn 1998; Myers 2004):

    Hypothesis 1: Family-related ideals or norms (e.g., adolescents adhering to parentalrules or reflecting positively upon their parents) will be more frequently adhered towhen religion becomes more important in adolescents lives or when they increasetheir participation in organized religion.However, not all religious changes improve family relations. When adolescents draw

    upon a common religious base, regardless of its significance in their lives (i.e., whethervery important or not), this lends itself to relationship stability, if not better relations.When religious conversion or reaffiliation to a different religious perspective occurs, con-flicts can be exacerbated (Mahoney 2005):

    Hypothesis 2: When adolescents change religious traditions (e.g., from Catholic toevangelical Protestant), their subsequent report of family relations will be poorer.Moreover, to the extent that born again experiences also entail changes in religioustraditions, these too will contribute to poorer family relations.Finally, given the evidence that certain deviant attitudes and behaviors (e.g., height-

    ened alcohol use, delinquency, etc.) tend to both aggravate family relations and reducereligiosity, we should see the influence of religious change on family relations shrinkwhen we account for such deviant attitudes or behaviors (Thornton 1985):

  • Mark D. Regnerus and Amy Burdette

    Religious Change and Adolescent Family Dynamics

    The Sociological Quarterly

    47

    (2006) 175194 2006 Midwest Sociological Society

    181

    Hypothesis 3: After accounting for change in behaviors that aggravate both familyrelations and religiosity, the positive influence of heightened attendance or religioussalience will diminish or disappear.

    DATA AND METHODS

    Data

    The data for this analysis come from the first two waves of the Add Health, a representa-tive study of adolescents in grades 7 to 12. Add Health is a school-based study of health-related behaviors. It is designed to explore the causes of these behaviors, with an emphasison social context and social networks. The Add Health study was funded by the NationalInstitute of Child Health and Human Development (NICHD) and 17 other federal agen-cies. Fieldwork was conducted by the National Opinion Research Center of the Universityof Chicago. Data were gathered from adolescents, their parents, siblings, friends, roman-tic partners, fellow students, and school administrators. The schools ranged in size fromless than 100 students to more than 3,000 students, and included urban, suburban, andrural locations. The schools, and the students in them, are thus representative samples.

    Respondents were interviewed in their homes twice, with an approximate one-yearinterval between interviews. The first wave of in-home interviews was conductedbetween April and December 1995. A parent was also interviewed once at the first wave.The second wave of data collection took place between April and August 1996. Prior tolist-wise deletion of missing values, the working sample size of respondents for whomthere is Waves I and II data, as well as a parent interview and survey weights, is 13,303 ado-lescents. Given its multiple waves and large sample size, Add Health is arguably the bestcurrent data source for examining religious effects on adolescent outcomes, especially forscholars concerned with change in behavior over time, and explanatory rather than asso-ciational relationships.

    Measures

    MotherChild and FatherChild Relationship Quality

    . These two dependent variables areidentical, summed indices of five measures, and were collected at both study waves. Theindices were intended to reflect comparable measures employed in a study by Ellis,Thomas, and Rollins (1976). Adolescent respondents were first queried about the pres-ence of a residential mother and father in their households. These were typically a biol-ogical parent, although they were not required to be. Of the baseline Wave II surveyrespondents, 1,071 reported having no residential mother or mother-like figure, and justfewer than 4,300 reported having no residential father or father-like figure. Those respon-dents who reported having one or the other (or both) were asked several questions in twodifferent sections about them. The first was, How close do you feel to your mother (orresidential mothers name)? The respondents could respond on a 5-point range from 1(not close at all) to 5 (extremely close). Next, they were asked, How much do youthink she (or residential mothers name) cares for you? Answers could range from 1(not at all) to 5 (very much). Identical questions were then asked about the

  • 182

    The Sociological Quarterly

    47

    (2006) 175194 2006 Midwest Sociological Society

    Religious Change and Adolescent Family Dynamics

    Mark D. Regnerus and Amy Burdette

    respondents residential father. Later in the survey, the respondents were presented with aseries of statements about their relationships with their parents: Most of the time, yourfather is warm and loving toward you, You are satisfied with the way you and your fathercommunicate with each other, and Overall, you are satisfied with your relationship withyour father. The respondents were also asked to evaluate the same statements about theirmother, and could respond to each with an answer ranging from 1 (strongly agree) to 5(strongly disagree). The index is comparable to the solidarity models measure of affec-tual solidarity (Silverstein and Bengtson 1997). The alpha coefficients of reliability for theset of five measures, split according to sex of parent and study wave, ranged from 0.83 to0.87. All change scores are simply computed by subtracting the Wave I score from that ofWave II. Table 1 presents the summary statistics for all variables.

    Family Satisfaction

    . This outcome is constructed as a summed index of three indicators ofthe adolescent respondents satisfaction with their family lifea topic broached consid-erably later in the survey than the parental relationship questions. The respondents wereasked, How much do you feel that people in your family understand you? Questionswere also asked concerning how much the respondent felt their family has fun togetherand how much they felt their family pays attention to them. The answer categories foreach ranged from 1 (not at all) to 5 (very much). The alpha coefficients of reliabilityfor the set of three variables were 0.78 and 0.79 at Waves I and II, respectively.

    Religion

    . We employ four adolescent religion measures in this study: (1) religious affilia-tion, (2) importance of religion, (3) attendance, and (4) self-identity as a born againChristian. The adolescents were asked about their religious affiliation during both waves.We grouped religious affiliations into eight categories in step with Steensland et al.s(2000) classification scheme, including evangelical, mainline, and black Protestants,Roman Catholics, Jews, Latter-Day Saints/Mormons, the religiously unaffiliated, and acategory of other religious groups. The change-in-affiliation measure taps whetheradolescents reported a religious affiliation at Wave II that was outside their original reli-gious tradition (among the eight listed previously). Additionally, we include a dichoto-mous measure tapping whether the adolescent and their parent cite the same or differentreligious traditions (e.g., mainline Protestant versus Catholic, conservative Protestantversus Mormon).

    We also include a dichotomous measure of whether the adolescent considers himself/herself to be a born again Christian, and whether this status changed between waves(i.e., whether the respondent said no and yes at Waves I and II, respectively). Perhypothesis, we distinguish between born again religious experiences that occur withina stable religious tradition and those that entail a new religious affiliation or tradition.

    Religious salience taps a more private and subjective form of religiosity in the respon-dents life. At both waves, the adolescents were asked, How important is religion toyou? Four response categories ranged from not important at all to very important.Unfortunately, those who indicated that they were not religious in the first religion sec-tion question were not asked to answer any subsequent religion questions. Rather than

  • Mark D. Regnerus and Amy Burdette

    Religious Change and Adolescent Family Dynamics

    The Sociological Quarterly

    47

    (2006) 175194 2006 Midwest Sociological Society

    183

    TABLE 1.

    Means, Ranges, and Standard Deviations (SDs) of Measures

    Variables Range

    Wave I

    mean

    Wave I

    SD

    Wave II

    mean

    Wave II

    SD

    Motherchild relationship quality

    (N

    =

    10,574)

    525 22.24 3.07 21.86 3.16

    Change in motherchild relationship quality

    (N

    =

    10,071)

    18 to

    +

    17

    0.38 2.79

    Fatherchild relationship quality (N

    =

    7,664) 525 21.37 3.68 20.87 3.71

    Change in fatherchild relationship quality

    (N

    =

    7,341)

    16 to

    +

    18

    0.48 3.04

    Family satisfaction index 315 11.29 2.43 11.24 2.51

    Change in family satisfaction

    12 to

    +

    12

    0.05 2.28

    Adolescents religious service attendance 14 2.77 1.20

    Change in service attendance habits

    between Waves I and II

    3 to

    +

    3

    0.05 0.96

    Adolescents self-rated importance of religion 14 3.07 1.04

    Change in importance of faith between

    Waves I and II

    3 to

    +

    3

    0.06 0.88

    Change in religious tradition 0,1 0.26 0.44

    Self-identifies as born again at Wave I 0,1 0.27 0.44

    Becomes born again between Waves I and II 0,1 0.08 0.27

    Religious tradition of parent and child are

    congruent

    0,1 0.73 0.45

    Level of personal autonomy 06 4.26 1.37

    Change in personal autonomy between

    Waves I and II

    6 to

    +

    6 0.35 1.43

    Female 0,1 0.51 0.50

    White 0,1 0.58 0.49

    Age 1219 15.24 1.57

    Average parent education 01 0.26 0.39

    Household size 115 3.62 1.57

    Stepfamily 0,1 0.14 0.35

    All other household structure types 0,1 0.07 0.25

    Single-parent household 0,1 0.24 0.43

    Biological two-parent family 0,1 0.55 0.50

    Likes taking risks 15 3.54 1.05

    Strategic 525 18.20 2.88

    Change in drinking habits between Waves

    I and II

    6 to

    +

    6 0.09 1.40

    Change in marijuana use between Waves I

    and II

    1 to

    +

    1 0.00 0.43

    Change in delinquency between Waves I

    and II

    18 to

    +

    18

    0.44 2.56

    Unless otherwise noted, Wave I variables N

    =

    11,094; measures of change between waves N

    =

    10,493.

  • 184

    The Sociological Quarterly

    47

    (2006) 175194 2006 Midwest Sociological Society

    Religious Change and Adolescent Family Dynamics

    Mark D. Regnerus and Amy Burdette

    lose a considerable number of cases, Add Health data users have typically assigned thelowest value on attendance and religious salience to youth who indicate no religiousaffiliation.

    Religious service attendance is a reliable and traditional measure of the public andcollective expression of religion, and captures involvement in an adultchild moral com-munity. The measure is ordinal (ranges 1 to 4), and was asked at both waves as follows: Inthe past 12 months, how often did you attend religious services? Response categoriesranged from never to once a week or more. While single items may not be the idealmeasures of public and private religiosity, their concurrent use within models is extensivewithin sociological research (Pearce and Axinn 1998; Nonnemaker, McNeely, and Blum2003; Myers 2004).

    Behavior/Status Change Measures

    . We include several behavioral change measures inorder to evaluate their effects on change in family relations. The measures consisted ofWave II comparisons with Wave I outcomes on frequency of alcohol use, the occurrenceof marijuana use, the frequency of a series of minor delinquent acts (e.g., damaging prop-erty, lying to parents, rowdiness, etc.), and change in the level of adolescents autonomy(a six-item summed index of the respondents freedom to decide what they wear, whothey hang around with, what time they go to bed on weeknights, etc.).

    Control Variables

    . We include several control variables that are thought to be related bothto religiosity and family relationship quality. They should function to limit spuriousassociations between religiosity and the three family relations outcomes. Socioeconomicbackground has in several studies affected parentchild relations, though not extensivelyso (Rossi and Rossi 1990; Pearce and Axinn 1998); thus we include a measure of averageparental education. Family structure has been linked with family relations as well as fam-ily religiosity (Bahr and Chadwick 1988; Booth and Amato 1994). We include two dichot-omous variables indicating whether the respondent is living in a stepfamily or in someother type of family (i.e., single parent, adopted, another type) compared to a biologi-cally intact, two-parent family. We also controlled for household size, age, and race(white

    =

    1). Parentchild value consensus and cohesiveness are expected to vary bygender, with sons closer to fathers and daughters to mothers (Pearce and Axinn 1998;Silverstein and Bengtson 1997). We include measures of gender for both the adolescentand parent respondent. Proclivity for risk is a single-item measure that consists of thelevel of agreement with the statement: You like to take risks. Respondents could rangefrom strongly agree to strongly disagree. It was not asked at Wave I; it is a Wave IImeasure that is treated as indicating an underlying personality type. Similarly, the vari-able dubbed strategic is a five-item summed index of how strategic a decision maker therespondent is. The measures include responses to questions such as When making deci-sions, you usually go with your gut feeling without thinking too much about the conse-quences of each alternative, and After carrying out a solution to a problem, you usuallytry to analyze what went right and what went wrong. The alpha coefficient of reliabilityfor this set of measures was 0.63.

  • Mark D. Regnerus and Amy Burdette

    Religious Change and Adolescent Family Dynamics

    The Sociological Quarterly

    47

    (2006) 175194 2006 Midwest Sociological Society

    185

    Modeling Religious Influences on Family Relationships

    Our modeling approach seeks to isolate religious influences on the family from othereffects (Holden 2001). We employ two types of models that should enhance the ability toassess causal influences of religion: (1) lagged dependent variable (LDV) models to assessthe influence of religiosity on change in the three family relations outcomes, while alsocontrolling for unmeasured heterogeneity in the outcome; and (2) pure change models,predicting interwave change in the outcomes as a function of interwave change in religi-osity and in a set of behaviors. By its nature, the LDV model is conservative; that is, onlycoefficients that significantly shape change over the course of the year between waves willdisplay significance. The LDV model is most appropriate when attitudes or behaviors aresubject to reinforcement or habit formation, as is very likely the case in family relations(Allison 1990).

    Alternatively, the pure change model is equally, if not more, conservative. Neithertime-invariant effects (e.g., race, gender, etc.) nor variables that are equally shared amongrespondents (i.e., change in age, etc.) should be included. Although change scores havetheir detractors, their confirmation as valid and reliable measures of individual growthappears to be increasing (Rogosa 1995). The change variables employed here are simpledifference scores: the Wave I variable value subtracted from the Wave II variable value.Together, these two methods should be understood as complementary; the LDV modelsestimate a baseline effect of religiosity on shifts in family relations, while the pure changemodels estimate the influence of religious change on discrete, measurable shifts in familyrelations.

    The estimates from each regression model included a school and region cluster term,as well as a weight to account for unequal probability of respondents being sampled. Theschool cluster increases the standard error of coefficients to account for homogeneitywithin schools, the primary sampling unit for Add Health. To accomplish this, we gener-ated our ordinary least squares (OLS) regression models in Stata using its svyreg func-tion (StataCorp 2001). Since the motherchild and fatherchild relationship outcomes,but not the family satisfaction index, are limited to those respondents who report a resi-dential mother or father or comparable figures, the two entail a considerable risk of sam-ple selection bias. As a possible solution, we evaluated Heckman models, but they provedto be poor choices for the motherchild analysis, resulting in erratic estimates and modelconvergence problems. While the Heckman approach was more stable for the fatherchild models, the results did not differ notably from OLS models.

    RESULTS

    Table 2 displays results from LDV regression models (in OLS) predicting adolescentsself-reports about their relationship with their mother, father, and overall family satisfac-tion, at Wave II. The models are predicting baseline effects (of the independent variables)on change in the outcomes. Adolescents who self-report that religion is important tothem indicated improved fatherchild relationship quality and overall family satisfac-tion. Specifically, a unit increase in baseline importance of religion corresponds with a

  • 186

    The Sociological Quarterly

    47

    (2006) 175194 2006 Midwest Sociological Society

    Religious Change and Adolescent Family Dynamics

    Mark D. Regnerus and Amy Burdette

    0.203-unit improvement in the quality of the respondents relationship with their fatherbetween survey waves, as well as a 0.095-unit improvement in overall family satisfaction.No religiosity effect appears associated with improvement in motherchild relations.However, a baseline effect of being born again alters only the motherchild relation-ship, improving it by just under 0.2 units. Given its dichotomous nature, this effect is lesspowerful than the religious salience effect on improvement in the fatherchild relation-ship. It should be apparent, then, that while the stock or level of religiosity may paybenefits more immediately (in cross-sectional associations with family relations, notevaluated here), it certainly pays benefits over time. Religious salience, however, appearsconsiderably more efficacious than attendance.

    Other variables seldom displayed significance in the LDV models. Those worth not-ing include aggravating effects of stepfamilies and single-parent households on both the

    TABLE 2.

    Ordinary Least Squares (OLS) Lagged Dependent Variable Regression Estimates of

    MotherChild, FatherChild, and Overall Family Relations, Wave II

    Effect

    Motherchild

    relationship

    Fatherchild

    relationship Family satisfaction

    Importance of religion 0.078 (0.05) 0.203** (0.06) 0.095** (0.03)

    Religious service attendance

    0.038 (0.04)

    0.106

    +

    (0.06)

    0.047 (0.03)

    Self-identifies as a born

    again Christian

    0.195* (0.08) 0.104 (0.12) 0.097 (0.07)

    Adolescents religious

    tradition same as parents

    0.071 (0.08) 0.117 (0.11) 0.064 (0.07)

    Female

    0.156* (0.07)

    0.354*** (0.08)

    0.045 (0.06)

    White

    0.088 (0.07) 0.050 (0.11)

    0.013 (0.07)

    Age 0.018 (0.02)

    0.016 (0.03) 0.025 (0.02)

    Average parental education

    0.079 (0.09) 0.146 (0.12) 0.072 (0.08)

    Number of persons in

    household

    0.005 (0.02)

    0.037 (0.04)

    0.008 (0.02)

    Stepfamily

    0.214* (0.09)

    0.248

    +

    (0.13)

    0.127 (0.08)

    Single-parent household

    0.169* (0.08)

    0.668** (0.25)

    0.097 (0.07)

    All other family structures

    (except bio-intact)

    0.009 (0.15) 0.475+ (0.26) 0.192 (0.12)

    Level of personal autonomy 0.020 (0.02) 0.041 (0.03) 0.033 (0.02)

    Likes taking risks 0.040 (0.03) 0.106* (0.04) 0.141*** (0.03)

    Strategic 0.022+ (0.01) 0.003 (0.02) 0.017 (0.01)

    Lagged dependent variable,

    from Wave I

    0.593*** (0.02) 0.634*** (0.01) 0.556*** (0.01)

    Intercept 8.249*** (0.54) 7.920*** (0.64) 4.782*** (0.43)

    Model fit statistics

    OLS R-square 0.347 0.420 0.319

    N 10,406 7,563 11,094

    Standard errors appear below the coefficient, in parentheses.+p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001 (two-tailed tests).

  • Mark D. Regnerus and Amy Burdette Religious Change and Adolescent Family Dynamics

    The Sociological Quarterly 47 (2006) 175194 2006 Midwest Sociological Society 187

    motherchild and fatherchild relationships, where significant decline in quality is notedbetween the two waves. Adolescent girls also displayed a declining relationship with boththeir mother and father between study waves, although the deterioration in their rela-tionship with their father was over twice the magnitude of that with their mother. Ourmeasure of proclivity for risk, included to help account for personality aspects of religios-ity, nevertheless displays an independent, aggravating effect. With each unit increase inproclivity for risk, fatherchild relationships and family satisfaction deteriorate by overone tenth of a unit.

    Table 3 displays the results from a pair of pure change models for each of the threeoutcomes. The first model evaluates only the effects of change in religious measures, netof change in adolescents level of autonomy or personal independence. The second modeladds a set of three change variableseach forms of delinquent behaviorthat comprisesa test of our third hypothesis, which suggests that it is not so much about change in reli-giosity but rather change in delinquent behavior that will alter both religiosity and familyrelations.

    While heightened autonomy does indeed correspond to improved family relationsacross all models, this does not account for the consistently strong and positive effects onfamily relations that are a function of heightened religious salience during adolescence.Indeed, for each unit increase in religious salience between study waves, we see a 0.15- to0.19-unit improvement in each measure of family relations. Only one other change-in-religion measure matters: youth who became born again within their original religioustradition display an improvement of nearly one half a unit in fatherchild relationshipquality, compared to youth who did not undergo such religious change. For the mostpart, however, the action remains (as in Table 2) with the self-rated importance of reli-gion in ones life.

    The models are clearly impacted by the addition of the change in delinquent behaviorvariables. New drug use consistently (and considerably) corresponds to declining qualityof family relations. Greater alcohol use is modestly related with poorer fatherchild rela-tions and overall family satisfaction, while more frequent minor delinquency is associ-ated with the declining quality of motherchild relations and family satisfaction. Each ofthese are clearly relationship aggravators. Their collective effects appear to outweigh thecontribution of religious change. Nevertheless, their effects do very little to mitigate theinfluence of heightened religious salience on improved family relations. That is, in con-trast to the third hypothesis, religious effects on family relations are not simply spurious,explicable by introducing troubling behavior that is thought to do harm to both religios-ity and family ties.

    To briefly revisit the hypotheses here, we found considerable support for the firsthypothesisthat personal religious salience, as in other studies of family relations, is aconsistent relationship builder. Church attendance, and even heightened attendance overtime, contributes little to this. We subjected this hypothesis to considerable rigor, usingboth LDV and pure change models, yet the consistency of religious salience effectsappeared stable. Additionally, when adolescents became born again within their samereligious tradition, we noted improved fatherchild relations. No support was found for

  • 188 The Sociological Quarterly 47 (2006) 175194 2006 Midwest Sociological Society

    Religious Change and Adolescent Family Dynamics Mark D. Regnerus and Amy Burdette

    TAB

    LE 3

    .E

    stim

    ates

    from

    Pu

    re C

    han

    ge M

    odel

    s: O

    rdin

    ary

    Leas

    t Squ

    ares

    (O

    LS)

    Reg

    ress

    ion

    Pre

    dict

    ing

    Ch

    ange

    in F

    amily

    Rel

    atio

    ns

    as a

    Fu

    nct

    ion

    of C

    han

    ge in

    Rel

    igio

    sity

    an

    d O

    ther

    Beh

    avio

    rs

    Eff

    ect

    Ch

    ange

    in m

    oth

    erc

    hild

    rela

    tion

    ship

    Ch

    ange

    in fa

    ther

    ch

    ild r

    elat

    ion

    ship

    Ch

    ange

    in fa

    mily

    sat

    isfa

    ctio

    n

    Mod

    el 1

    Mod

    el 2

    Mod

    el 1

    Mod

    el 2

    Mod

    el 1

    Mod

    el 2

    Ch

    ange

    in a

    tten

    dan

    ce-0

    .006

    (0.

    05)

    -0.0

    01 (

    0.05

    )0.

    052

    (0.0

    7)0.

    060

    (0.0

    7)0.

    041

    (0.0

    4)0.

    047

    (0.0

    4)

    Ch

    ange

    in im

    port

    ance

    of r

    elig

    ion

    0.18

    7***

    (0.

    05)

    0.17

    0***

    (0.

    05)

    0.17

    9* (

    0.07

    )0.

    157*

    (0.

    07)

    0.17

    4***

    (0.

    04)

    0.15

    6***

    (0.

    04)

    Bec

    ame

    bor

    n a

    gain

    , sa

    me

    trad

    itio

    n0.

    008

    (0.2

    0)0.

    001

    (0.2

    0)0.

    485*

    (0.

    23)

    0.47

    2* (

    0.23

    )0.

    019

    (0.1

    3)0.

    017

    (0.1

    4)

    Bec

    ame

    bor

    n a

    gain

    , di

    ffer

    ent

    trad

    itio

    n

    0.02

    2 (0

    .26)

    0.04

    5 (0

    .26)

    -0.0

    61 (

    0.27

    )-0

    .043

    (0.

    27)

    -0.0

    99 (

    0.19

    )-0

    .089

    (0.

    19)

    Ch

    ange

    in r

    elig

    iou

    s af

    filia

    tion

    0.00

    5 (0

    .10)

    -0.0

    01 (

    0.10

    )-0

    .054

    (0.

    12)

    -0.0

    62 (

    0.12

    )0.

    021

    (0.0

    6)0.

    016

    (0.0

    6)

    Ch

    ange

    in a

    uto

    nom

    y0.

    064*

    * (0

    .02)

    0.06

    8**

    (0.0

    2)0.

    072*

    (0.

    03)

    0.07

    7* (

    0.03

    )0.

    066*

    ** (

    0.02

    )0.

    071*

    ** (

    0.02

    )

    Ch

    ange

    in d

    rin

    kin

    g be

    hav

    ior

    -0.0

    58+ (

    0.03

    )-0

    .123

    ***

    (0.0

    4)-0

    .073

    ** (

    0.02

    )

    Ch

    ange

    in m

    ariju

    ana

    use

    -0.3

    70**

    * (0

    .08)

    -0.4

    61**

    * (0

    .13)

    -0.3

    38**

    * (0

    .08)

    Ch

    ange

    in m

    inor

    del

    inqu

    ency

    -0.0

    53**

    * (0

    .02)

    -0.0

    37+ (

    0.02

    )-0

    .060

    ***

    (0.0

    1)

    Inte

    rcep

    t-0

    .436

    ***

    (0.0

    5)-0

    .442

    ***

    (0.0

    5)-0

    .552

    ***

    (0.0

    7)-0

    .539

    ***

    (0.0

    7)-0

    .116

    * (0

    .04)

    -0.1

    25**

    (0.

    04)

    Mod

    el fi

    t sta

    tist

    ic

    OLS

    R-s

    quar

    e0.

    005

    0.01

    30.

    006

    0.01

    70.

    008

    0.02

    2

    N10

    ,071

    10,0

    717,

    341

    7,34

    110

    ,493

    10,4

    93

    + p