You are on page 1of 16

17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Paul R. Amato and Valarie King The Pennsylvania State University
Maggie L. Thorsen Montana State University∗

Parent–Child Relationships in Stepfather Families


and Adolescent Adjustment: A Latent Class
Analysis

In the current study the authors drew on Because of high rates of divorce, nonmarital
Waves I and III from Add Health to examine childbearing, cohabitation, and remarriage, an
the closeness of parent–adolescent relation- increasing number of children are growing up
ships in married mother–stepfather families apart from their biological fathers and living
(N = 1,934). They used latent class analysis to with stepfathers. The transition to stepfamily liv-
identify family constellations defined by ado- ing presents a number of risks for children, and
lescents’ relationships with all of their parents: children in stepfamilies exhibit more internaliz-
mothers, stepfathers, and biological nonresident ing and externalizing problems than do children
fathers. In particular, the authors (a) identi- in two-biological-parent households, on average
fied the most common underlying patterns of (Bray, 1999). And despite improvements in chil-
adolescent–parent relationships in stepfamilies; dren’s standard of living when custodial mothers
(b) determined the background characteristics remarry, children in stepfather families are no
that predict membership in these groups; and better off on most emotional and behavioral indi-
(c) examined how adolescents in these groups cators than are children in single-mother house-
fare with respect to depressive symptoms, delin- holds (Amato, 2010; Sweeney, 2010).
quency, and substance use. The results indicate A focus on average differences in children’s
that adolescents’ relationships can be repre- adjustment, however, obscures the hetero-
sented with 4 latent classes. Adolescents in these geneity in outcomes among children living in
classes differ on measures of adjustment, and stepfamilies (Coleman, Ganong, & Russell,
many of these differences persist into the early 2013). Why do some children in stepfamilies
adult years. thrive while others flounder? Although a variety
of factors contribute to children’s adjustment
this para is copied to statement of problem
in stepfamilies, almost all observers agree
that the role of parents is central (Bornstein,
Department of Sociology, 201 Oswald Tower, University 2002). Close and supportive relationships with
Park, PA 16802 (pxa6@psu.edu). parents foster children’s healthy development
∗ Department of Sociology & Anthropology, Montana State in all types of families, including stepfami-
University, 2-109 Wilson Hall, Bozeman, MT 59717. lies, yet establishing and maintaining strong
This article was edited by Robert Crosnoe. parent–child ties in stepfamilies is challenging,
Key Words: development/outcomes, National Longitudinal especially for adolescents (Bray & Easling,
Study of Adolescent Health, parent–adolescent relations, 2005; Hetherington & Clingempeel, 1992;
stepfamilies. Hetherington & Stanley-Hagan, 2000).
482 Journal of Marriage and Family 78 (April 2016): 482–497
DOI:10.1111/jomf.12267
17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Parent–Child Relationships in Adolescence 483

In this study we drew on Waves I and with their stepchildren and others remaining dis-
III from the National Longitudinal Study of engaged and emotionally distant (King, 2006).
Adolescent to Adult Health (Add Health; Despite the usefulness of this research, few stud-
http://www.cpc.unc.edu/projects/addhealth) to ies of stepfamilies have studied systems larger
examine the closeness of parent–adolescent than dyads. In one exceptional study, Baxter,
relationships in married mother–stepfather fam- Braithwaite, and Bryant (2006) examined triadic
ilies and the implications of these relationships relationships among college students living with
for adolescent adjustment. We focused on step- a biological parent and a stepparent. The most
father families because the number of children in common pattern to emerge from their qualitative
stepmother households is comparatively small analysis was one in which young adults related
(Stewart, 2007), and their representation in to the stepparent (to whom they were moderately
Add Health is too limited to conduct a detailed close) primarily through the resident biological
analysis. The current study also was limited to parent (to whom they were very close). Other
married stepfathers because adolescents in the patterns involved youth who were close to the
Add Health study who lived with their mothers resident biological parent but not the stepparent,
and cohabiting partners were not asked ques- youth who were not close to either parent, and
tions about their relationships with stepfather youth were very close to both parents, with the
figures. Stepfamilies that began as cohabiting last group being the least common.copied to theoretical
partnerships and transitioned into marriage Family systems theory provides folder
a general
prior to the Wave I interview, however, were framework for our research. This perspective
included in the sample. Despite some sample focuses on patterns of closeness and commu-
limitations, the Add Health data set is appropri- nication between family members, how these
ate for the current topic because it is large, is patterns are maintained over time, and the
nationally representative, and provides detailed implications of these patterns for individual and
information on parent–child relationships in family development (Broderick, 1993; Kerr &
stepfamilies. Moreover, Add Health makes Bowen, 1988; Minuchin, 1974). The assumption
it possible to study the associations between that all parts of a family system are interrelated
stepfamily relationships and multiple aspects of shifts the focus away from particular dyadic rela-
adolescent adjustment. tionships and toward more general patterns that
Previous stepfamily research has focused on characterize family relationships. In the present
children’s relationships with each parent sepa- study we considered adolescents’ relationships
rately (e.g., Buchanan, Maccoby, & Dornbusch, with three parental figures: (a) stepfathers, (b)
2000; King, 2006). In contrast, we identified mothers, and (c) nonresident biological fathers.
family constellations defined by adolescents’ Although most studies of stepfamilies have
relationships with mothers, stepfathers, and not incorporated information on nonresident
biological nonresident fathers—an approach fathers, bringing nonresident fathers into the
broadly consistent with family systems theory. picture makes it possible to study systems larger
In particular, we (a) identified the most com- than those defined by the household. Moreover,
mon underlying patterns of adolescent–parent children’s contact with nonresident fathers has
relationships in stepfamilies; (b) determined the increased in recent decades (Amato, Meyers, &
background characteristics that predict these Emery, 2009), and the quality of these relation-
patterns; and (c) examined how different patterns ships is related to multiple aspects of children’s
of relationships are associated with symptoms adjustment (Adamsons & Johnson, 2013; Amato
of depression, delinquency, and substance use & Gilbreth, 1999). For these reasons, including
in adolescence (using cross-sectional data) and nonresident fathers provides a more compre-
young adulthood (using longitudinal data). hensive picture of parent–child relationships
in stepfamilies and how these relationships are
related to adolescent adjustment. (Adolescents
Background in Add Health were not asked about nonres-
Many researchers have studied stepfam- idential stepmothers, so we were unable to
ily dyads, with a particular focus on the incorporate information on these relationships.)
stepfather–stepchild relationship. This research Family systems theory, like virtually all
has revealed a striking degree of variability, with family theories, assumes that parent–child rela-
some stepfathers developing close emotional ties tionships are central to children’s development
17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
484 Journal of Marriage and Family

and adjustment. Of course, as children grow and empirical indicators are available to deter-
into adolescence, parent–child conflict increases mine the optimal number. Moreover, whereas
and engagement in shared activities declines cluster analysis assigns individuals to clusters
(Smetana, Campione-Barr, & Metzger, 2006). absolutely, LCA calculates each individual’s
When these relationships remain emotionally probability of membership in each class. (Prob-
close, however, parents continue to be valuable abilities of less than 1 are assumed to be due to
resources for their adolescent children. Indeed, measurement error.) Studies based on LCA (or
a large research literature shows positive asso- latent transition analysis, its close cousin) have
ciations between the quality of parent–child appeared in the research literature on family
relationships and multiple aspects of adolescent relationships in recent years. For example, LCA
adjustment across a variety of family structures has been used to study patterns of interaction
(Buchanan, Maccoby, & Dornbusch, 2000; Gray between adult children and their parents in the
& Steinberg, 1999; Steinberg, 2001). Netherlands (van Gaalen & Dykstra, 2006),
Family transitions like divorce and remarriage the structure of intergenerational relations in
pose challenges for children’s relationships with rural China (Guo, Chi, & Silverstein, 2012),
parents. Parental divorce (or union disruption) and patterns of father–infant interaction in
tends to weaken children’s ties to nonresident two-parent families (Goodman, Crouter, Lanza,
fathers (Amato, 2010), and maternal repartner- Cox, & Vernon-Feagans, 2011). We know of no
ing often creates tension between children and studies that have used LCA to study stepfamily
mothers (Cavanagh, Schiller, & Riegle-Crumb, relationships.
2006; Day & Acock, 2004). Moreover, many Following systems theory, we assume that
children reject their stepfathers, especially when family relationships tend to “crystallize” into
remarriages occur during early adolescence particular configurations based on an underlying
(Hetherington & Jodl, 1994; Hetherington & logic. Because LCA is an exploratory rather than
Stanley-Hagan, 2000). Nevertheless, a great a confirmatory method, it is difficult to predict
deal of variability exists in adolescents’ ties what these configurations will be. One possi-
with parents in stepfamilies (King, 2006), and bility is that adolescents tend to provide similar
these relationships continue to be important ratings of closeness to mothers, stepfathers,
contexts for understanding adolescent adjust- and nonresident fathers. This would happen
ment (Hetherington, Henderson, & Reiss, 1999; if positivity (or negativity) in one relationship
Pryor & Rodgers, 2001). spills over and influences other relationships.
Correspondingly, well-adjusted and socially
skilled adolescents may build positive connec-
The Present Study tions with family members, whereas troubled
The current study extends prior research by adolescents may withdraw emotionally from
using latent class analysis (LCA) to study their families. With respect to the last possibility,
patterns of relationships in stepfamilies. LCA Dunn, Cheng, O’Connor, and Bridges (2004)
is a statistical method for identifying unob- found that children’s reports of conflict with
served subgroups within populations based mothers, stepfathers, and nonresident fathers
on observed indicators (Collins & Lanza, were positively correlated. They reasoned that
2010; McCutcheon, 1987). LCA is also a “easy” or “difficult” children tend to have sim-
person-centered rather than a variable-centered ilar relationships with all three parents because
approach. In a person-centered approach like they elicit similar responses in different people.
LCA, people are placed into groups on the If children are similarly close to everyone in
basis of the similarity of their responses (in the their families, then the latent classes would
present case, adolescents’ reports of closeness reflect adolescents with either close or distant
to parents). A person-centered approach shifts ties with parents, with perhaps an intermediate
attention away from the differences between group also emerging. With respect to adolescent
variables and toward the differences between rel- adjustment, a great deal of research suggests
atively homogeneous subgroups of adolescents. that adolescents have the most positive out-
LCA has several advantages over earlier comes when family relationships are close and
methods, such as cluster analysis. Unlike cluster supportive (Buchanan, Maccoby, & Dornbusch,
analysis, LCA does not require researchers 2000; Gray & Steinberg, 1999; Steinberg,
to specify the number of classes in advance, 2001). A straightforward hypothesis, therefore,
17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Parent–Child Relationships in Adolescence 485

is that adolescents report fewer symptoms of Brown, Manning, & Stykes, 2015), we included
depression, less delinquency, and less substance variables that reflect the presence of half- and
use when stepfamily relationships are generally stepsiblings in the household.
close and more symptoms of depression, more We then determined whether forms of ado-
delinquency, and more substance use when lescent adjustment at Wave I—symptoms
stepfamily relationships are generally distant. of depression, delinquency, and substance
Family systems theory allows for other rela- use—vary with latent class membership. Given
tionship configurations, however, given the the existence of feedback loops in family sys-
tendency of family members to form coalitions tems, we assumed that the links between youth
and alliances. Some adolescents may be close adjustment and patterns of stepfamily close-
to both resident parents but distant from non- ness are bidirectional; that is, that troubled
resident fathers—a pattern that may occur when parent–child relationships increase the risk of
nonresident fathers withdraw from their chil- problems such as delinquency and substance
dren’s lives or when resident parents establish use, and these problems, in turn, create further
strong boundaries around themselves and their tension in parent–child relationships. (For the-
children. Some adolescents may be close to oretical and empirical work on bidirectional
mothers and distant from stepfathers as well as and child effects, see Crouter & Booth, 2003,
biological fathers, particularly in cases where and Hawkins, Amato, & King, 2007). Although
mothers act as gatekeepers or form coalitions we could not test for bidirectionality with our
with their children. Yet other adolescents may data, we used the family and individual char-
be close to both biological parents and distant acteristics listed above as controls to ensure
from stepfathers, or close to nonresident fathers that the links between family configurations
but distant from mothers and stepfathers. Given and adolescent adjustment were not due to
the exploratory nature of the current study, we their mutual correlations with other variables.
cannot frame specific hypotheses about the We could not control for unobserved factors
number, nature, and frequency of these patterns. that may be producing spurious associations
Nevertheless, we hypothesized that adolescents between family configurations and adolescent
in mixed configurations experience moderate adjustment.
levels of adjustment: less positive than when all Finally, to determine whether the differences
relationships are close but more positive then in adjustment between latent classes are sta-
when all relationships are distant. ble over time, we examined class differences
The first step in the current analysis identified in adjustment at Wave III, approximately 6–7
latent classes and provided population-level years later, when respondents were in emerging
estimates of the proportion of adolescents in adulthood (Arnett, 2000). During the transition
each class. After determining the number and to adulthood youth have many developmental
size of the latent classes, we examined family tasks to accomplish. The process is not always
and individual characteristics that predict mem- smooth, however, and the prevalence of sev-
bership in these classes. We drew on prior theory eral types of risky behavior, including substance
and research on parent–child relationships in use, peaks during emerging adulthood, not ado-
stepfamilies for this purpose. These variables lescence (Arnett, 2000). But despite the fact
include adolescent gender, adolescent age, that establishing independence from parents is a
adolescent race, whether the adolescent was defining feature of adulthood, parents continue
U.S. born, adolescent religiosity, mother and to serve as crucial sources of support for youth
stepfather education, household income, years as they make this transition (Stewart, 2007). Our
in a stepfamily, and the number of siblings in the goal in this final analysis was to see whether
household (Hetherington & Clingempeel, 1992; relationships with parents in adolescence have
Pryor, 2014; Stewart, 2007; Sweeney, 2010). implications that persist beyond the teen years.
To capture aspects of family history, we also
included whether the adolescent was born in
marriage and the total number of father figures Method
to whom the adolescent had been exposed
Sample
since birth. Finally, because recent work on
family complexity suggests the importance We used data from Waves I and III of the
of looking at different types of siblings (e.g., Add Health study. When weighted, these data
17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
486 Journal of Marriage and Family

are nationally representative of adolescents in their biological fathers (21%) were not asked
Grades 7 through 12 in the United States during the question about closeness. Because our goal
the 1994–1995 school year (Harris et al., 2009). was to examine relationship patterns among
We drew on the subset of adolescents in Add all adolescents, it was necessary to include
Health who participated in the in-home inter- these cases in the analysis. Consequently, we
views at Wave I (N = 20,745). The analytic sam- included a binary indicator in the analysis
ple for the cross-sectional analysis was restricted (0 = does not know nonresident biological
to adolescents with valid sample weights who father, 1 = knows nonresident biological father)
were living with a biological mother and a and assigned a value of 1 (the lowest possible
stepfather at Wave I, excluding those whose value) on the closeness variable to adolescents
nonresident biological father was known to be who did not know their nonresident biological
deceased (n = 1,934). Daughters made up about father.
half (51%) of the sample, and the mean age was
15.4 years. The sample was mostly non-Hispanic Predictors of adolescent–parent relationship
White (70%), with 13% non-Hispanic Black, classes. We examined several individual- and
11% Hispanic, and 6% other. The typical ado- family-level characteristics that may predict
lescent had been in a stepfamily for 7.6 years. membership in the latent classes. These vari-
Additional details on the sample are available in ables also served as controls when we examined
the first column of Table 2. associations between latent class membership
For the longitudinal analysis, we drew on and youth adjustment. Adolescent variables
Wave III data collected in 2001–2002, when included whether the adolescent was a daughter
youth were in their early adult years (ages (0 = son, 1 = daughter), the adolescent’s age at
18–26). The Wave III sample size was 1,408, Wave I (in years), and if the adolescent was
or 73% of the original Wave I sample. Attrition not a citizen (0 = a U.S. citizen, 1 = not a U.S.
between waves was more common among men citizen). Adolescent race/ethnicity was captured
than women, among youth without stepsiblings with four dummy variables: (a) non-Hispanic
than with stepsiblings, and among youth who did White (reference group), (b) non-Hispanic
not know their nonresident fathers. Attrition was Black, (c) Hispanic, and (d) other. Ordered
not related to any of the other variables used in measures of mother’s education and stepfather’s
the analysis. education were used to measure social class
(range: 1 = less than high school education
to 4 = college education or more), along with
Analysis
family income in logged dollars. The number
Data analysis was conducted using Mplus of years in a stepfamily involved the length
version 6 (Muthén & Muthén, 2010) with of time the adolescent had lived in the same
full-information maximum-likelihood estima- household with the stepfather, regardless of
tion to handle missing data. This approach uses whether the union had begun with cohabitation
all available data when estimating parameters, or marriage.
thereby reducing missing data biases (Enders & Information about siblings was drawn
Bandalos, 2001). Results are based on weighted from the household roster and included a
data, with standard errors adjusted for clustering count variable for the number of full sib-
and stratification in the Add Health sampling lings and two binary indicators indicating
design. whether the respondent lived with any step-
siblings (0 = no stepsiblings, 1 = stepsiblings)
or any half-siblings (0 = no half-siblings,
Measures 1 = half-siblings). To measure the number of
Parent–child relationship measures. Closeness father figures, we drew on questions about the
to mother was measured with a single item mother’s relationship history and the number of
asking adolescents how close they felt to their coresidential relationships (cohabitations and
resident biological mothers (range: 1 = not at marriages) to which the adolescent had been
all close to 5 = very close). Identical questions exposed since birth. A binary variable indicated
were used to measure closeness to stepfather whether the adolescent was born in marriage
and closeness to nonresident father. Adolescents (0 = not born in marriage, 1 = born in marriage).
who stated that they never saw or did not know Finally, religiosity was based on the mean of
17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Parent–Child Relationships in Adolescence 487

three standardized items dealing with how often something with a value under $50, stole some-
the adolescent attended religious services, the thing with a value over $50, entered onto
importance of religion, and participation in someone else’s property with the intention of
religious activities (𝛼 = .82). Information on stealing, sold drugs, got rowdy in a public place,
the mother’s education, stepfather’s education, got into a serious physical fight, used or threat-
family income, the number of father figures, ened to use a weapon, took part in a fight with
and whether the adolescent had been born in a group of their friends against another group
marriage was obtained from the mother inter- of people, shot or stabbed someone, and got
view; all other variables were derived from the into a fight in which the other person was seri-
adolescent interview. ously injured. These items were dichotomized
(0 = never, 1 = at least once) and summed to cre-
Measures of adjustment. We examined three ate an index of adolescent delinquency (range:
aspects of adjustment during adolescence and 0–15, M = 2.26, SD = 2.53).
young adulthood: (a) depressive symptoms, (b) The corresponding variable for young adults
delinquency, and (c) substance use. Scales were was derived from Wave III questions asking
constructed from items drawn from the Wave I respondents whether (in the past year) they
in-home interview for the cross-sectional anal- deliberately damaged someone else’s property,
ysis and from the Wave III in-home interview stole something with a value under $50, stole
for the longitudinal analysis. Fewer items were something with a value over $50, entered onto
available for some constructs in Wave III than someone else’s property with the intention of
in Wave I. stealing, sold drugs, used or threatened to use a
We measured depressive symptoms with weapon, took part in a fight with a group of their
items from the Center for Epidemiological friends against another group of people, or got
Studies Depression Scale (Radloff, 1977). The
into a fight in which the other person was seri-
Wave I scale asked how often during the previ-
ously injured. These items were dichotomized
ous week adolescents had the following feelings
(0 = never, 1 = at least once) and summed to cre-
or experiences: felt sad, were depressed, felt
ate a scale of young adult antisocial activity
lonely, felt fearful, felt disliked by others, could
(range: 0–8, M = 0.49, SD = 1.20).
not shake off the blues, were too tired to do
things, felt that their lives were not worth living, We also assessed adolescents’ and young
were bothered by things more than usual, had adults’ use of three substances: (a) cigarettes
trouble focusing, had a poor appetite, talked (tobacco), (b) alcohol, and (c) marijuana. Ques-
less than usual, did not enjoy life, felt happy, tions were identical at Wave I and Wave III. To
felt hopeful, and felt that they were as good as measure smoking, respondents were asked about
other people (𝛼 = .88). Responses ranged from the number of days in the past month on which
0 (rarely or never) to 3 (most or all of the time). they had smoked any cigarettes. We created a
The corresponding Wave III young adult score dichotomous item (0 = no smoking, 1 = smoking
was based on a subset of nine items from the on 1 or more days) to indicate whether respon-
adolescent interview (𝛼 = .81). The Wave I and dents had used cigarettes (or had been cigarette
Wave III symptoms scales were standardized to free) during the past month. Nearly one third of
have means of 0 and standard deviations of 1 adolescents (31%) and nearly one half of young
within each wave. adults (49%) reported smoking in the previous
Delinquency was based on 15 different month. Frequent binge drinking was captured
delinquent activities during adolescence and with a question asking how often during the
eight different delinquent (or criminal) activ- past year respondents had been “drunk” or “very
ities during young adulthood. The adolescent high” on alcohol (0 = drunk once per month or
delinquency variable was drawn from a series less, 1 = drunk more than once a month). Scores
of Wave I questions that asked whether respon- of 1 on this item were assigned to 18% of ado-
dents had engaged in the following activities lescents and 24% of young adults. Finally, a
during the past year: painted graffiti on someone question asked whether respondents had smoked
else’s property, deliberately damaged someone marijuana in the past month (0 = no, 1 = yes).
else’s property, lied to their parents about where Positive responses to this question were pro-
they were or who they were with, shoplifted, vided by 15% of adolescents and 24% of young
took someone’s car without permission, stole adults.
17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
488 Journal of Marriage and Family

Results of relative closeness). Adolescents in Class 1


The Latent Classes (9% of the sample) did not report being partic-
ularly close to any of their parents. The mean
We estimated solutions with two, three, four, ratings (raw scores) for stepfathers, mothers, and
five, and six latent classes and relied on three nonresident biological fathers were 2.58, 2.77,
measures to determine the best solution. Entropy and 2.61, respectively. All three means were
is a measure of how well individual cases can below 3, the midpoint of the response options.
be classified unambiguously and ranges from Correspondingly, the mean Z scores for stepfa-
0 to 1, with larger values indicating a clearer thers and mothers were negative and substantial
delineation of classes. The Bayesian Informa- (−0.99 and −2.35, respectively), although the
tion Criterion is a measure of model fit, with mean Z score for nonresident biological fathers
lower values indicating that a given model is (−0.05) was only slightly negative and close to
more likely to be the true model. Finally, the the grand mean of 0. Sixteen percent of adoles-
Lo–Mendell–Rubin test indicates whether a cents in this group had no contact with their bio-
solution with K classes provides a significantly logical fathers. We refer to this class as not close
better fit to the data than a solution with K − 1 to resident parents.
classes. Adolescents in Class 2 (20% of the sam-
In the current study, the three measures pro- ple) reported being moderately close to all of
vided inconsistent results. Entropy increased their parents, although the mean rating for non-
from two to four classes and then declined, resident fathers (2.49) was below the midpoint
which suggested that the four-class solution of the response scale. All three of the mean Z
was optimal. Bayesian Information Criterion scores were negative. Nineteen percent of ado-
values were lowest for the five-class solution, lescents in this group did not know their biolog-
which suggested that this was the optimal solu- ical fathers. We refer to this class as moderately
tion. And the Lo–Mendell–Rubin test was not close to resident parents.
significant for solutions with more than three Adolescents in Class 3 (16% of the sample)
classes, which suggested that the three-class reported being close to their stepfathers (4.2) and
solution was optimal. When fit indices yield very close to their mothers (5.0). Both of the cor-
contrary conclusions it is necessary to examine responding Z scores were positive. No adoles-
all of the best-fitting solutions to see which cents in this group knew their biological fathers.
has the most heuristic value (Collins & Lanza, (The proportion not knowing their fathers was
2010). Our examination of the three-, four-, and 1.0). By default, these adolescents had been
five-class solutions revealed that the four-class assigned the lowest possible score of 1 on the
solution had the clearest interpretation. The father closeness rating, as noted earlier. We label
three-class solution largely combined Classes 3 this group close to resident parents–don’t know
and 4, which masked the important distinction nonresident father.
between children with either close or nonexis- Finally, adolescents in Class 4 (55% of the
tent relationships with nonresident fathers (see sample) reported being close to their stepfathers
below), whereas the five-class solution essen- (with a mean score of nearly 4) and very close
tially divided Class 2 into two subclasses that to their mothers (with a mean score of exactly
differed in what appeared to be substantively 5). Moreover, all of these adolescents (100%)
unimportant ways. The four-class solution not knew their nonresident biological fathers. These
only had a clear interpretation but also had an adolescents also were relatively close to their
excellent entropy value (.99), which indicated fathers, as reflected in a mean raw score above
that the cases could be classified into four groups the midpoint of the response scale and a positive
with a high degree of certainty. mean Z score. We refer to this class as close to
The means of the four variables used to gen- all parents.
erate the latent classes are shown in Table 1.
The first column shows the overall means for
the full sample, and subsequent columns show Differences Between Latent Classes
the means for the four latent classes. For the on Background Variables
three closeness ratings, we report means based The data in Table 2 show the means for all of
on raw scores (to provide a sense of absolute the background variables by class membership.
closeness) as well as Z scores (to provide a sense The first column shows the overall mean for
17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Parent–Child Relationships in Adolescence 489

Table 1. Means of Relationship Indicators by Latent Class Membership

Variable Full sample Class 1a Class 2b Class 3c Class 4d

Closeness to stepfather
Mean raw score 3.70 2.58 3.10 4.20 3.95
SE 0.15 0.08 0.11 0.05
Mean Z score 0.00 −0.99 −0.54 0.42 0.21
Closeness to mother
Mean raw score 4.61 2.77 4.00 5.00 5.00
SE 0.05 0.00 0.00 0.00
Mean Z score 0.00 −2.35 −0.74 0.57 0.57
Closeness to nonresident father
Mean raw score 2.63 2.61 2.49 1.00 3.16
SE 0.14 0.09 0.00 0.07
Mean Z score 0.00 −0.05 −0.12 −1.01 0.33
Don’t know nonresident father
Mean raw score 0.21 0.16 0.19 1.00 0.00
SE 0.04 0.03 0.00 0.00
N (unweighted) 197 383 301 1,053
Proportion (weighted) .09 .20 .16 .55
Note. Means are based on weighted data. Standard errors are adjusted for weighting, clustering, and stratification.
a Not close to resident parents (9%). b Moderately close to resident parents (20%). c Close to resident parents–don’t know

nonresident father (16%). d Close to all parents (55%).

the full sample, and subsequent columns show was significant. In addition, adolescents in this
the means for the four latent classes. To sup- class were less likely than other adolescents to
plement the means shown in Table 2 we used have been born within marital unions (two of
multinomial logistic regression to regress class the three differences were significant). Taken
membership onto the background variables. We together, these results indicate that the close to
conducted three regression analyses and rotated resident parents–don’t know nonresident father
the excluded group to provide contrasts among class was made up largely of adolescents born to
all four classes. Significant differences between unwed mothers with comparatively little educa-
groups (based on the multivariate results) tion who had formed new unions (and married
are reported in the final column of Table 2. their partners) when their children were rela-
(The full multinomial results are available tively young. Given this constellation of traits,
on request.) it is not surprising that these adolescents knew
The analysis was clearest in distinguish- little about their biological fathers (Cheadle,
ing adolescents in the close to resident Amato, & King, 2010). It is noteworthy that
parents–don’t know nonresident father class despite their somewhat disadvantaged circum-
(Class 3) from other adolescents. Table 2 shows stances, these families not only stayed together
that adolescents in this class were younger than but also maintained close relationships between
adolescents in the other three classes, and the adolescents and stepfathers.
multivariate analysis indicated that these differ- Adolescents in the not close to resident par-
ences were statistically significant. Adolescents ents class (Class 1) were especially likely to
in this class also had been in stepfamilies for be women, with the differences between this
the longest time (more than 9 years), and the class and two other classes being statistically
differences between this class and the other significant in the multivariate analysis. Other
three classes were significant. Adolescents in researchers have noted that adolescent daughters
this class had the lowest mean scores for mother are more likely than sons to report friction in
and stepfather education, although most of these stepfamilies (Hetherington & Jodl, 1994). Ado-
differences were not significant. Although ado- lescents in this class also tended to be somewhat
lescents in this class were the most likely to be older than other adolescents. Other than these
Hispanic, only one difference between classes differences, adolescents in this class were not
17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
490 Journal of Marriage and Family

Table 2. Means (and Standard Errors, in Parentheses) of Background Variables by Latent Classes

Variable Total Class 1a Class 2b Class 3c Class 4d Differences, p < .05

Daughter .51 .66 .54 .48 .48 1 > 3, 4


(.02) (.06) (.03) (.03) (.02)
Age 15.39 15.78 15.64 15.00 15.35 1, 2, 4 > 3;
(0.13) (0.18) (0.14) (0.15) (0.15) 1>4
Years stepfamily 7.58 7.36 7.42 9.21 7.07 3 > 1, 2, 4
(0.17) (0.53) (0.29) (0.34) (0.22)
Number full sibs 0.70 0.71 0.65 0.69 0.72
(0.04) (0.09) (0.09) (0.07) (0.05)
Any half-sibs .41 .42 .38 .54 .38
(.02) (.05) (.03) (.04) (.02)
Any stepsibs .11 .06 .09 .10 .12
(.01) (.02) (.02) (.02) (.02)
Mother education 2.52 2.49 2.64 2.24 2.57
(0.04) (0.13) (0.07) (0.09) (0.04)
Stepfather education 2.55 2.60 2.53 2.23 2.63 4>3
(0.05) (0.12) (0.08) (0.09) (0.06)
Log income 3.50 3.54 3.71 3.36 3.45 2>4
(0.05) (0.12) (0.05) (0.07) (0.08)
Hispanic .11 .13 .08 .19 .10 3>2
(.02) (.03) (.02) (.04) (.02)
Black .13 .14 .09 .16 .13 4>2
(.02) (.03) (.02) (.03) (.02)
Other race .06 .04 .09 .06 .06
(.01) (.02) (.02) (.02) (.01)
Child not citizen .04 .07 .03 .06 .02
(.01) (.02) (.01) (.02) (.01)
Number of father figures 1.94 1.86 2.01 1.84 1.97
(0.03) (0.07) (0.04) (0.07) (0.03)
Marital birth .76 .71 .76 .56 .82 4 > 2, 3; 2 > 3
(.02) (.07) (.03) (.05) (.02)
Religiosity Z −0.03 −0.01 −0.08 0.01 0.00
(0.04) (0.07) (0.07) (0.07) (0.04)
Note. Total N = 1,934. Means are based on weighted data. Standard errors are adjusted for weighting, survey clustering,
and stratification. Significance tests for group differences are based on multinomial logistic regression.
a Not close to resident parents (9%). b Moderately close to resident parents (20%). c Close to resident parents–don’t know

nonresident father (16%). d Close to all parents (55%).

notably different from other adolescents on the shortest duration (with one significant differ-
background variables. ence between classes). Adolescents in this class
Adolescents in the moderately close to resi- also were the most likely to have been born
dent parents group (Class 2) were comparable within marriage, and two of the three differ-
to those in Class 1 (not close to resident parents) ences between this class and the other classes
in being older than average. They were the least were statistically significant.
likely of any class to be non-Hispanic Black,
although the differences among classes were
modest. They also tended to have the highest Latent Class Differences in Adjustment
family income (significantly higher than Class The next step in the analysis compared the
4). Otherwise, members of this group did not four groups of adolescents on the measures of
differ appreciably from the other classes. adjustment in adolescence and early adulthood.
Adolescents in the close to all parents group To accomplish this goal, we conducted a series
(Class 4) had been in stepfamilies for the of regression analyses with dummy variables to
17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Parent–Child Relationships in Adolescence 491

represent the latent classes. We relied on linear difference between those who did not know
regression for depressive symptoms, Poisson their nonresident biological fathers (Class 3) or
regression for the count measure of delinquency were close to their nonresident fathers (Class 4).
(or criminality among young adults), and logis- This pattern was not replicated in early adult-
tic regression for the binary substance use hood; instead, the bivariate model indicated
measures. Table 3 shows the results from two a reversal, with individuals who had been in
regression models with Class 4 (close to all par- Class 1 as adolescents (not close to resident
ents) serving as the omitted comparison group. parents) showing the lowest level of delinquent
We conducted additional regression analyses behavior. No differences between classes were
with the other classes serving as the omitted significant in the multivariate model, however,
comparison group. This made it possible to so we do not discuss the results for young adults
examine all possible contrasts between groups, further.
and the significant differences are summarized in With respect to substance use, adolescents in
the table. Model 1 is bivariate, whereas Model 2 the close to resident parents–don’t know nonres-
controls for all of the background variables listed ident father group (Class 3) were less likely to
in Table 2. Results for adolescents (Wave I) are report smoking cigarettes than were adolescents
shown on the left side of the table, and results for in the other three groups, and the same trend was
young adults (Wave III) are shown on the right. apparent in the bivariate and multivariate results.
With respect to symptoms of depression, ado- To provide an idea of the magnitude of these dif-
lescents who were not close to resident parents ferences, 20% of adolescents in Class 3 reported
or moderately close to resident parents (Classes smoking cigarettes in the previous month com-
1 and 2) reported more symptoms than did ado- pared with 34% in Class 1, 37% in Class 2, and
lescents with stronger ties to parents (Classes 3 32% in Class 4. The results for young adults
and 4). The same pattern was apparent in the were similar, which indicates that the tendency
bivariate and multivariate models. These find- for adolescents who were close to their resi-
ings are consistent with the notion that having dent parents (but not their nonresident fathers) to
close ties with parents protects adolescents from avoid cigarette smoking continued into the early
experiencing symptoms of depression. There adult years.
was no difference, however, between adoles- With respect to alcohol use, the bivariate
cents in the close to resident parents–don’t know results show that adolescents in the close to
nonresident father group (Class 3) and the close resident parents–don’t know nonresident father
to all parents group (Class 4). The results for (Class 3) group were less likely than adoles-
young adults were similar, although only the dif- cents in the moderately close to resident parents
ference between the not close to resident parents group (Class 2) and the close to all parents group
group (Class 1) and the close to all parents class (Class 4) to engage in frequent binge drinking.
(Class 4) was statistically significant. The unadjusted percentages were 14% for ado-
With respect to delinquency, only one dif- lescents in Class 3 compared with 19% in Class
ference among groups was significant in Model 1, 27% in Class 2, and 19% in Class 4. The dif-
1. Controlling for the background variables in ferences between classes no longer were signif-
Model 2 revealed several additional differences, icant in the multivariate analysis, however, and
with adolescents in the not close to resident the same pattern of results was apparent among
parents group (Class 1) reporting more delin- young adults.
quency than adolescents in the close to resident Finally, although not all contrasts were sta-
parents–don’t know nonresident father group tistically significant, adolescents in the moder-
(Class 3) and the close to all parents group ately close to resident parents group (Class 2)
(Class 4). In addition, adolescents in the mod- were the most likely to have used marijuana
erately close to resident parents group (Class 2) in the past month, and adolescents in close to
scored higher than did adolescents in the close resident parents–don’t know nonresident father
to all parents group (Class 4). These findings group (Class 3) were the least likely. The unad-
are consistent with the notion that having close justed percentages were 17% in Class 1, 24%
ties with parents protects adolescents from in Class 2, 8% in Class 3, and 16% in Class
drifting into antisocial activities. Once again, 4. Similar trends were apparent in the mul-
however, among adolescents who had close tivariate as well as the bivariate analysis in
ties with mothers and stepfathers there was no both waves.
17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
492 Journal of Marriage and Family

Table 3. Regression Analysis of Adjustment Measures on Latent Classes, Waves I and III

Adolescence (Wave I) Adulthood (Wave III)


Predictor Model 1 Model 2 Model 1 Model 2

Depressive symptoms
Class 1a .31 .27 .24 .24
(.07) (.07) (.09) (.08)
Class 2b .18 .16 .10 .09
(.05) (.05) (.06) (.06)
Class 3c −.02 −.04 .08 .04
(.05) (.05) (.08) (.08)
Class 4d .00 .00 .00 .00
Differences, p < .05 1, 2 > 3, 4 1, 2 > 3, 4 1>4 1>4
Delinquency
Class 1a .22 .30 −.25 −.04
(.09) (.10) (.09) (.09)
Class 2b .14 .15 .03 .11
(.08) (.08) (.12) (.11)
Class 3c .08 .03 −.07 −.11
(.09) (.09) (.12) (.13)
Class 4d .00 .00 .00 .00
Differences, p < .05 1>4 1 > 3, 4; 2 > 4 1 < 2, 4 None
Smoking
Class 1a .14 .02 .07 .24
(.25) (.24) (.25) (.25)
Class 2b .24 .10 .05 .03
(.18) (.20) (.17) (.19)
Class 3c −.60 −.58 −.49 −.48
(.21) (.21) (.25) (.24)
Class 4d .00 .00 .00 .00
Differences p < .05 1, 2, 4 > 3 1, 2, 4 > 3 2, 4 > 3 1, 2, 4 > 3
Binge drinking
Class1a .07 −.04 −.17 .03
(.30) (.33) (.20) (.22)
Class 2 .44 .35 .15 .21
(.17) (.19) (.20) (.20)
Class 3c −.38 −.17 −.54 −.32
(.26) (.29) (.22) (.25)
Class 4d .00 .00 .00 .00
Differences, p < .05 2, 4 > 3 None 2, 4 > 3 None
Marijuana
Class 1a .10 .12 −.03 .16
(.29) (.34) (.28) (.28)
Class 2b .56 .56 .29 .40
(.20) (.20) (.18) (.19)
Class 3c −.74 −.58 −.31 −.18
(.38) (.40) (.24) (.25)
Class 4d .00 .00 .00 .00
Differences, p < .05 2 > 3, 4; 4 > 3 2 > 3, 4 2>3 2 > 3, 4
Note. Numbers in parentheses are standard errors. Total N = 1,934. Table values are linear regression coefficients for
depression; Poisson regression coefficients for delinquency; and logistic regression coefficients for smoking, binge drinking,
and marijuana. Coefficients in Model 1 are from bivariate analyses. Model 2 includes controls for all background variables.
Class 4 serves as the omitted comparison group, although the comparison group was rotated to provide group comparisons for
significance testing. Results are weighted, and standard errors are adjusted for weighting, survey clustering, and stratification.
a Not close to resident parents (9%). b Moderately close to resident parents (20%). c Close to resident parents–don’t know

nonresident father (16%). d Close to all parents (55%).


17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Parent–Child Relationships in Adolescence 493

Overall, adolescents who were close to their of stepfamilies with adolescents. The remaining
mothers and stepfathers but not to their biolog- adolescents (16% of the total) were close to
ical fathers (Class 3) tended to report the low- both resident parents but had no relationship
est levels of substance use. Moreover, across all with their nonresident biological fathers. These
five outcomes, adolescents in this class were no groups overlap a good deal with the groups
worse off than were adolescents who were close observed by Baxter et al. (2006)—the only
to all of their parents, including their nonresident other study to our knowledge that attempted to
fathers (Class 4). The same trends were apparent describe broad configurations of relationships
in early adulthood, although somewhat attenu- in stepfamilies. Their study, however, did not
ated. We return to these unexpected findings in include nonresident fathers, involved university
the discussion section. students as respondents, and used qualitative
rather than quantitative methods. For these
reasons, it is difficult to compare the results of
Discussion the two studies.
To understand variation within stepfamilies We did not see evidence of household
and how this variation is related to children’s coalitions—relationship patterns often dis-
adjustment, previous researchers have divided cussed by systems theorists—in our data (e.g.,
stepfamilies into groups based on structural Kerr & Bowen, 1988). For example, no classes
characteristics, such as the stepparent’s gen- emerged in which children were close to resident
der or the presence of step- or half-siblings mothers but excluded their stepfathers. Instead,
in the household (e.g., Fine & Kurdek, 1992; adolescents’ relationships with both resident
Ganong & Coleman, 1986; Hetherington & parents tended to be either distant (Class 1),
Stanley-Hagan, 2000). Although structural moderately close (Class 2), or very close
characteristics of stepfamilies are important, (Classes 3 and 4). This finding is consistent with
few researchers have attempted to distinguish prior research showing that closeness to mothers
between stepfamilies on the basis of relation- is positively correlated with closeness to stepfa-
ship characteristics. This omission is curious, thers (King, Thorsen, & Amato, 2014). Adoles-
given systems theory’s emphasis on emotions cents who were close to both resident parents,
and its view of families as networks of inter- however, were either distant (Class 3) or close
locking relationships (Broderick, 1993; Kerr (Class 4) to their nonresident fathers, so con-
& Bowen, 1988; Minuchin, 1974). To explore sistency in adolescent–parent relationships did
this approach, we focused on adolescents’ not extend beyond the household. These results
reports of closeness to mothers, stepfathers, suggest that children develop close ties with
and nonresident biological fathers. We relied on stepfathers either when (a) their biological
LCA—an appropriate method when the number fathers continue to be actively involved in their
and characteristics of groups are not known children’s lives or (b) their stepfathers “take
a priori. To our knowledge, no other study the place” of completely absent fathers. The
has relied on LCA to understand stepfamily presence of both dynamics would account for
relationships. the absence of a linear correlation between
Although the number of relationship patterns closeness to fathers and closeness to stepfathers
that characterize stepfamilies is potentially (Dunn et al., 2004; Jensen & Shafer, 2013; King
large, parent–adolescent relationships in the et al., 2014).
present study fell into four general groups. Adolescents in Class 3 (close to resident
Some adolescents (9%) were distant from their mothers and stepfathers but not close to non-
mothers and stepfathers, whereas others (20%) resident biological fathers) are of particular
were moderately close to their mothers and interest. Most of these adolescents were born
stepfathers. Adolescents in both groups gener- outside of marriage and entered stepfamilies
ally knew their nonresident fathers but were not at early ages (age 5, on average). The young
particularly close to them. Most adolescents, age of these children at the time of stepfam-
however, were close to their mothers and step- ily formation and the many years of residing
fathers, and some of these adolescents (55% of together probably contributed to the closeness of
the total) also were close to their nonresident adolescent–stepfather ties in this group (Heth-
biological fathers. So, a pattern of close relation- erington & Jodl, 1994). It is noteworthy that
ships with all parents characterized the majority these closely knit stepfamilies stayed together
17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
494 Journal of Marriage and Family

(more than 9 years, on average) despite being reflect a potential disadvantage of being close
disadvantaged socioeconomically. to nonresident fathers. Adolescents who spend
As we anticipated, the latent classes that significant amounts of time with nonresident
emerged from the analysis were related to fathers are likely to travel regularly between
aspects of adolescent adjustment. Consistent two households, and it is possible that splitting
with our first general hypothesis, adolescents time across two households makes it difficult
with weak ties to resident parents (Class 1) for parents to effectively monitor their adoles-
reported the most symptoms of depression and cents’ peer networks and experimentation with
the largest number of delinquent activities. substances. Moreover, spending time in two
Moreover, the longitudinal analysis revealed households may increase adolescents’ expo-
that these individuals continued to report a high sure to people (peers and adults) who smoke
number of depressive symptoms 6–7 years later, cigarettes, including, in some cases, nonresident
although they did not report an elevated number fathers and their new partners.
of antisocial behaviors. Controlling for a variety Our findings also appear to clash with
of individual and family characteristics helped previous research showing that closeness to
ensure that these trends were not spurious, with nonresident fathers is negatively associated
most of the significant group differences in the with children’s emotional and behavioral prob-
bivariate analysis continuing to be significant lems (Adamsons & Johnson, 2013; Amato &
in the multivariate analysis. These results are Gilbreth, 1999). It may be, however, that many
broadly consistent with the notion that close adolescents require only one close father figure
parent–child relationships protect adolescents in their lives. If this is true, then adolescents
from a broad range of internalizing and exter- with close ties to their stepfathers may not
nalizing problems (Buchanan, Maccoby, & “need” their biological fathers, although they
Dornbusch, 2000; Gray & Steinberg, 1999; may value continuing contact. This conclusion
Steinberg, 2001)—a principle that applies to is consistent with that reached by King (2006),
stepfamilies as well biological-parent families who found that close ties to stepfathers were
(Stewart, 2007). These results also are consis- more important than close ties to nonresident
tent with the notion that parents find it easier biological fathers with respect to two out of
to bond with adolescents who are emotionally three adolescent outcomes. Moreover, most
adjusted and well behaved (Hawkins, Amato, & adolescents in Class 3 (close to resident par-
King, 2007). Parent–adolescent closeness and ents but not to nonresident fathers) had been
adolescent behavior are likely to be related in a born outside of marriage and were relatively
reciprocal fashion, although demonstrating this disadvantaged socioeconomically. It may be
point is beyond the scope of the current study. that in disadvantaged populations, close rela-
With respect to substance use, adolescents tionships with nonresident fathers involve costs
with weak ties to resident parents (Class 1) or as well as benefits (Thomas, Farrell, & Barnes,
moderately strong ties to resident parents (Class 1996). This might be the case when fathers are
2) were the mostly likely to report smoking struggling with problems often associated with
cigarettes, binge drinking, and marijuana use poverty, such as unemployment, discrimination,
(despite some variation across models and time substance abuse, incarceration, and chronic
periods). Adolescents who were close to their psychological distress. The costs of maintaining
resident parents but did not know their nonres- close relationships under these circumstances
ident fathers (Class 3) consistently reported the might cancel any benefits and account for why
lowest levels of substance use, and this tendency adolescents in Class 3 showed little evidence of
persisted into early adulthood. Indeed, adoles- impairments despite having no involvement with
cents in Class 3 were significantly less likely to their biological fathers. Because Add Health
report smoking cigarettes than were adolescents contains minimal information on nonresident
in Class 4 at both waves. These results clash fathers, testing this explanation was not possible
with our hypothesis that adolescents who were in the current study, although it would make a
close to all of their parents (Class 4) would useful starting point for further research.
exhibit the most positive outcome profile and Like all studies, the current one involves
that adolescents with a mixed pattern of close- significant limitations. Because the Add Health
ness (Class 3) would have an intermediate level interview did not include questions about
of adjustment. We suspect that these findings adolescents’ closeness to mothers’ cohabiting
17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Parent–Child Relationships in Adolescence 495

partners, we were unable to incorporate infor- are related to youth development—would be


mation on cohabiting stepfamilies. (Stepfamilies useful to counselors, therapists, and educators
that began as cohabiting partnerships and turned who work with stepfamilies (e.g., Lucier-Greer
into marriages were included in the analysis, & Adler-Baeder, 2012).
however.) We also lacked information on step-
fathers’ (or other parents’) feelings of closeness Note
to adolescents. Although we assume that ado-
lescents’ and stepfathers’ feelings are positively This research was supported by funding from the National
Science Foundation (NSF) to Valarie King, Principal Inves-
correlated, a significant minority of cases might tigator (SES-1153189) and by funding from the Eunice
exist in which adolescents and stepfathers (or Kennedy Shriver National Institute of Child Health and
other parents) hold discrepant feelings toward Human Development (NICHD) to the Population Research
one another. Similarly, we were unable to Institute at The Pennsylvania State University for Popu-
lation Research Infrastructure (R24 HD41025) and Fam-
include questions about closeness to nonresi- ily Demography Training (T-32HD007514). This research
dent stepmothers. In addition, we did not have a uses data from Add Health, a program project directed
sufficient number of cases to perform a compa- by Kathleen Mullan Harris and designed by J. Richard
rable analysis for resident stepmother families. Udry, Peter S. Bearman, and Kathleen Mullan Harris at the
Moreover, we focused on a single relationship University of North Carolina at Chapel Hill, and funded
by Grant P01-HD31921 from the Eunice Kennedy Shriver
dimension, closeness to parents, and this was National Institute of Child Health and Human Development,
measured with a single item. Although emo- with cooperative funding from 23 other federal agencies
tional closeness is a central relationship feature, and foundations. Special acknowledgment is due Ronald
broadening the focus to include other relation- R. Rindfuss and Barbara Entwisle for assistance in the
original design. Information on how to obtain the Add
ship dimensions (e.g., the frequency of sharing Health data files is available on the Add Health website
activities or parental monitoring and supervi- (http://www.cpc.unc.edu/addhealth). No direct support was
sion) might provide more detailed distinctions received from Grant P01-HD31921 for this analysis.
between a more nuanced set of latent classes.
Finally, nonresident father–child contact has
References
increased in recent decades (Amato, Meyers,
& Emery, 2007). Given that the Wave 1 Add Adamsons, K., & Johnson, S. K. (2013). An updated
Health data were collected in the mid-1990s, and expanded meta-analysis of nonresident father-
ing and child well-being. Journal of Family Psy-
our results for nonresident fathers may not apply
chology, 27, 589–599. doi:10.1037/a0033786
to current cohorts of adolescents. Amato, P. R. (2010). Research on divorce:
In conclusion, family systems theory shifts Continuing trends and new developments. Jour-
our attention away from dyads and toward larger nal of Marriage and Family, 72, 650–666.
patterns of relationships within stepfamilies. doi:10.1111/j.1741-3737.2010.00723.x
The present study indicates that LCA can distin- Amato, P. R., & Gilbreth, J. (1999). Nonresident
guish among groups of stepfamilies on the basis fathers and children’s well-being: A meta-analysis.
of relationship closeness in a manner that yields Journal of Marriage and the Family, 61, 557–573.
compelling classes. Moreover, the present study doi:10.2307/353560
demonstrates that these relationship configura- Amato, P. R., Meyers, C. E., & Emery, R. E. (2009).
Changes in nonresident father–child contact from
tions are bound up with multiple forms of adjust- 1976 to 2002. Family Relations, 58, 41–53.
ment that persist from adolescence into the early Arnett, J. J. (2000). Emerging adulthood: A the-
adult years. Our analysis also shows that adjust- ory of development from the late teens through
ment is not a simple function of the number of the twenties. American Psychologist, 55, 469–480.
positive relationships in the family network. In doi:10.1037/0003-066X.55.5.469
particular, when adolescents have close relation- Baxter, L. A., Braithwaite, D. O., & Bryant, L. E.
ships with stepfathers they appear to receive little (2006). Types of communication triads perceived
additional benefit from having close relation- by young-adult stepchildren in established step-
ships with nonresident fathers. Future studies families. Communication Studies, 57, 381–400.
doi:10.1080/10510970600945923
can build on the current findings by using LCA
Bornstein, M. (2002). Handbook of parenting: Being
to explore naturally occurring variation within and becoming a parent. Mahwah, NJ: Erlbaum.
stepfamilies (as well as in other family forms) Bray, J. H. (1999). From marriage to remarriage and
with more detailed relationships characteristics. beyond: Findings from the developmental issues in
A clearer understanding of how stepfamilies stepfamilies research project. In E.M. Hethering-
differ from one another—especially in ways that ton (Ed.), Coping with divorce, single parenting,
17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
496 Journal of Marriage and Family

and remarriage: A risk and resiliency perspective Ganong, L. H., & Coleman, M. (1986). A comparison
(pp. 253–271). Mahwah, NJ: Erlbaum. of clinical and empirical literature on children in
Bray, J. H., & Easling, I. (2005). Remarriage and stepfamilies. Journal of Marriage and the Family,
stepfamilies. In W. M. Pinsof & J. L. Lebow 48, 309–318. doi:10.2307/352398
(Eds.), Family psychology: The art of the science Goodman, W. B., Crouter, A. C., Lanza, S., Cox, M.,
(pp. 267–294). New York: Oxford University & Vernon-Feagans, L. (2011). Paternal work stress
Press. and latent profiles of father–infant parenting qual-
Broderick, C. (1993). Understanding family process. ity. Journal of Marriage and Family, 73, 588–604.
Newbury Park, CA: Sage. doi:10.1111/j.1741-3737.2011.00826.x
Brown, S. L., Manning, W. D., & Stykes, J. B. (2015). Gray, M. R., & Steinberg, L. (1999). Unpacking
Family structure and child well-being: Integrating authoritative parenting: Reassessing a multidimen-
family complexity. Journal of Marriage and Fam- sional construct. Journal of Marriage and the Fam-
ily, 77, 177–190. ily, 61, 574–587. doi:10.2307/353561
Buchanan, C. M., Maccoby, E. E., & Dornbusch, S. Guo, M., Chi, I., & Silverstein, M. (2012). The
M. (2000). Adolescents after divorce. Cambridge, structure of intergenerational relations in rural
MA: Harvard University Press. China: A latent class analysis. Journal of Marriage
Cavanagh, S. E., Schiller, K. S., & Riegle-Crumb, and Family, 74, 1114–1128. doi:10.1111/j.1741-
C. (2006). Marital transitions, parenting, and 3737.2012.01014.x
schooling: Exploring the linkage between fam- Harris, K. M., Halpern, C. T., Whitsel, E., Hussey,
ily structure history and adolescents’ academic J., Tabor, J., Entzel, P., & Udry, J. R. (2009).
success. Sociology of Education 79, 329–354. The National Longitudinal Study of Adoles-
doi:10.1177/003804070607900403 cent Health: Research design. Retrieved from
Cheadle, J., Amato, P. R., & King, V. (2010). Pat- http://www.cpc.unc.edu/projects/addhealth/design
terns of nonresident involvement. Demography, Hawkins, D., Amato, P. R., & King, V. (2007). The
47, 205–226. doi:10.1353/dem.0.0084 relationship between nonresident father involve-
Coleman, M., Ganong, L., & Russell, L. T. (2013). ment and adolescent well-being: Parent effects or
Resilience in stepfamilies. In D. S. Becvar (Ed.), child effects? American Sociological Review, 72,
Handbook of family resilience (pp. 85–103). 990–1010. doi:10.1177/000312240707200607
New York: Springer. doi:10.1007/978-1-4614- Hetherington, E. M., & Clingempeel, W. G. (1992).
3917-2_6 Coping with marital transitions. Monographs of
Collins, L. M., & Lanza, S. T. (2010). Latent class and the Society for Research in Child Development,
latent transition analysis: With applications in the 57(2–3), Serial No. 227. doi:10.2307/1166050
social, behavioral, and health sciences. Hoboken, Hetherington, E. M., Henderson, S. H., & Reiss,
NJ: Wiley. D. (1999). Adolescent siblings in stepfamilies:
Crouter, A. C., & Booth, A. (2003). Children’s influ- Family functioning and adolescent adjustment.
ence on family dynamics: The neglected side of Monographs of the Society for Research in
family relationships. Mahwah, NJ: Erlbaum. Child Development, 64(4), Serial No. 259.
Day, R. D., & Acock, A. (2004). Youth ratings of doi:10.1111/1540-5834.00045
family processes and father role performance of Hetherington, E. M., & Jodl, K. M. (1994). Stepfami-
resident and nonresident fathers. In R. D. Day & lies as settings for child development. In A. Booth
M. E. Lamb (Eds.), Conceptualizing and measur- & J. Dunn (Eds.), Stepfamilies: Who benefits? Who
ing father involvement (pp. 273–292). Mahwah, does not? (pp. 55–79). Hillsdale, NJ: Erlbaum.
NJ: Erlbaum. Hetherington, E. M., & Stanley-Hagan, M. (2000).
Dunn, J., Cheng, H., O’Connor, T. G., & Bridges, Diversity among stepfamilies. In D. H. Demo,
L. (2004). Children’s perspectives on their K. R. Allen, & M. A. Fine (Eds.), Handbook of
relationships with their nonresident fathers: Influ- family diversity (pp. 173–196). New York: Oxford
ences, outcomes and implications. Journal of University Press.
Child Psychology and Psychiatry, 45, 553–566. Jensen, T. M., & Shafer, K. (2013). Stepfamily func-
doi:10.1111/j.1469-7610.2004.00245.x tioning and closeness: Children’s views on sec-
Enders, C. K., & Bandalos, D. L. (2001). The rel- ond marriages and stepfather relationships. Social
ative performance of full information maximum Work, 58, 127–136. doi:10.1093/sw/swt007
likelihood estimation for missing data in structural Kerr, M. E., & Bowen, M. (1988). Family evaluation:
equation models. Structural Equation Modeling, 8, The role of the family as an emotional unit that gov-
430–457. doi:10.1207/S15328007SEM0803_5 erns individual behavior and development. New
Fine, M. A., & Kurdek, L. A. (1992). The adjust- York: Norton.
ment of adolescents in stepfather and stepmother King, V. (2006). The antecedents and consequences
families. Journal of Marriage and the Family, 54, of adolescents’ relationships with stepfathers
725–736. doi:10.2307/353156 and nonresident fathers. Journal of Marriage
17413737, 2016, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/jomf.12267 by Akbar Ali - Universiteitsbibliotheek Gent , Wiley Online Library on [26/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Parent–Child Relationships in Adolescence 497

and Family, 68, 910–928. doi:10.1111/j.1741- Smetana, J. G., Campione-Barr, N., & Metzger,
3737.2006.00304.x A. (2006). Adolescent development in interper-
King, V., Thorsen, M. L., & Amato, P. R. (2014). sonal and societal contexts. Annual Review
Factors associated with positive relationships of Psychology, 57, 255–284. doi:10.1146/
between stepfathers and adolescent stepchildren. annurev.psych.57.102904.190124
Social Science Research, 47, 16–29. doi:10.1016/ Steinberg, L. (2001). We know some things:
j.ssresearch.2014.03.010 Parent–adolescent relationships in retrospect and
Lucier-Greer, M., & Adler-Baeder, F. (2012). Does prospect. Journal of Research on Adolescence, 11,
couple and relationship education work for indi- 1–19. doi:10.1111/1532-7795.00001
viduals in stepfamilies? A meta-analytic study. Stewart, S. D. (2007). Brave new stepfamilies:
Family Relations, 61, 756–769. doi:10.1111/ Diverse paths toward stepfamily living. Thousand
j.1741-3729.2012.00728.x Oaks, CA: Sage.
McCutcheon, A. L. (1987). Latent class analysis. Sweeney, M. M. (2010). Remarriage and stepfami-
Newbury Park, CA: Sage. lies: Strategic sites for family scholarship in the
Minuchin, P. (1974). Families and family therapy. 21st century. Journal of Marriage and Family, 72,
Cambridge, MA: Harvard University Press. 667–684. doi:10.1111/j.1741-3737.2010.00724.x
Muthén, L. K., & Muthén, B. O. (2010). Mplus user’s Thomas, G., Farrell, M. P., & Barnes, G. M. (1996).
guide (6th ed.). Los Angeles: Muthén & Muthén. The effects of single-mother families and nonresi-
Pryor, J. (2014). Stepfamilies: A global perspective dent fathers on delinquency and substance abuse in
on research, policy, and practice. New York: Rout- Black and White adolescents. Journal of Marriage
ledge. and the Family, 58, 884–894. doi:10.2307/353977
Pryor, J., & Rodgers, B. (2001). Children in changing van Gaalen, R. I., & Dykstra, P. A. (2006). Solidarity
families: Life after parental separation. Malden, and conflict between adult children and parents:
MA: Blackwell. A latent class analysis. Journal of Marriage
Radloff, L. S. (1977). The CES–D Scale: A self-report and Family, 68, 947–960. doi:10.1111/j.1741-
depression scale for research in the general 3737.2006.00306.x
population. Applied Psychological Measurement,
1, 385–401. doi:10.1177/014662167700100306

You might also like