81/0)4'6;5/%51<-891:?81/0)4'6;5/%51<-891:?
'%#+063)898+01<-'%#+063)898+01<-
)+;3:?!;*31+):1659
)413?#:8;+:;8-)5,013,&-33-15/15)65=-9:-8565:->:)413?#:8;+:;8-)5,013,&-33-15/15)65=-9:-8565:->:
$0-"63-6.!)8-5:A013,"-3):1659)5,!)8-5:)365D1+:15#6;:0$0-"63-6.!)8-5:A013,"-3):1659)5,!)8-5:)365D1+:15#6;:0
68-)68-)
65):0)5)8<19
81/0)4'6;5/%51<-891:?!86<6
903-?)89-51**?
81/0)4'6;5/%51<-891:?!86<6
)903-?(/1**?*?;-,;
12)-3);.;8
81/0)4'6;5/%51<-891:?!86<6
#0)5)!81*-90
3,6415165%51<-891:?
6336=:019)5,),,1:165)3=6829):0::799+063)89)8+01<-*?;-,;.)+7;*
!)8:6.:0- :0-8#6+1)3)5,-0)<168)3#+1-5+-9644659
81/15)3!;*31+):1651:):165 81/15)3!;*31+):1651:):165
)8<1965):0)5903-?)89-51**?12)-3);.;8)5,#0)5)!81*-90B)413?
#:8;+:;8-0)5/-)5,-0)<168)3!86*3-4915#6;:068-)C!67;3):165"-9-)8+0)5,!631+?
"-<1-=
'%#+063)898+01<-1:):165'%#+063)898+01<-1:):165
)8<1965):0)51**?903-?)89-5;.;812)-3))5,!81*-90#0)5))413?#:8;+:;8-)5,
013,&-33-15/15)65=-9:-8565:->:$0-"63-6.!)8-5:A013,"-3):1659)5,!)8-5:)365D1+:15
#6;:068-)
)+;3:?!;*31+):1659
0::799+063)89)8+01<-*?;-,;.)+7;*
$019!--8"-<1-=-,8:1+3-19*86;/0::6?6;.68.8--)5,67-5)++-99*?'%#+063)898+01<-:0)9*--5
)++-7:-,.6815+3;916515)+;3:?!;*31+):1659*?)5);:0681@-,),41519:8):686.'%#+063)898+01<-68468-
15.684):16573-)9-+65:)+:-33-5()4):)5/-36*?;-,;
Vol.:(0123456789)
Population Research and Policy Review (2020) 39:439–464
https://doi.org/10.1007/s11113-020-09586-8
1 3
ORIGINAL RESEARCH
Family Structure andChild Well‑Being inaNon‑western
Context: The Role ofParent–Child Relations andParental
Conict inSouth Korea
JonathanA.Jarvis
1
· AshleyLarsenGibby
2
· MikaelaJ.Dufur
1
·
ShanaPribesh
3
Received: 25 September 2018 / Accepted: 25 April 2020 / Published online: 1 May 2020
© Springer Nature B.V. 2020
Abstract
Research suggests that children who live with two biological married parents are less
likely to exhibit behavioral problems than children who do not. While research on
childhood outcomes often focuses on the nature of the family structure itself, recent
studies have pointed to relationship quality between parents, and between parents
and children, as important mechanisms through which family structure affects chil-
dren. However, the independent influence of these two mechanisms—relationship
quality between parents and between parents and children—on childhood outcomes
has never, to our knowledge, been tested outside of a high-income western envi-
ronment. Using the Korean Youth Panel Study (KYPS), we use cross-sectional and
longitudinal GEE models to examine the influence of family structure, parent–child
relationship quality, and parental conflict on internalizing and externalizing behavior
problems among 2844 adolescents in South Korea, a high-income, non-western con-
text with relatively little family structure instability. We found that family structure,
parental conflict, and parent–child relationship quality were all significantly related
to child well-being. Children living in family structures other than with their two
married biological parents were more likely to report internalizing and externalizing
behavioral problems. Further, parental conflict was negatively related to child well-
being, while parent–child relationship quality was positively related to child well-
being. These findings show that interpersonal relationships are important to children
across family structures and should be included in studies examining how families
impact child well-being. Our study also reinforces the importance of looking at fam-
ily structures and processes in non-western contexts.
Keywords Family structure change· Parent–child relations· Parental conflict·
Child behavioral problems· Korea
* Jonathan A. Jarvis
Extended author information available on the last page of the article
440
J.A.Jarvis et al.
1 3
Introduction
Research on family structure suggests that children who live with two biological
married parents are less likely to experience behavioral problems than children who
do not, despite the normalization of non-traditional family structures in many high-
income, western countries (Lee and McLanahan 2015; Hoffmann 2017). In high-
income, non-western nations, such as Taiwan, Singapore, and South Korea, where
divorce rates have historically been much lower than in western nations (Chen and
Yip 2018; OECD 2018a), non-traditional family structures remain less common and
attitudes towards divorce more negative (Eun 2007). While the positive effects for
children living in two-parent married families are generally agreed upon, the mech-
anisms through which family structure affects childhood outcomes are less clear,
especially in these non-western nations where family structure has been more stable.
Recent studies increasingly point to the role of relationship quality between parents,
and between parents and children, in explaining positive child outcomes.
Although an in-depth literature connects relationship quality, family structure,
and child outcomes (e.g., Amato and Keith 1991; Booth and Amato 2001; Yu etal.
2010), this literature is limited in three major ways. First, many studies conceptual-
ize household conflict and relationship quality at the parent level (among parents
themselves); however, few consider relationship quality at the parent–child level.
Second, the influence of parent–child relationship quality on behavioral outcomes,
independent of parental conflict and family instability, has rarely been tested. Third,
these studies are almost solely focused on western nations where divorce and fam-
ily instability are relatively high. In order to assess the universality of such findings,
studies in non-western nations are needed.
Using data from the Korean Youth Panel Study (KYPS) gathered from youth
in South Korea (hereafter Korea), we estimate cross-sectional regression and lon-
gitudinal generalized estimating equation (GEE) models to examine the independ-
ent influences of family structure, parent–child relationship quality, and parental
conflict on internalizing and externalizing behaviors among Korean children. We
address the aforementioned gaps in the literature with the following questions: (1)
to what extent does parent–child relationship quality account for the association
between family structure and child behavior? and (2) how do parent–child relation-
ship quality, parental conflict, and family structure affect child behavioral problems,
independently of each other? Examining these patterns among Korean children also
provides a window into how these processes operate in contexts with very differ-
ent family structure instability patterns than those in the better-understood western
context.
Family Structure andChild Outcomes
A robust literature connects family structure to a variety of child outcomes, includ-
ing educational achievement and attainment, deviance and delinquency, and later
family formation (see Brown’s 2010 review). Of particular interest are children’s
mental and emotional health, as living in family structures with access to both
441
1 3
Family Structure andChild Well-Being inaNon-western Context:…
biological parents has a negative association with a number of health challenges,
such as eating disorders and attention deficit hyperactivity disorder (Carballo etal.
2013). Family structure appears to affect childhood outcomes, whether the biologi-
cal parent is removed by marital or relationship dissolution (Kim 2011), by death
(Amato and Anthony 2014), by military deployment (Gorman etal. 2010), by immi-
gration (Creighton etal. 2009), or by incarceration (Wildeman etal. 2013). In addi-
tion, children who lose access to a parent are often exposed to additional related
stressors, such as having to move from a shared family home (Pribesh and Downey
1999; McCulloch etal. 2000). These changes in family structure are associated with
problematic behavior in children (McLanahan and Sandefur 1994; Hoffmann 2002,
2017), patterns that are generally attributed to factors related to marital disruption
and family structure change, such as decreased familial resources (Thomson and
McLanahan 2012), stressors related to transition and instability (Waldfogel et al.
2010; Lee and McLanahan 2015), and selectivity issues related to family structure
and parenting (Dufur and Alexander 2017).
Beyond a child’s current family structure, transitions into and out of family struc-
tures during a child’s life are also critical for child outcomes. Research has found
that family instability can negatively affect child outcomes (Waldfogel etal. 2010),
perhaps most negatively for children transitioning out of two-parent families (Lee
and McLanahan 2015). While research has examined the effects of family stressors
associated with the removal of a parent, adding a parental figure causes stress on
children as well (Fomby and Cherlin 2007; Bachman etal. 2011). This may be a
result of residential moves or the negotiation of new family guidelines (Aquilino
2006).
The Role ofConict
It is also possible that family processes related to but distinct from family struc-
ture instability contribute to these negative outcomes. Beyond instability and fam-
ily structure, parental conflict may contribute to negative child behavior. Conflict
is an integral part of the negative impact of divorce on child outcomes (Amato and
Keith 1991; Amato etal. 1995). Marital conflict lowers closeness with parents, and
divorce lowers it further (Booth and Amato 2001). Parental conflict leading up to
and during marital disruption can also be extremely stressful for children (Cum-
mings etal. 1981; Cummings 1987), which is associated with a number of negative
outcomes.
Some researchers have concluded that the presence of conflict between parents is
so damaging that it outweighs the effects of family transitions. Supporting this conclu-
sion, several studies have found that children in intact but high-conflict family struc-
tures actually exhibit more behavior problems than children in high-conflict disrupted
family structures (Amato and Keith 1991; Cherlin etal. 1991; Jekielek 1998; Morrison
and Coiro 1999), perhaps because they are exposed to ongoing conflict. This research
suggests that there is a spectrum, or rank order, of well-being for children ranging from
two-parent intact families with low conflict as the best environment for a child, to two-
parent intact families with high-conflict families as the most problematic (Amato etal.
442
J.A.Jarvis et al.
1 3
1995). Therefore, while divorce is a negative experience for children, it might be more
desirable for children than remaining in an intact two-parent family with high levels
of conflict. Disruption may even be a form of stress relief for children in high-conflict
environments (Wheaton 1990).
Other studies, however, show that separation and divorce lead to problem behaviors
regardless of parental conflict (Morrison and Coiro 1999). These findings contribute
to a mixed picture of how family transitions and exposure to parental conflict work
independently and together to increase child problem behavior. Further complicating
our understanding, these mixed findings have focused almost entirely on western con-
texts with relatively high rates of family instability. The role of parental conflict in child
behavior problems in low-divorce settings remains unknown. It is possible that in such
contexts, parental conflict is even more influential on problem behavior, as the stress
relief associated with divorce (Wheaton 1990) is less available to troubled families.
On the other hand, divorce and living in non-traditional family structures in these con-
texts may be so non-normative that experiencing such transitions becomes especially
difficult.
Parent–Child Relationships
In addition to having a direct effect on children’s behavior, both family structure transi-
tions and exposure to parental conflict likely have a negative association with the rela-
tionships between parents and children. Shifts in parenting during high-conflict circum-
stances or during family dissolution, when parents may become distracted, depleted,
or preoccupied (Amato and Booth 1996), may damage pre-existing parent–child rela-
tionships. Parental conflict can also lead to parent–child relations changing as family
boundaries thin and children become too involved in adult issues (for a review, see Cox
and Paley 1997). Resentment about family dissolution or stressors generated by hav-
ing to adjust to new residential and family circumstances may damage communication
and trust between children and parents (Erel and Burman 1995; Kouros etal. 2014;
Braithwaite etal. 2015). Given these patterns, it is surprising that few representative
studies explore the parent–child relationship as a mechanism by which family structure,
instability, and parental conflict can affect child outcomes. Further, the patterns that are
well established have been observed in western societies that began experiencing low
levels of fertility and high levels of family dissolution and reconstitution, or a second
demographic transition, 40years ago (Lesthaeghe 2010). To our knowledge, no repre-
sentative studies have observed the influence of the parent–child relationship independ-
ent of family structure, instability, and parental conflict in a low-divorce context like
Korea where these demographic changes have begun to occur much more recently. It
is possible that parent–child relationships are shaped in different ways by parental con-
flict in societies where such conflict has little chance of being resolved through marital
dissolution.
443
1 3
Family Structure andChild Well-Being inaNon-western Context:…
Family Structure Change inNon‑western Nations
Examining family structure, parent–child relationship quality, and exposure to
parental conflict in contexts where a second demographic transition is still in early
stages can shed light on these questions. Research on family structure has predomi-
nantly focused on patterns and consequences of change in family structures in high-
income, western nations (Park 2008). And yet, family structures in high-income,
non-western countries not only exhibit different patterns and rates of change, but are
doing so in different social, cultural, and historical contexts. Therefore, observing
family structure in these different contexts allows us to examine the universality of
findings, such as the importance of family structure for children’s behavioral prob-
lems and the mechanisms by which family structure influences children.
Recent research examining family structures in high-income Asian nations where
intense and rapid social and demographic transitions have occurred more recently
(Park 2007, 2008, 2014; Chung and Emery 2010; Park and Raymo 2013), suggests
that the impact of family structure on children can both follow patterns observed
in western nations (Park 2008) and at times follow more regional patterns (Park
and Raymo 2013). The few inquiries focusing on the effects of family structure and
transition in these countries have largely examined educational outcomes; less is
known about how rapidly changing family structures in Asian countries affect devi-
ance and mental health issues. A single study of 454 Korean children found both
similarities to and differences from western patterns of effects of parental divorce
on child behavior problems (Chung and Emery 2010). One key difference was that
unlike children in the United States, where divorce had the largest effect on their
externalizing problems, the children in the Korean sample who had experienced
divorce had significantly more internalizing, but not externalizing, problems than
those in married families. While suggestive, these findings are not generalizable
and do not include investigations of parental conflict or parent–child relations that
might explain potential relationships between family structure and child behavior
problems. These are especially salient mechanisms in a country like Korea where
divorce has historically been much less common than in western countries, and is
thus less likely to provide a possible stress relief for high-conflict families (Amato
etal. 1995).
The Korean context is also one that experienced intense social change, as well as
substantial economic and educational growth, in the latter half of the twentieth cen-
tury. Such change may not only mean that divorce becomes more common as women
become increasingly educated and able to compete in the labor market, but also that
women may be able to use their income, employment, and education as protective fac-
tors against child behavior problems in the face of family dissolution. Yet cultural pres-
sures to maintain traditional and “successful” families—or at least the facades of suc-
cessful families—are still operative in non-western nations such as Korea, Taiwan, or
Singapore. While cohabitation has become a common alternative to marriage in many
western countries, the marginalization of cohabiting couples and infrequency of cohab-
itation in non-western countries (Yoo 2015; Raymo etal. 2015) has led to such reluc-
tance in the reporting of cohabitation that the Korean government has not been able to
obtain reliable statistics on cohabitation (Yoo 2015). And, even as the crude divorce
444
J.A.Jarvis et al.
1 3
rate in Korea began increasing in the 1990s to two divorces per 1000 (Lee 2006; Dom-
maraju and Jones 2011), traditional family values remain important, and the social
stigma of divorce and normative sanctions against divorce in Korea remains especially
high (Eun 2007; Chung and Emery 2010; Park and Raymo 2013). Not only is divorce
less common, but when divorce occurs in Korea it is also more likely to occur in the
second decade of marriage when Korean children are older (Lee 2006). As a result,
parents may more often remain in high-conflict marriages when compared to parents
in western countries. Under such circumstances, children in high-conflict, two-parent
families may experience more adverse effects on their behavior than children whose
parents have divorced or remarried (Jekielek 1998; Morrison and Coiro 1999), even as
these non-traditional family forms remain relatively uncommon in the culture. While
family relationship quality matters across contexts (Cox and Paley 1997), we suspect
that these measures may be even more important in a context where instability is rare
and, consequently, individuals in poor relationships are more likely to remain in those
relationships. Therefore, children living with parents who have a low-quality relation-
ship will be exposed to conflict on a much more regular basis.
This Study
Our study contributes to the literature in three major ways: (1) we test the influence
of parent–child relationship quality, a rarely-tested mechanism through which family
instability influences children, and examine the extent to which parent–child relation-
ship quality accounts for the association between family structure and child behavior,
(2) we consider the independent influence of parental conflict, parent–child relationship
quality, and family structure on child well-being to assess their importance and their
mediating roles, and (3) we examine child behavioral problems in Korea, a low-insta-
bility context where two-parent families predominate. Past research in western con-
texts (Amato and Keith 1991; Cherlin etal. 1991; Jekielek 1998; Morrison and Coiro
1999) suggests that with so little family instability, parental conflict may be especially
impactful on children’s behavioral problems. As a result, we expect parental conflict to
negatively affect behavioral problems in Korean children. Also, with transitions being
unusual in the Korean context and two- parent, married family structures so normative,
we expect the impact of family structure transitions or living in a non-normative family
structure to be associated with worse child behavioral problems even after controlling
for parental conflict, parent–child relations, family resources, and selectivity factors.
Finally, in the Korean context where family structure change is less common than in
western nations, we expect the effects of family structure on child behavioral problems
to be mediated by family relationship measures such as parent-child relations or paren-
tal conflict.
445
1 3
Family Structure andChild Well-Being inaNon-western Context:…
Data andMethods
We use data from the Korea Youth Panel Survey (KYPS). KYPS is a longitudinal
cohort study aimed at understanding the career development and deviant behav-
iors of adolescents in Korea. The survey applied a three-stage stratified sampling
design—sampling from school districts, schools, and classes—to obtain a nation-
ally representative sample. It includes two cohorts. The younger cohort, whom we
examine here, were followed from fourth to eighth grade (2003–2008), creating five
waves of data. Answers to survey questions were self-reported with the help of an
interviewer (National Youth Policy Institute 2012).
The KYPS data are exceptionally well-suited for examining family structure,
family dynamics, and the well-being of children. Most importantly, KYPS includes
information regarding children’s internalizing and externalizing behaviors, par-
ent–child relationship quality, and parental conflict. This unique information allows
us to explore the impact of relationship quality in a context with high family stability
and low rates of divorce.
Sample
Participants were interviewed once a year for 5years, totaling five waves of data. For
our cross-sectional models, we include all participants who were in the sample at
the fifth and final wave (N = 2448). Our longitudinal models include all participant-
waves (N = 14,220 participant-waves across N = 2844 children). Most of the sampled
children, 86%, remained in the study for all five waves. Although there is no way to
entirely avoid the effects of attrition, weighting is a commonly used strategy (Vande-
casteele and Debels 2006). We employ weighting here (described below).
In order to more fully observe the effect of attrition, we compared children in the
first wave of data (the “original” sample) to children that remained in the fifth and
final wave (the analytic sample). Demographically, our analytical sample and the
original sample had few differences. Most children resided in a two-parent, stable
household (94.8% in the analytic sample and 94.7% in the original sample). Further,
the analytic and original samples were similar in the percentage of children who
were female (46.8% in the analytic sample and 46.4% in the original sample), the
average age (9.9 for both samples, SD = 0.4), and average number of siblings (1.1
for both samples, SD = 0.6). Household monthly income was very similar across
samples (around $2700 in both samples) as well as mother’s employment (around
50% employed full-time in both samples) and father’s employment (around 96%
employed full-time in both samples). Finally, in both samples similar numbers of
mothers had a high school education (61% of the original sample and 63% of the
analytic sample) or a college degree (25% in the original sample and 22% in the
analytic sample).
There were very little missing data overall. Among the participants who were in
the sample at the fifth wave, some variables had no missing data (child age, and
child gender), other variables had fewer than 5% of values missing (parental conflict,
parent–child relationship quality, household monthly income, mother’s education,
446
J.A.Jarvis et al.
1 3
and father’s education), and three variables had fewer than 9% of values missing
(respondent family structure, mother’s employment and father’s employment). In
order to maintain the entire sample across models, we employed multiple impu-
tation. Due to issues with multicollinearity, imputation was performed on indices
rather than on items in the indices. Imputation of new data sets can be limitless;
however, 20 iterations is an acceptable threshold (Johnson and Young 2011). We
performed 25 iterations to ensure confidence in our results. We follow von Hippel
(2007) in excluding cases on which the dependent variables (internalizing and exter-
nalizing) were imputed after the imputations were completed.
Measures
Dependent Variables
Our two dependent variables are internalizing and externalizing behaviors. We
measure internalizing behaviors with a standardized index including the follow-
ing items: “sometimes I feel extremely anxious with no apparent reason,” “I worry
about everything,” “sometimes I think that I am a bad person,” “sometimes I think
that I am a worthless person,” “sometimes I feel extremely lonely with no apparent
reason,” “sometimes I feel extremely sad and gloomy with no apparent reason,” “I
am not interested in anything,” “I try to assuage my anger whenever I get angry,”
and “sometimes I feel suicidal with no apparent reason.” Participants responded to
these statements on a five-point Likert scale of “strongly disagree,” “disagree,” “nei-
ther disagree nor agree,” “agree,” and “strongly agree.” Responses were averaged
across all items and then standardized to ensure a mean of 0 and standard deviation
of 1. We use this index (α = 0.82) as a continuous measure of general internalizing
behavior problems. Values greater than zero indicate more internalizing behaviors
while values less than zero indicate fewer internalizing behaviors.
We measure externalizing behaviors using this same process. The externalizing
scale includes the following items: “I may hit other people when I feel annoyed,”
“I will hit back at a person who hits me,” “I fight more frequently than others do,”
“sometimes I can’t suppress an impulse to hit other people,” “I am often seized by
an impulse to throw an object whenever I get angry,” and “I consider myself as an
explosive soon to be blown off.” Like the internalizing measure, we standardized the
externalizing behavior problem index (α = 0.86) and used it as a continuous measure
of general externalizing behavior problems. Values greater than zero indicate more
negative externalizing behaviors while values less than zero indicate fewer external-
izing behaviors.
Independent Variables
Our key independent variables are family structure, parent–child relationship qual-
ity, and parental conflict. In each wave, we measure family structure by construct-
ing three categories that reflect both the number of biological parents available
to the child and the number of family transitions (i.e., instability) a child likely
447
1 3
Family Structure andChild Well-Being inaNon-western Context:…
experienced: living with two biological parents, living with a single parent (either
mother or father), and living in a two-parent family that includes a non-biological
parent (generally created by a remarriage). We also measure instability by exam-
ining whether family structure has changed from the structure indicated in previ-
ous waves. Because of the low levels of instability and non-marital births in Korea
(Statistics Korea 2018b), most children living with a single parent have experienced
one transition, and children living in an unstable, two-parent household with a non-
biological parent most likely experienced two or more transitions. Due to small
sample sizes in categories representing respondents living with a single father, we
did not distinguish further based on gender of single parent (all included in the sin-
gle-parent category). The unstable, non-biological two-parent family category also
includes a very small number of cases where biological parents married during the
survey years; however, almost all of the cases in the unstable two-parent family cat-
egory represent a remarriage of a biological parent to a non-biological parent. We
include family structure information from Wave 5 for cross-sectional models and
from Waves 1–5 in longitudinal models.
We also measure the relationship quality between the participants and their par-
ents in each wave. We refer to this as parent–child relationship quality. This index
includes the items “my parents and I have frequent conversations,” “my parents
and I candidly talk about everything,” “my parents always treat me with love and
affection,” and “my parents and I try to spend much time together.” Participants
responded to these statements with a five-point Likert scale from “very untrue,”
coded as 1, to “very true,” coded as 5. Responses to these items were averaged to
create the relationship quality index (α = 0.76), which was included as a continuous
measure. A value of one indicates the lowest relationship quality and a value of five
indicates the highest relationship quality.
Finally, we created a measure of parental conflict in each of the five waves. Past
literature suggests that parents’ relationship quality, often measured as the degree
of conflict, has important implications for children’s well-being and development
(e.g., Margolin 1988, Fincham and Osborne 1993, Katz and Gottman 1993, Katz
and Gottman 1994). To measure parental conflict, children were presented with the
following statements and asked how true these statements were from “very untrue,”
coded as 1, to “very true,” coded as 5: “I frequently see my parents badmouth or
swear at each other” and “I frequently see one of my parents hit the other one.” The
responses to these questions were averaged to create an index (α = 0.81), which was
included as a continuous variable. While interparental conflict is ideally assessed
using multiple measures from both parents and children, the two measures we used
were the only parental conflict measures available in the KYPS data. We acknowl-
edge that our two measures capture intense and relatively extreme forms of conflict;
however, these types of conflict are more common in this context than in contexts
like the United States (Ko and Park 2020; Park etal. 2017; Han etal. 2017; Kim
etal. 2009). As a result, both the possibility that more extreme measures of conflict
might be especially associated with child outcomes and the possibility that the pres-
ence of what westerners would consider extreme forms of conflict might have differ-
ent associations with child outcomes in Korea exist. Therefore, examining this type
of serious parental conflict and its relation to family structure and child well-being is
448
J.A.Jarvis et al.
1 3
useful and relevant. To distinguish the relative severity of the interparental conflict
measures we have available here, we refer to this index as serious parental conflict
throughout the rest of the paper. We include information on both parent–child rela-
tionship and serious parental conflict from Wave 5 for cross-sectional models and
from Waves 1–5 in longitudinal models.
Control Variables
We also account for several factors that influence family structure, family dynam-
ics, and internalizing and externalizing behaviors. First, we account for characteris-
tics of the focal child. Due to past literature finding significant gender differences in
both externalizing and internalizing behaviors (Leadbeater etal. 1999), we account
for the child’s sex (1 female, 0 male). We also include the child’s age (in years)
and number of siblings (a count variable). Second, we include characteristics of the
household. We include household monthly income as a continuous measure in dol-
lars. In light of findings suggesting that less-educated mothers across many settings
are more likely to divorce (Park and Raymo 2013; Pew Research 2015; Dufur and
Alexander 2017), and in order to examine the role of selectivity on family structure
change and subsequent child behavioral problems, we account for both mothers’ and
fathers’ education. Both are coded as categorical variables with the categories “less
than a high school degree”, “high school”, “some college”, and “college or higher.”
Finally, we measure both parents’ employment status as a categorical variable that
includes the categories “unemployed”, “employed part-time”, and “employed full-
time.” With the exception of child gender, which was asked in Wave 1, we include
information on each of these control variables from Wave 5 for cross-sectional
models and from Waves 1–5 in longitudinal models. Maternal age at birth and fam-
ily structure at birth are commonly controlled for as measures of selectivity when
examining the impact of family structure on child outcomes; however, neither of
these measures is available in the KYPS data. From Korean population data, we
find that Korean average maternal age at first birth is 31.6years old (Statistics Korea
2018a), the highest in the OECD (2016), but we cannot tell from the KYPS data
whether single or remarried mothers might have given birth at younger ages. With
regards to family structure at birth, unwed childbirth remains especially uncommon
in Korea, as 98% of births in 2017 were to married parents (Statistics Korea 2018b),
also the highest among OECD countries (OECD 2018b). This suggests that despite
the lack of direct measures in our data, the vast majority of Korean children in our
sample were born to married biological parents, with mothers who were older than
most mothers in western nations.
Our data are not without limitations. We are not always able to untangle exactly
whom respondents are referring to when they are asked about “parents.” For example,
when asked about paternal education, it is unclear if children in stepfamilies were refer-
ring to a stepfather or a non-residential biological father when answering. Respondents
were asked to consider their parent in answering and then, if they did not live with a
parent, a legal guardian. However, all children in the KYPS data lived with at least
one biological parent. Similarly, in the parent–child relationship quality measure, we
could not differentiate between how respondents feel about their mothers and fathers. A
449
1 3
Family Structure andChild Well-Being inaNon-western Context:…
child could have a high-quality relationship with his or her mother but not with his or
her father, a pattern that is indistinguishable in our data. Further, information on non-
resident parents could be unavailable for children living in single-parent households,
though we found many respondents in single-parent families provided information on
both parents. To test this latter concern, we observed patterns in missing data. Although
relatively few values were missing generally, children in single-parent families had
higher levels of missing data on father’s employment status (15–45% across waves)
than children in two-parent families (generally less than 10%). In order to keep these
children in the sample and utilize father’s employment status as a control variable, we
imputed this information as described above. We also estimated models without imput-
ing data and models without including father’s employment as a control (results avail-
able upon request). Both approaches garnered substantively similar results to the results
presented here.
Analytic Strategy
To examine the relationships between family structure, relationship quality, and inter-
nalizing and externalizing behaviors, we first report descriptive statistics (Table1). We
then examine child behavioral problems cross-sectionally in the fifth and final wave of
data (Table2). For these models, other than gender, which is measured only in Wave 1,
all variables were measured in the fifth wave of data. A cross-sectional approach allows
us to focus on the context and circumstances the children are in at a particular moment
and how various factors at this point in time may be associated with their behavioral
problems. Due to the continuous nature of our dependent variables, we utilize ordi-
nary least squares (OLS) regressions in our cross-sectional analysis. Next, we employ
generalized estimating equation (GEE) models in order to utilize all five waves of data
while still accounting for correlation among repeated or clustered measures (Zorn
2001). GEE models allow us to assess variability in our dependent variables and how
this variability corresponds to our independent variables within individuals across time.
This approach allows us to use information across all five waves of data, accounting for
developmental differences as children age. Other than gender, which is measured only
in Wave 1, each variable included in our models was measured at each wave of data
and allowed to vary across time in the longitudinal models.
In each regression table, our first models include family structure, the primary inde-
pendent variable, alone. The second models add parent–child relationship quality and
our measure of serious parental conflict. Our final models include all the control varia-
bles. All models are weighted using survey weights (wave five survey weights for OLS
models and wave one survey weights for GEE models). All models are estimated with
robust standard errors.
450
J.A.Jarvis et al.
1 3
Table 1 Characteristics of children aged 13–15 (grade eight), means and percentages, Korea, 2008
(N = 2448)
Data: Korean Youth Panel Survey, Cohort 2, Wave 5
Mean or percent Standard dev Minimum Maximum
Family structure
Biological, two-parent, stable 88.44%
Single parent 6.68%
Non-biological/unstable two-parent 4.88%
Parent–Child Relationship Index (total)
a
3.73 0.69 1.00 5.00
Often talk with parents 3.61 0.96 1.00 5.00
Speak honestly with parents 3.29 1.07 1.00 5.00
Feels loved by parents 3.76 0.90 1.00 5.00
Spends time with parents 3.47 0.93 1.00 5.00
Parental Conflict Index (total)
a
1.75 0.92 1.00 5.00
Parents do not fight often (verbally) 1.92 1.07 1.00 5.00
One parent does not hit the other 1.58 0.92 1.00 5.00
Child characteristics
Female 46.77%
Age 13.85 0.36 13.00 15.00
Number of siblings 1.15 0.58 0.00 5.00
Household characteristics
Household monthly income 3117.88 1637.51 0.00 27,000.00
Mother’s education
Less than high school 7.18%
High school 62.96%
Some college 8.12%
College or higher 21.75%
Father’s education
Less than high school 5.95%
High school 44.83%
Some college 10.08%
College or higher 39.13%
Mother’s employment
Full time 48.92%
Part time 6.11%
Unemployed 44.97%
Father’s employment
Full-time 95.04%
Part-time 1.45%
Unemployed 3.51%
451
1 3
Family Structure andChild Well-Being inaNon-western Context:…
Table 2 OLS regression predicting internalizing and externalizing behavior scales, Korea (2008), coefficients presented (N = 2448)
Model 1
(internalizing)
Model 2
(externalizing)
Model 3
(internalizing)
Model 4
(externalizing)
Model 5
(internalizing)
Model 6
(externalizing)
Family structure
a
Single parent 0.22* 0.14† 0.16* 0.09 0.15
†
0.11
(0.09) (0.08) (0.08) (0.08) (0.08) (0.08)
Two-parent, unstable 0.37*** 0.27** 0.25** 0.18
†
0.25** 0.19
†
(0.09) (0.10) (0.09) (0.10) (0.09) (0.10)
Parent–child relationship quality − 0.31*** − 0.17*** − 0.32*** − 0.17***
(0.03) (0.03) (0.03) (0.03)
Parental conflict 0.19*** 0.19*** 0.20*** 0.19***
(0.02) (0.03) (0.02) (0.03)
Child characteristics
Female 0.39*** 0.02
(0.04) (0.04)
Age 0.06 0.02
(0.05) (0.05)
Number of siblings − 0.01 0.00
(0.03) (0.04)
Household characteristics
Household income
b
− 0.0001 0.02
(0.01) (0.01)
Mother’s education
c
High school 0.10 0.09
(0.11) (0.10)
Some college 0.21
†
0.13
(0.13) (0.11)
College or higher 0.15 0.18
(0.12) (0.11)
452
J.A.Jarvis et al.
1 3
Table 2 (continued)
Model 1
(internalizing)
Model 2
(externalizing)
Model 3
(internalizing)
Model 4
(externalizing)
Model 5
(internalizing)
Model 6
(externalizing)
Father’s education
c
High school − 0.13 − 0.12
(0.10) (0.10)
Some college − 0.14 − 0.15
(0.12) (0.13)
College or higher − 0.16 − 0.22*
(0.11) (0.11)
Mother’s employment
d
Part-time 0.08 0.07
(0.07) (0.08)
Full-time 0.07
†
0.13**
(0.04) (0.04)
Father’s employment
d
Part-time 0.04 − 0.12
(0.15) (0.18)
Full-time 0.06 − 0.05
(0.09) (0.12)
Data: Korean Youth Panel Survey, Cohort 2
†
p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001
Standard errors in parentheses; estimates weighted using wave five cross-sectional weights
a
Reference: biological, stable
b
In thousands of dollars
c
Reference: less than high school
d
Reference: not employed
453
1 3
Family Structure andChild Well-Being inaNon-western Context:…
Results
Approximately 88% of our sample are living with their biological mother and
father who are married. Substantially fewer children live with a single-parent
(6.7%) or in an unstable, two-parent family (4.9%) (see Table1). These family
patterns, signaling high levels of stability, are consistent with past literature on
Korea (Park 2008) and demonstrate how different the Korean context is from
western contexts where children’s family instability is higher on average. Because
of this low instability and, consequently, low variation in family structure, par-
ent–child relationship quality and serious parental conflict indices may become
increasingly important characteristics. Using these indices, we find that children,
on average, report relatively high relationship quality with their parents (an aver-
age of 3.7 with a maximum of 5.0, SD = 0.7) and low levels of serious paren-
tal conflict (an average of 1.8, SD = 0.9). We do not present means and standard
deviations from the externalizing and internalizing indices in Table 1 because
they are standardized indices, each with a mean of 0 and standard deviation of 1.
Turning to child characteristics, almost half of our sample is female (47%).
The respondents are 14 years old on average (SD = 0.4) and have one sibling
(SD = 0.6). In terms of household characteristics, the average household monthly
income is USD 3,117.88 with a fairly wide distribution. Korea is widely known
for its highly educated population (OECD 2018c) and strict gendered norms
(Eun 2007). This is reflected in our data. In our sample, very few parents have
not graduated from high school (7% of mothers and 6% of fathers), while about
one-third of mothers and half of fathers have at least some college education or
higher. Further, while only 49% of mothers report working full-time, almost all
fathers (95%) are working full-time.
Findings from our cross-sectional OLS models show that family structure and
relationship quality are independently and significantly associated with children’s
behavior issues (see Table2). Models 1 and 2 show results for internalizing and
externalizing behaviors, respectively, using only family structure as an independ-
ent variable. Children living in single-parent families experience significantly
more internalizing (b = 0.22, p < 0.05) and externalizing behaviors (b = 0.14,
p < 0.10) than children living in two-parent, stable families, though the coefficient
is marginally significant for externalizing behavior problems. Similarly, children
living in two-parent, unstable families report significantly more internalizing
(b = 0.37, p < 0.001) and externalizing (b = 0.27, p < 0.01) behaviors than children
living in two-parent, stable families. The coefficients for two-parent, unstable
families are much larger than for single-parent families (over 1.5 times larger for
internalizing and almost twice as large for externalizing), suggesting that greater
exposure to instability is detrimental to child behavior.
After introducing parent–child relationship quality and serious parental con-
flict, family structure remains significantly associated with children’s internaliz-
ing and externalizing behaviors (see Table2, Models 3 and 4). Children living
with a single parent experience 0.16 of a standard deviation more internaliz-
ing behaviors than children living with two biological parents, and this finding
454
J.A.Jarvis et al.
1 3
is statistically significant (p < 0.05). Similarly, children living in a two-parent,
unstable household experience a fourth of a standard deviation (b = 0.25) more
internalizing behaviors than children living with two biological parents, and this
finding is statistically significant (p < 0.01).
In Models 3 and 4, parent–child relationship quality is also significantly and
negatively related to both internalizing (b = − 0.31, p < 0.001) and externalizing
(b = − 0.17, p < 0.001) behaviors. That is, children who report higher-quality rela-
tionships with their parents experience fewer internalizing and externalizing behav-
iors. Like parent–child relationship quality, serious parental conflict is also signifi-
cantly related to children’s internalizing and externalizing behaviors. Children who
report higher levels of serious parental conflict report significantly more internaliz-
ing (b = 0.19, p < 0.001) and externalizing (b = 0.19, p < 0.001) behaviors than chil-
dren who report lower levels of parental conflict. These coefficients remain similar
in size and significance accounting for child and household characteristics (Table2,
Models 5 and 6).
To further explore whether relationship quality and serious parental conflict
might explain the effects of family structure on child outcomes, we estimated formal
mediation tests across all 25 multiply imputed iterations of our data (xtsgmediate in
Stata 15) (results available upon request). We found that in the OLS cross-sectional
models, parent–child relationships did not significantly mediate the relationship
between family structure and child internalizing and externalizing behavioral prob-
lems, and evidence of serious parental conflict mediating these effects was weak
and inconsistent across iterations. This suggests that while parent–child relationship
and serious parental conflict are useful predictors of child behavior problems, fam-
ily structure exerts an effect independently of the ways it may affect relationships
between parents and between parents and children.
Few of our control variables are significantly related to internalizing or external-
izing behaviors. As found in prior studies, girls report significantly more internaliz-
ing behaviors than boys (b = 0.39, p < 0.001) (e.g., Leadbeater etal. 1999). Further,
children whose mothers work full-time experience significantly more internalizing
(b = 0.07, p < 0.10) and externalizing behaviors (b = 0.13, p < 0.01) compared to
children whose mothers do not work, albeit, the findings for internalizing behaviors
are only marginally significant. Considering Korea’s unusually large gender wage
and employment gaps (OECD 2017), mothers full-time work is still considered non-
normative and likely reflective of economic disadvantage.
To ensure confidence in our results, as well as to fully utilize all of the data availa-
ble in KYPS, we further tested these relationships using longitudinal models. Unlike
OLS models, GEE models allow for examinations of observations over time with
observations clustered within individuals. Longitudinal data such as KYPS include
repeated measures of behavior problems, relationships, and other key variables as
respondents age across the survey waves. These data allow us to better account for
important differences as children develop, such as respondent maturation. We use
GEE models to better account for unobserved heterogeneity and correlation between
repeated measurements within individual children.
Results from the GEE models largely support the conclusions from the OLS
models (see Table 3): family structure, parent–child quality, and serious parental
455
1 3
Family Structure andChild Well-Being inaNon-western Context:…
Table 3 GEE regression predicting internalizing and externalizing behavior scales, Korea (2008), coefficients presented (N = 14,220)
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
(internalizing) (externalizing) (internalizing) (externalizing) (internalizing) (externalizing)
Family structure
a
Single parent 0.10
†
0.14** 0.05 0.10
†
0.04 0.12*
(0.05) (0.05) (0.05) (0.05) (0.05) (0.05)
Two-parent, unstable 0.15* 0.14* 0.09 0.09 0.09 0.12
†
(0.06) (0.06) (0.06) (0.06) (0.06) (0.06)
Parent–child relationship quality − 0.25*** − 0.13*** − 0.25*** − 0.13***
(0.01) (0.01) (0.01) (0.01)
Parental conflict 0.19*** 0.16*** 0.20*** 0.16***
(0.01) (0.01) (0.01) (0.01)
Child characteristics
Female 0.19*** − 0.07**
(0.03) (0.03)
Age − 0.02*** − 0.01*
(0.01) (0.01)
Number of siblings 0.02 0.03
(0.02) (0.02)
Household characteristics
Household monthly income
b
− 0.01* − 0.003
(0.001) (0.001)
Mother’s education
c
High school 0.03 0.08
(0.06) (0.06)
Some college 0.02 0.05
(0.07) (0.07)
College or higher 0.04 0.08
(0.07) (0.07)
456
J.A.Jarvis et al.
1 3
Table 3 (continued)
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
(internalizing) (externalizing) (internalizing) (externalizing) (internalizing) (externalizing)
Father’s education
c
High school − 0.08 − 0.05
(0.06) (0.06)
Some college − 0.07 − 0.03
(0.08) (0.07)
College or higher − 0.11 − 0.03
(0.07) (0.07)
Mother’s employment
d
Part-time 0.07
†
0.01
(0.04) (0.05)
Full-time 0.07** 0.02
(0.02) (0.02)
Father’s employment
d
Part-time 0.04 − 0.09
(0.09) (0.09)
Full-time 0.03 − 0.01
(0.07) (0.07)
Data: Korean Youth Panel Survey, Cohort 2
†
p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001
Standard errors in parentheses; estimates weighted using wave one cross-sectional weights
a
Reference: biological, stable
b
In thousands
c
Reference: less than high school
d
Reference: not employed
457
1 3
Family Structure andChild Well-Being inaNon-western Context:…
conflict independently and significantly relate to children’s internalizing and exter-
nalizing behaviors. In Models 1 and 2, the results show that children living in sin-
gle-parent families experience more internalizing behaviors (b = 0.10, p < 0.10) and
significantly more externalizing behaviors (b = 0.14, p < 0.05) than children living in
two-parent, stable families. Similarly, children living in two-parent, unstable fami-
lies experience significantly more internalizing (b = 0.15, p < 0.05) and externalizing
behaviors (b = 0.14, p < 0.05).
After introducing parent–child relationship quality and serious parental conflict
(see Table3, Models 3 and 4), living in a single-parent family is no longer signifi-
cantly related to higher levels of internalizing behaviors (b = 0.05) and only mar-
ginally related to externalizing behaviors (b = 0.10, p < 0.10). Further, living in a
two-parent, unstable family is also no longer significantly related to internalizing or
externalizing behaviors. One possible explanation for the lack of significance is the
absence of variability in family structure and instability across waves. Longitudinal
tests of within-person variation in behavior problems are stricter, more conserva-
tive tests, explaining variation in variables that reflect narrower distributions than
the cross-sectional tests we report above. Alternatively, however, it is possible that
parent–child relationship quality explained variation in this relationship. Using a
formal mediation test with the xtsgmediate program in Stata 15, we found that par-
ent–child relationship quality significantly mediated the relationship between family
structure and internalizing and externalizing behaviors in the GEE models. How-
ever, we found no evidence that serious parental conflict significantly mediated this
relationship.
Mirroring the findings in OLS models, parent–child relationship quality and seri-
ous parental conflict are significantly related to children’s internalizing and external-
izing behaviors (Models 3 and 4). Children who report higher-quality relationships
with their parents experience significantly fewer internalizing (b = − 0.25, p < 0.001)
and externalizing behaviors (b = − 0.13, p < 0.001) than children who report lower-
quality relationships with their parents. Substantively similar findings emerge for
serious parental conflict. Children who report higher levels of serious parental con-
flict experience significantly more internalizing (b = 0.19, p < 0.001) and external-
izing behaviors (b = 0.16, p < 0.001) than children who report lower levels of serious
parental conflict. These relationships remain significant after accounting for child
and household characteristics (see Table3, Models 5 and 6).
In the full GEE models, several child and household variables are significantly
related to internalizing and externalizing behaviors. Similar to the cross-sectional
OLS models, female children experience significantly higher increases in internal-
izing (b = 0.19, p < 0.001) and significantly lower increases in externalizing behav-
iors (b = − 0.07, p < 0.01) than boys, matching past research (Leadbeater etal. 1999).
Older children experience significantly fewer internalizing (b = − 0.02, p < 0.001)
and externalizing behaviors (b = − 0.01, p < 0.05) than younger children. Increases
in household monthly income are negatively and significantly related to internal-
izing behaviors (b = − 0.01 in thousands, p < 0.05). And children whose mothers are
employed full-time experience significantly more internalizing behaviors (b = 0.07,
p < 0.01) than children whose mothers are not employed.
458
J.A.Jarvis et al.
1 3
Sensitivity Analyses
We estimated our models under a number of additional specifications to ensure
our findings were not statistical artifacts, especially since we use data that has
a relatively small proportion of children in non-traditional family structures: (1)
we used the number of family transitions a child had experienced during the sur-
vey period as our primary independent variable instead of family structure, (2)
we estimated our models under before imputing missing data and compared them
to post-imputation results, and (3) because our parent–child relationship quality
measures were ambiguous in asking about parents in general, rather than speci-
fying which parent, we stratified models by family structure; in these stratified
models, family structure was excluded as an independent variable.
Results across these specifications (available upon request) were very simi-
lar to the results we present here. Across all models, children living in unstable
two-parent families, or in families who had experienced two or more transitions,
experienced more internalizing and externalizing behaviors than children living
with two biological parents, independent of parent–child relationship quality and
serious parental conflict. This finding was significant in every model except when
estimating the number of family transitions using a GEE model; however, that
result is also in line with the way family structure is no longer significant in the
full GEE models reported above, again providing reassurance that those findings
are not statistical artifacts. Estimating the number of family transitions as the pri-
mary independent variable using the OLS models garnered similar significant
results to the models we report above. Further, findings for parent–child relation-
ship quality and serious parental conflict remained significant and in the expected
directions across children in all family structures. Overall, then, our sensitivity
tests suggest that we can be confident in our key findings: Korean children who
live in non-traditional family structures or who transition into those structures,
who experience serious interparental conflict in their homes regardless of fam-
ily structure, or who have worse relationships with their parents experience more
behavior problems and increasing behavior problems over time.
Discussion
In this study, we make three important contributions. First, we examine a relatively
neglected mechanism through which family change may affect children: parent–child
relationship quality. Second, we examine the independent influence of family struc-
ture and instability, parent–child relationship quality, and parental conflict on child
outcomes. And third, we use data from Korean families to examine the universality
of research related to family structure in a high income, non-western context that has
more recently experienced a second demographic transition.
We find several notable results. As expected, we find that Korean families in
our study were much less likely to experience marital disruption than families
459
1 3
Family Structure andChild Well-Being inaNon-western Context:…
in high-income, western nations. And yet, despite this comparative stability in
family structure, we see that family structure is meaningful for Korean children.
Being in a stable, married, two biological parent family generally led to fewer
behavioral problems among Korean children, even after accounting for both fam-
ily relationship quality and a wide range of demographic variables.
We also included mechanisms through which family structure may be affecting
children’s behavior. We find that lack of exposure to serious parental conflict, as
well as positive parent–child relations lead to fewer behavioral problems in Korean
children, net of each other. These mechanisms are important to this analysis because
of the low levels of divorce in general in Korea, and the unusual tendency for Kore-
ans to divorce in the second decade of marriage when children are older (Lee 2006).
As a result, Korean children in the age range we examined are much more likely to
be in a family where married parents may have conflict without the possible resolu-
tion provided by a divorce.
Serious forms of parental conflict have a significant impact on children’s inter-
nalizing and externalizing behavioral problems, even when controlling for family
resources, parental characteristics, and family structure. Korean children have fewer
problems when their parents fight less. Still, our models generally suggest that expe-
riencing family transitions is problematic for Korean children net of serious paren-
tal conflict. These results support findings from research on western families where
both family structure change and parental conflict affected children’s behavior prob-
lems (Jekielek 1998; Riggio 2004; Yu etal. 2010), strengthening claims for the uni-
versality of such patterns.
We find similar results when examining the impact of parent–child relations on
Korean children’s problem behaviors. Positive parent–child relations have a protec-
tive effect against behavior problems in Korean children, even when controlling for
conventional explanations such as depleted resources or parental characteristics.
Further, we found some evidence that parent–child relations significantly mediated
the relationship between family structure and child well-being, validating its impor-
tance. However, even amidst this mediation, family structure remained significant.
Therefore, our results support the notion that family structure and family dynamics
are important correlates for child well-being.
These findings suggest that more efforts should be made to target the health of
Korean children in all non-traditional families, but particularly children in blended
families, as the process of separating adults from and adding adults to children’s
lives is associated with behavioral problems. In a nation like Korea where the vast
majority of children are raised in two-parent married families, marriage norms are
powerful and intact, and two-parent families are held in high regard. And yet, as the
number of divorces in Korea has increased, so too has the number of remarriages
(Statistics Korea 2013). As we see from this research, the remarriage process can be
detrimental to the children experiencing these changes, perhaps because these fam-
ily transitions are still culturally viewed as non-normative. Social programs targeting
children and parents considering remarriage may prove especially beneficial.
One potential pathway to aid parents and children in such circumstances is pro-
grams that would facilitate positive parent–child relationships. While the effects
of positive parent–child relationships did not completely overcome the detrimental
460
J.A.Jarvis et al.
1 3
effects of family structure change, we found that such positive intrafamilial dynam-
ics did reduce Korean children’s behavioral problems, even for children in non-tra-
ditional family situations. Because of the high-stress and intense educational system
in Korea, school-aged children spend much of their time in various forms of edu-
cational training, and not with their families. Therefore, programs emphasizing the
importance of spending time together and helping to facilitate parent–child relation-
ships through activities like frequent and candid conversations may help children
navigate family structure change more successfully. Lastly, when children perceived
low levels of serious parental discord, they were much less likely to report behav-
ioral problems. Helping parents in all family structures navigate their own relation-
ships and reduce serious conflict could help their children’s emotional well-being
and behavior.
Finally, this research captures the impact of family structure on children’s behav-
ior and the mechanisms through which family structure may be affecting children’s
behavior in a high-income, non-western nation that is still in the earlier stages of a
second demographic transition. This means that these results are capturing change
in important social behaviors and the ramifications of these behaviors (in this case,
child behavioral problems) in a context at the cusp of more dramatic social changes.
Based on what western nations experienced decades ago, this may lead to further
normalization of the experiences of being in a non-traditional family structure or
experiencing family structure change; on the other hand, we may see the emergence
of uniquely Asian or Korean versions of this change where historical and cultural
influences continue to stigmatize non-traditional family structures or mitigate the
trend of cohabitation before or instead of marriage.
This study was subject to several limitations. First, the data do not include infor-
mation about family structure at birth, instead beginning in the fourth grade of ele-
mentary school. This prevents an examination of important factors that may occur
before children are in school and are more dependent on their parents. In a society
only beginning to experience the second demographic transition, family structure
changes in the “tender years” (Amato and Anthony 2014) of early childhood may
exert stronger or longer lasting negative effects on children. This also limits our abil-
ity to test whether there are associations between child outcomes and the amount of
time spent in particular family structures. Second, our parent–child quality measures
are somewhat ambiguous as the questions used do not specify which parent the child
is reporting about (the questions simply refer to “parents”). This limits the conclu-
sions we can make about specific parents (e.g., fathers versus mothers, stepparents,
etc.). In order to test the robustness of this finding, we estimated models stratified
by union status, rather than including family structure as an independent variable,
to assess whether findings were stronger for children in households with both bio-
logical parents, who experienced less ambiguity around this question. We found
that parent–child relationship quality and serious parental conflict were significantly
related to internalizing and externalizing behaviors, in the expected directions,
across all family structures. Therefore, despite this ambiguity, children’s perceptions
of their relationship with their parents, however they may define them, are impor-
tant for their well-being. Finally, measures of parental conflict in this data are lim-
ited to two items and represent severe behaviors, perhaps capturing some elements
461
1 3
Family Structure andChild Well-Being inaNon-western Context:…
of abusive relationships. While abusive behavior can be part of parental conflict,
conflating arguments or discordance with abuse is problematic. To our knowledge,
these are the best data to address both parental conflict variables and a context in
which divorce is still relatively uncommon, but data with more nuanced measures
of parental conflict could help examine the degree to which common forms of con-
flict both influence child behavior and explain potential effects of family structure or
transitions.
Overall, our results provide a compelling new direction for future research. More
studies should incorporate measures of parent–child relationship quality, as it influ-
ences child well-being above and beyond both parental conflict and family structure.
Reasons for why this happens could be explored more fully. Further, while we found
that family structure, serious parental conflict and parental-child relationship quality
had independent impacts on child behavioral problems, parent–child relationships
may be an important mechanism through which family structure affects child behav-
ior. These patterns represent both similarities and differences between the high-
income, non-western context we study here and the western contexts upon which
the bulk of the family literature is based. Instead of conceptualizing parent–child
relationship quality as a mechanism explaining away the effects of family context,
more research is needed to explore this concept as both an independent influence
on children across family structures, and a way that family structure change affects
child outcomes.
Acknowledgements We want to acknowledge the contributions of John Hoffmann, Alyssa Alexander,
Carolina Otero, and Can Cheng from Brigham Young University. This research was supported by funding
from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD)
to the Population Research Institute at The Pennsylvania State University for Population Research Infra-
structure (P2C HD041025) and Family Demography Training (T32 HD007514).
Compliance with Ethical Standards
Conict of interest The authors declare that they have no conflicts of interest.
References
Amato, P. R., & Keith, B. (1991). Parental divorce and the well-being of children: A meta-analysis. Psy-
chological Bulletin, 110(1), 26–46.
Amato, P. R., & Booth, A. (1996). A prospective study of divorce and parent-child relationships. Journal
of Marriage and the Family, 58(2), 356–365.
Amato, P. R., Loomis, L. S., & Booth, A. (1995). Parental divorce, marital conflict, and offspring well-
being during early adulthood. Social Forces, 73(3), 895–915.
Amato, P. R., & Anthony, C. J. (2014). Estimating the effects of parental divorce and death with fixed
effects models. Journal of Marriage and Family, 76(2), 370–386.
Aquilino, W. S. (2006). The noncustodial father–child relationship from adolescence into young adult-
hood. Journal of Marriage and Family, 68(4), 928–946.
Bachman, H. J., Coley, R. L., & Carrano, J. (2011). Maternal relationship instability influences on chil-
dren’s emotional and behavioral functioning in low-income families. Journal of Abnormal Child
Psychology, 39(8), 1149–1161.
Booth, A., & Amato, P. R. (2001). Parental predivorce relations and offspring postdivorce well-being.
Journal of Marriage and Family, 63(1), 197–212.
462
J.A.Jarvis et al.
1 3
Braithwaite, S., Steele, E., Spjut, K., & Dowdle, K. K. (2015). Parent-child connectedness mediates the
association between marital conflict and children’s internalizing/externalizing outcomes. Journal of
Child Family Studies, 24(12), 3690–3699.
Brown, S. L. (2010). Marriage and child well-being: Research and policy perspectives. Journal of Mar-
riage and Family, 72, 1059–1077.
Carballo, J., García-Nieto, R., Álvarez-García, R., Caro-Cañizares, I., López-Castromán, J., Muñoz-Lor-
enzo, L., etal. (2013). Sibship size, birth order, family structure and childhood mental disorders.
Social Psychiatry & Psychiatric Epidemiology, 48(8), 1327–1333.
Chen, M., & Yip, P. S. F. (2018). Decomposing the crude divorce rate in five countries: Singapore, Tai-
wan, South Korea, the UK, and Australia. Asian Population Studies, 2, 137–152.
Cherlin, A. J., Furstenberg, F. F., Jr., Chase-Linsdale, P. L., Kiernan, K. E., Robins, P. K., Morrison,
D. R., etal. (1991). Longitudinal studies of effects of divorce on children in great Britain and the
United States. Science, 252(5011), 1386–1389.
Chung, Y., & Emery, R. (2010). Early adolescents and divorce in South Korea: Risk, resilience and pain.
Journal of Contemporary Family Studies, 41(5), 855–870.
Cox, M. J., & Paley, B. (1997). Families as systems. Annual Review of Psychology, 48(1), 243–267.
Creighton, M. J., Park, H., & Teruel, G. M. (2009). The role of migration and single motherhood in upper
secondary education in Mexico. Journal of Marriage and Family, 71(5), 1325–1339.
Cummings, E. M., Zahn-Waxler, C., & Radke-Yarrow, M. (1981). Young children’s responses to expres-
sions of anger and affection by others in the family. Child Development, 52(4), 1274–1282.
Cummings, E. M. (1987). Coping with background anger in early childhood. Child Development, 58(4),
976–984.
Dommaraju, P., & Jones, G. (2011). Divorce Trends in Asia. Asian Journal of Social Science, 39(6),
725–750.
Dufur, M. J., & Alexander, A. (2017). Does childhood family structure help create stratification in adult
education and labor market attainment? An argument for the selectivity perspective. Sociology
Compass, 11(1), 1–13.
Erel, O., & Burman, B. (1995). Interrelatedness of marital relations and parent-child relations: A meta-
analytic review. Psychological Bulletin, 118(1), 108–132.
Eun, K. S. (2007) Family values in Korea: A comparative analysis. In Paper presented at the 2007 World
Family Policy Forum, Brigham Young University, Provo, UT.
Fincham, F. D., & Osborne, L. N. (1993). Marital conflict and children: retrospect and prospect. Clinical
Psychology Review, 13(1), 75–88.
Fomby, P., & Cherlin, A. J. (2007). Family instability and child well-being. American Sociological
Review, 72(2), 181–204.
Gorman, G. H., Eide, M., & Hisle-Gorman, E. (2010). Wartime military deployment and increased pedi-
atric mental and behavioral health complaints. Pediatrics, 126(6), 1058–1066.
Han, Y. R., Jeong, G. H., & Kim, S. (2017). Factors influencing beliefs about intimate partner violence
among adults in South Korea. Public Health Nursing, 34, 412–421.
Hoffmann, J. P. (2002). The community context of family structure and adolescent drug use. Journal of
Marriage and Family, 64(2), 314–330.
Hoffmann, J. P. (2017). Family structure and adolescent substance use: An international perspective. Sub-
stance Use and Misuse, 52(13), 1667–1683.
Jekielek, S. M. (1998). Parental conflict, marital disruption and children’s emotional well-being. Social
Forces, 76(3), 905–936.
Johnson, D. R., & Young, R. (2011). Toward best practices in analyzing datasets with missing data: Com-
parisons and recommendations. Journal of Marriage and Family, 73(5), 926–945.
Katz, L. F., & Gottman, J. M. (1993). Patterns of marital conflict predict children’s internalizing and
externalizing behaviors. Developmental Psychology, 29(6), 940–950.
Katz, L. F., & Gottman, J. M. (1994). Patterns of marital interaction and children’s emotional devel-
opment. In R. D. Parke & S. G. Kellam (Eds.), Family research consortium: Advances in family
research. Exploring family relationships with other social contexts (pp. 49–74). Hillsdale, NJ: Law-
rence Erlbaum Associates Inc.
Kim, H. S. (2011). Consequences of parental divorce for child development. American Sociological
Review, 76(3), 486–511.
Kim, J., Park, S., & Emery, C. R. (2009). The incidence and impact of family violence on mental health
among South Korean women: Results of a national survey. Journal of Family Violence, 24, 193.
463
1 3
Family Structure andChild Well-Being inaNon-western Context:…
Ko, Y., & Park, S. (2020). Building a new intimate relationship after experiencing intimate partner vio-
lence in victim-SURVIVORS of South Korea. Journal of Interpersonal Violence, 35(1–2), 3–24.
https ://doi.org/10.1177/08862 60518 81426 5.
Kouros, C. D., Papp, L. M., Goeke-Morey, M. C., & Cummings, E. M. (2014). Spillover between marital
quality and parent-child relationship quality: Parental depressive symptoms as moderators. Journal
of Family Psychology, 28(3), 315–325.
Leadbeater, B. J., Kuperminc, G. P., Blatt, S. J., & Hertzog, C. (1999). A multivariate model of gender
differences in adolescents’ internalizing and externalizing problems. Developmental Psychology,
35(5), 1268–1282.
Lee, Y. J. (2006). Risk factors in the rapidly rising incidence of divorce in Korea. Asian Population Stud-
ies, 2(2), 113–131.
Lee, D., & McLanahan, S. S. (2015). Family structure transitions and child development: Instability,
selection, and population heterogeneity. American Sociological Review, 80(4), 738–763.
Lesthaeghe, R. (2010). The unfolding story of the second demographic transition. Population and Devel-
opment Review, 36(2), 211–251.
Margolin, G. (1988). Marital conflict is not marital conflict is not marital conflict. In R. D. Peters &
R. J. McMahon (Eds.), Social learning and systems approaches to marriage and the family (pp.
193–216). Philadelphia, PA: Brunner/Mazel.
McCulloch, A., Wiggins, R. D., Joshi, H. E., & Sachdev, D. (2000). Internalising and externalising chil-
dren’s behaviour problems in Britain and the US: Relationships to family resources. Children &
Society, 14(5), 368–383.
McLanahan, S. S., & Sandefur, G. (1994). Growing up with a single parent: What hurts, what helps.
Cambridge, MA: Harvard University Press.
Morrison, D. R., & Coiro, M. J. (1999). Parental conflict and marital disruption: Do children benefit
when high-conflict marriages are dissolved? Journal of Marriage and Family, 61(3), 626–637.
National Youth Policy Institute. (2012). Korea Youth Panel Survey- Elementary 4th gradeUser’s guide.
Seoul, South Korea: National Youth Policy Institute
OECD. (2016). SF2.3.B: Mean age of women at first birth, 1995 and 2016 or latest available. OECD
Family Database. Retrieved July 25, 2019 from https ://www.oecd.org/els/soc/SF_2_3_Age_mothe
rs_child birth .pdf.
OECD. (2017). How does Korea compare? The Pursuit of gender equality: An uphill battle. Paris: OECD
Publishing. https ://doi.org/10.1787/97892 64281 318-en
OECD. (2018a). SF3.1: Marriage and divorce rates. OECD Family Database. Retrieved July 20, 2019
from https ://www.oecd.org/els/famil y/SF_3_1_Marri age_and_divor ce_rates .pdf.
OECD. (2018b). SF2.4.A: Share of births outside of marriage: Proportion (%) of all births where the
mother’s marital status at the time of birth is other than married, 2016. OECD Family Database.
Retrieved July 26, 2019 form https ://www.oecd.org/els/famil y/SF_2_4_Share _birth s_outsi de_marri
age.pdf.
OECD. (2018c). Population with tertiary education (indicator). Retrieved January 15, 2020 from https ://
data.oecd.org/eduat t/popul ation -with-terti ary-educa tion.htm.
Park, H. (2007). Single parenthood and children’s reading performance in Asia. Journal of Marriage and
Family, 69(3), 863–877.
Park, H. (2008). Effects of single parenthood on educational aspiration and student disengagement in
Korea. Demographic Research, 18(13), 377–408.
Park, H., & Raymo, J. (2013). Divorce in Korea: trends and educational differentials. Journal of Mar-
riage and Family, 75(1), 110–126.
Park, H. (2014). Single parenthood and children’s education in Republic of Korea: An update. In H. Park
& K.-K. Kim (Eds.), Korean education in changing economic and demographic 153 contexts, edu-
cation in the Asia-Pacific region: Issues, concerns and prospects (Vol. 23). Dordrecht: Springer.
https ://doi.org/10.1007/978-981-4451-27-7_9
Park, G. R., Park, E. J., Jun, J., & Kim, N. S. (2017). Association between intimate partner violence and
mental health among Korean married women. Public Health, 152, 86–94.
Pew Research Center. (2015). The link between a college education and a lasting marriage. Retrieved
November 13, 2018 from https ://pewrs r.ch/1lf7x u9.
Pribesh, S., & Downey, D. B. (1999). Why are residential and school moves associated with poor school
performance? Demography, 36(4), 5212–5534.
Raymo, J. M., Park, H., Xie, Y., & Yeung, W.-J. J. (2015). Marriage and family in East Asia: Continuity
and change. Annual Review of Sociology, 41, 471–492.
464
J.A.Jarvis et al.
1 3
Riggio, H. R. (2004). Parental marital conflict and divorce, parent-child relationships, social support, and
relationship anxiety in young adulthood. Personal Relationships, 11(1), 99–114.
Statistics Korea. (2013). Status of divorces and remarriages for the past three decades. Marriage and
divorce. Retrieved August 5, 2019 from https ://kosta t.go.kr/porta l/eng/press Relea ses/8/11/index
.board ?bmode =downl oad&bSeq=&aSeq=31195 4&ord=1.
Statistics Korea. (2018a). Mean age of mother by birth order for provinces. Statistical database, vital
statistics. Retrieved August 15, 2019 from https ://kosis .kr/eng/stati stics List/stati stics ListI ndex.
do?menuI d=M_01_01&vwcd=MT_ETITL E&parmT abId=M_01_01&statI d=19620 04&t hema
Id=#Selec tStat sBoxD iv.
Statistics Korea. (2018b). Live births by marital legitimacy for provinces, annual 1981–2017. Statisti-
cal database, vital statistics. Retrieved August 5, 2019 from https ://kosis .kr/eng/stati stics List/stati
stics ListI ndex.do?menuI d=M_01_01&vwcd=MT_ETITL E&parmT abId=M_01_01&statI d=19620
04&thema Id=#Selec tStat sBoxD iv.
Thomson, E., & McLanahan, S. S. (2012). Reflections on “family structure and child well-being: eco-
nomic resources vs. parental socialization”. Social Forces, 91(1), 45–53.
Vandecasteele, L., & Debels, A. (2006). Attrition in panel data: The effectiveness of weighting. European
Sociological Review, 23(1), 81–97.
von Hippel, P. T. (2007). Regression with missing Ys: An improved strategy for analyzing multiply
imputed data. Sociological Methodology, 37(1), 83–117.
Waldfogel, J., Craigie, T., & Brooks-Gunn, J. (2010). Fragile families and child wellbeing. Future of
Children, 20(2), 87–112.
Wheaton, B. (1990). Life transitions, role histories, and mental health. American Sociological Review,
55(2), 209–223.
Wildeman, C., Wakefield, S., & Turney, K. (2013). Misidentifying the effects of parental incarceration?
A comment on Johnson and Easterling (2012). Journal of Marriage and Family, 75(1), 252–258.
Yoo, H. J. (2015). Living Cohabitation in the Republic of Korea: The Reported Experiences of Lesbians,
Gays and Heterosexuals. PhD thesis, University of York.
Yu, T., Petit, G. S., Lansford, J. E., Dodge, K. A., & Bates, J. E. (2010). The interactive effects of mari-
tal conflict and divorce on parent—adult children’s relationships. Journal of Marriage and Family,
72(2), 282–292.
Zorn, C. J. W. (2001). Generalized estimating equation models for correlated data: A review with applica-
tions. American Journal of Political Science, 45, 470–490.
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published
maps and institutional affiliations.
Aliations
JonathanA.Jarvis
1
· AshleyLarsenGibby
2
· MikaelaJ.Dufur
1
·
ShanaPribesh
3
Ashley Larsen Gibby
ashley_gibb[email protected]
Mikaela J. Dufur
Shana Pribesh
1
Department ofSociology, Brigham Young University, 2008 JFSB, Provo, UT84602, USA
2
School ofFamily Life, Brigham Young University, 2063 JFSB, Provo, UT84602, USA
3
Department ofEducational Foundations andLeadership, Old Dominion University, DCOE
2307, Norfolk, VA23529, USA
