Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Maltreated adolescents and their families: a longitudinal examination of family functioning, parenting attitudes, & youth mental health
(USC Thesis Other)
Maltreated adolescents and their families: a longitudinal examination of family functioning, parenting attitudes, & youth mental health
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
MALTREATED ADOLESCENTS AND THEIR FAMILIES:
A LONGITUDINAL EXAMINATION OF FAMILY FUNCTIONING, PARENTING
ATTITUDES, & YOUTH MENTAL HEALTH
By
Daniel SeungChul Lee
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(SOCIAL WORK)
August 2021
ii
ACKNOWLEDGEMENTS
Immeasurable gratitude goes to my parents, Susan and Peter, whose unwavering support
and sacrifice compelled me to persevere beyond challenges to become, not only the first in my
family to graduate from college, but to pursue this endeavor to this point and earn a Ph.D. I
would like to thank my better half, Rosie for her steadfast care and love. And to Theo: the
overflow of joy and laughter you bring to the household brighten even the toughest of days.
To my committee, Drs. Ferol Mennen, Jordan Davis, Julie Cederbaum, and Mary Helen
Immordino-Yang: thank you for your steady mentorship, guidance, and support throughout the
years. It has meant so much to me, and I have difficulty summing up my appreciation in words.
Drs. Anne-Marie Yamada, Chih-Ping Chou, Christopher Beam, Dorian Traube, Eric Pedersen,
Hans Oh, Michal Sela-Amit, and Olivia Lee: I thank for encouraging me along various parts of
this journey, as well as for the opportunities you gave me to learn and grow. I give special thanks
to Dr. Michael Hurlburt and Ms. Malinda Sampson for their diligence in helping ensure I – and
all my fellow colleagues in the PhD program – had what we needed from start to finish.
To the Prevention, Early intervention, and Addiction Recovery Lab (PEARL), the
Substance use And Mental health among Marginalized Youth (SAMMY) lab, and the Young
Adolescent Project (YAP) lab: I appreciate the help in shaping how I view and think about
various problems and their solutions, and for the chance to learn from and collaborate on many
works. A special thanks to Eric and Jordan. I am excited about our continued work together!
To my PhD colleagues and my cohort: thanks for your friendship over the years! I will
remember our meals, trips, and study sessions fondly. Lastly, to my friends and community who
helped provide me respite from my studies, I very much appreciate you all.
iii
TABLE OF CONTENTS
Acknowledgments ......................................................................................................................................... ii
List of Tables ................................................................................................................................................ iv
List of Figures ..................................................................................................................................v
Abstract .......................................................................................................................................... vi
Chapter 1: Introduction ....................................................................................................................1
Research questions ...............................................................................................................5
Chapter 2: Child maltreatment and family conflict across adolescence: A growth mixture
modeling approach ...........................................................................................................................6
Introduction .........................................................................................................................6
Methods .............................................................................................................................12
Results ...............................................................................................................................16
Discussion .........................................................................................................................24
Chapter 3: Externalizing symptoms, family conflict, and family cohesion: Testing competing
models of between and within-person associations across adolescence for maltreated youth .....32
Introduction .......................................................................................................................32
Methods .............................................................................................................................41
Results ...............................................................................................................................46
Discussion .........................................................................................................................57
Chapter 4: Parenting attitudes and adolescent depressive symptoms in maltreated youth: Are
family conflict and cohesion mediators? ......................................................................................66
Introduction .......................................................................................................................66
Methods .............................................................................................................................72
Results ...............................................................................................................................79
Discussion .........................................................................................................................85
Chapter 5: Implications and Conclusion ........................................................................................94
References ......................................................................................................................................99
iv
LIST OF TABLES
Table 1. Demographics and family conflict means by grouping ................................................................. 17
Table 2. Bivariate correlations ..................................................................................................................... 18
Table 3. Model fit indices ............................................................................................................................ 19
Table 4. Odds ratios for multinomial logistic regression with covariates ................................................... 20
Table 5. Differences between adolescent-report and parent-report family conflict .................................... 21
Table 6. Baseline demographics .................................................................................................................. 48
Table 7. Correlations, means, and standard deviations ............................................................................... 49
Table 8. Full model and multi-group model results .................................................................................... 50
Table 9. Demographics, means, and t/χ² tests for differences in study variables by maltreatment and sex 73
Table 10. Bivariate correlations .....................................................................................................77
v
LIST OF FIGURES
Figure 1. Family conflict classes ...................................................................................................23
Figure 2. Total sample models .......................................................................................................53
Figure 3. Rule breaking by maltreatment and comparison groups ................................................55
Figure 4. Aggression models by maltreatment and comparison groups ........................................56
Figure 5. Conceptual model ...........................................................................................................78
Figure 6. Family conflict models ...................................................................................................81
Figure 7. Family cohesion models ................................................................................................83
vi
Abstract
The aim of this dissertation is to investigate the role of family factors, including family
conflict, family cohesion, and parenting attitudes on the mental health outcomes of maltreated
adolescents. Given the proximal importance of the family on adolescent development (Koepke &
Dennisen, 2012), this dissertation, in 3 papers, examines a group of maltreated and non-
maltreated adolescents and their parents across four waves of data. Data for this dissertation
comes from the Young Adolescent Project, an on-going longitudinal study on the developmental
outcomes of maltreatment on a group of ethnically/racially diverse adolescents and their families
from the same urban neighborhoods (Negriff et al., 2020). To best study the influence of the
family, a subsample was chosen for these studies which include only adolescents who remained
with their families throughout the periods examined. Our sample includes 288 participants for
the first two studies and 220 for the third study.
The first study (Chapter 2) of this dissertation examines heterogeneity in family conflict
trajectories during earlier adolescence and asks whether maltreated adolescents face greater risk
for elevated or atypical levels of family conflict than their non-maltreated peers. It addresses a
gap in knowledge related to the longitudinal experiences of family conflict for earlier
adolescence, the period when family conflicts tend to increase as adolescents begin to negotiate
new roles in their family and strive for greater autonomy and independence and to extend it to
maltreated adolescents. Results support a four-class solution in which maltreated adolescents
have greater odds of belonging to a low-increasing and high-decreasing relative to the
persistently-low class of family conflict. This suggests that maltreated adolescents are at greater
risk of elevated and atypical trajectories of family conflict, which indicate a need for
practitioners and policy makers to take a longitudinal view of intervening with this population.
vii
The second study (Chapter 3) sought to clarify the relationship between family conflict,
family cohesion, and adolescent externalizing symptoms over time. Using quantitative
methodology that allows for the disaggregation of between- and within-person effects, this study
tested three competing conceptual frameworks: the interpersonal risk, symptom driven, and
transactional models, in order to best explain the longitudinal associations between family
conflict, family cohesion, and externalizing symptoms among maltreated and non-maltreated
adolescents. Results indicate that at the within-person level, a symptom driven model is indicated
for aggression, whereas transactional models are indicated for rule breaking, with greater than
one’s own typical externalizing symptoms driving increases in family conflict and decreases in
family cohesion (for rule breaking). Grouping by maltreatment, a symptom driven model applies
to maltreated adolescents, and an interpersonal risk for comparison adolescents. These results
indicate that maltreated adolescents may uniquely be at-risk of externalizing symptoms that drive
greater conflict and lower cohesion within their families. The implication is that practitioners
should address externalizing symptoms through the underlying mechanisms of externalizing
symptoms, which could pay dividends by decreasing related family conflicts.
The third study (Chapter 4), using the Social Information Processing model, examined
the influence of parenting attitudes on adolescent depressive symptoms, and whether this
relationship was mediated by family conflict and family cohesion. Results indicate that while
parenting attitudes and adolescent depressive symptoms are concurrently associated at the first
wave of the study, these effects fade across time and are not mediated by family conflict or
cohesion. Instead, for maltreated adolescents, family cohesion directly protects against
depressive symptoms in later years. These findings indicate that parenting attitudes may be a
viii
more suitable target for parents of younger adolescents, but that targeting family cohesion for
maltreated adolescents may confer some protection against depressive symptoms.
1
Chapter 1: Introduction
Child maltreatment can lead to mental health problems including depressive and
externalizing outcomes (Denholm et al., 2013; McLaughlin et al., 2010; Sousa et al., 2018).
More than 656,000 youth, in 2019, had substantiated reports of maltreatment (U.S. Department
of Health and Human Services [USDHHS], 2021). Of these youth, only 22.9% were removed
from their home, which suggests a majority remain with their families (USDHHS, 2021). This
number is likely to grow as a result of recent federal legislation supporting family preservation
services to keep youth with their families (Lindell et al., 2020). Some researchers have examined
family risks and mental health outcomes of maltreated youth and found that approximately 36%
experience clinical levels of externalizing symptoms (compared to 16.5% in the general
population), and 29%, internalizing symptoms (Campbell et al., 2012). Further, in their
nationally representative samples, Campbell and colleagues (2012) as well as Horwitz, Hurlburt,
and colleagues (2011) report that these youth (ages 0-18) and families experience higher family-
level risks including a lack of a supportive caregiver in the home (53.1%), high levels of family
stress (49.2%), a lack of social support (36.3%), primary caregivers with poor parenting skills
(27%), intrafamilial violence (22.1%), unrealistic expectations of the child (13.5%), and
excessive/inappropriate use of discipline (15.1%). Taken together, these figures suggest that
maltreated youth who remain at home face a great deal of familial adversities and a higher risk
for clinical levels of externalizing and internalizing symptoms.
Adolescence is an important period because family-level risks may become even more
acute. As adolescents begin to negotiate with parents for greater autonomy and independence
from the family, disruptions in important aspects of family relations can occur (Branje, 2018),
including increases in family conflict, defined as – criticism, tension, anger, and hostility within
2
and among family members (Branje, 2018; Fosco, Van Ryzin, Connell, et al., 2016), and
decreases in protective factors such as family cohesion (Tsai et al., 2013) – defined as feelings of
bonding and emotional connectedness shared between family members (Olson et al., 2019). In
parallel, there is a rising risk for onset and increases in depressive symptoms (Barker et al., 2019;
Keyes et al., 2019) and externalizing problems, which have long-term negative consequences
(Carter, 2019; Chan et al., 2013; Lee et al., 2014).
Key limitations exist in our knowledge of these relationships. For instance, while studies
have reported that family conflict is a strong risk factor correlated with child maltreatment
(Finkelhor, 1993; Stith et al., 2009; Vial et al., 2020), empirical evidence is lacking for whether
maltreated adolescents face greater risk for elevated and or atypical levels of family conflict.
Providing empirical evidence for the link between maltreatment and family conflict could aid in
informing practice and policies to benefit these youth and their families.
Increases in family conflict as well as decreases in family cohesion have been reported to
be associated with greater externalizing outcomes (Choe et al., 2014; Taylor et al., 2016).
However, these relationships differ in their temporal ordering, and a clearer understanding may
improve intervention efforts – particularly for maltreated adolescents. For instance, some studies
have reported that externalizing symptoms in adolescence precede family conflict (Steeger &
Gondoli, 2013), while others have reported that family conflict precedes externalizing problems
(Benson & Buehler, 2012); reciprocal relations across time in adolescents have also been
reported (Choe & Zimmerman, 2014). Similarly, for family cohesion, Elam et al. (2018) reported
externalizing problems being associated with decreased cohesion, while Barr et al. (2012)
reported that lower family cohesion is associated with future externalizing problems. These
represent different conceptual frameworks (i.e., interpersonal risk, symptom driven, and
3
transactional models), and clarification is needed of the relationship between family conflict,
family cohesion, and externalizing problems to determine which model best explains the
relationships.
Recent methodological advances have promoted disentangling the between- and within-
persons effects in order to better understand the variance explained at these different levels
(Curran et al., 2014). Berry & Willoughby (2017) have argued that processes relating to cross-
lagged effects (e.g., adolescent’s externalizing symptoms’ effect on future family conflict) are
more appropriate to understand at the within-person level. But with a few exceptions (e.g.,
Knopp et al., 2017; Mastrotheodoros et al., 2020; Zemp et al., 2018), most studies examining the
relationships between family conflict and externalizing symptoms have not disaggregated effects
at the between- and within-person level, limiting our understanding of the true relations between
family-level processes and adolescent externalizing problems. Understanding this would aid
prevention and intervention efforts targeting the precipitating (family conflict or externalizing
problems) and understanding the role of family cohesion as buffer.
Child maltreatment has a long been associated with parenting attitudes (Milner, 1993,
2003), with studies reporting maltreating parents as endorsing more inappropriate parenting
attitudes on average than non-maltreating parents (Camilo et al., 2020a). The Social Information
Processing model posits that parents have pre-existing attitudes about childrearing, which serves
to interpret their caregiving environment and in turn, influences how they interpret their child’s
behaviors and needs, and guides parental responses (Azar et al., 2013; Milner 2003). Parents who
attribute negative intent to their children’s behaviors may be more prone to responding
inappropriately, which may lead to maltreatment (Camilo et al., 2020a). This model helps
explain how parenting attitudes may influence youth outcomes. Some evidence has supported the
4
association between more inappropriate parenting attitudes and greater adolescent depressive
symptoms (Park et al., 2016; Weed et al., 2013). But depressive symptoms during adolescence,
are also influenced by the family environment – which includes family conflict and family
cohesion (Moos & Moos, 1994; Olson et al., 2019; Yap et al., 2014). Numerous studies have
found family conflict and family cohesion play important roles as risk and protective factors,
respectively, on adolescent depressive outcomes (Branje, 2018; Fosco et al., 2016; Moreira &
Telzer, 2015; Weymouth et al., 2016).
But despite these connections between parenting attitudes and adolescent depressive
symptoms, and the proximal influence of the family on adolescent outcomes, limited work has
been done to examine family conflict and family cohesion as mediators in the relationship
between parenting attitudes and adolescent depressive symptoms. Moreover, given the
differences in parenting attitudes and adolescent depressive symptoms by maltreatment (Camilo
et al., 2020a; Campbell et al., 2012; Norman et al., 2012) and sex (Lewis et al., 2015; Kaferly et
al., 2020; Salk et al., 2017), key limitations in our understanding exist. Specifically, because
parenting attitudes as well as family conflict and family cohesion are modifiable by interventions
(Clark et a., 2013; Fosco et al., 2016), it is important to understand whether family-level
mechanisms (i.e., conflict and cohesion) might mediate the relationship between parenting
attitudes and adolescent depressive symptoms, and whether these relationships differ by
maltreatment and sex. This clarifies targets for intervention for distinct populations.
These limitations in extant studies are addressed in this dissertation by use of quantitative
methods that examine the longitudinal experiences of family-level factors by maltreated
adolescents. Broadly, the goal of these studies is to inform intervention efforts and policies for
5
those who work with and on behalf of maltreated adolescents and their families. To this end, the
three distinct studies in this dissertation ask the following questions:
Study 1 (Chapter 2)
1) To what extent is there heterogeneity in trajectories of family conflict across early-to-mid
adolescence?
2) Are maltreated adolescents at greater risk for elevated and or atypical levels of family
conflict across adolescence?
Study 2 (Chapter 3)
1) What are the longitudinal between-person associations for family conflict, family
cohesion, and externalizing problems across adolescence?
2) Which conceptual model (interpersonal risk, symptom driven, or transactional model)
best explains the within-person associations for family conflict, family cohesion, and
externalizing problems across adolescence?
3) Do maltreated adolescents who remain home have different pathways of association with
family conflict, family cohesion, and externalizing problems than non-maltreated
adolescents?
Study 3 (Chapter 4)
1) What is the relationship between parenting attitudes and adolescent depressive
symptoms?
2) Do family conflict and family cohesion mediate the relationship between parenting
attitudes and adolescent depressive symptoms?
3) Are there differences in these relationships by maltreatment and sex?
6
Chapter 2: Child maltreatment and family conflict across adolescence: A growth mixture
modeling approach
Introduction
Family conflict, defined as hostility, criticism, anger, and tension among family members
(Fosco et al., 2016; Moos & Moos, 1994), has been linked to a host of serious problems,
including adverse interpersonal, psychosocial, and mental health outcomes (Clayborne et al.,
2018; Cummings & Schatz, 2012; Weymouth et al., 2016). While past studies have reported
relationships between family conflict and child maltreatment (Hamby et al., 2010; Thornberry et
al., 2014), few have examined the longitudinal experiences of family conflict following reported
child maltreatment. Further, a number of longitudinal studies report levels of family conflict vary
across developmental periods, with an increase in family conflict frequency occurring during
adolescence (Branje, 2018). These elevated and atypical levels of family conflict are associated
with deleterious mental, behavioral, and physical health outcomes (Bi et al., 2015; Choe et al.,
2014; Yu, 2019). This heterogeneity in levels of family conflict during adolescence has not been
explored in relation to child maltreatment, which could aid in understanding the potential sources
of family conflict risk and associated outcomes for maltreated adolescents. This study seeks to
investigate family conflict trajectories across early-to-mid adolescence to understand the risk that
maltreated adolescents may face in their family environment.
Maltreatment and Family Conflict
Child maltreatment is associated with a host of adverse downstream consequences over
time, including substance misuse, mood disorders, and family conflict (McLaughlin et al., 2010;
Sousa et al., 2018). Given that only 22.9% of youth who had substantiated maltreatment are
removed from the home, a sizable number of youths remain at home after maltreatment (U.S.
7
Department of Health and Human Services [USDHHS], 2021). These data align with efforts that
have been made to provide more families with family preservation services to prevent youth
from entering out-of-home placements (Lindell et al., 2020). These family preservation efforts
focus on keeping families together, but adequate support needs to be provided to the family to
make this happen and this may entail understanding the patterns of family conflict that
maltreated adolescents face during this period. As such, in this study, we examine adolescents
who remain with their biological families following a report of maltreatment.
Family conflict is important to study across adolescence because it co-occurs alongside
the onset of other risks, such as the development of depressive and externalizing symptoms
(Choe et al., 2014), and it can have a lasting impact on patterns of family conflict (Rothenberg et
al., 2016). In a large panel study of children and families, family conflict during adolescence was
the strongest and most consistent predictor of mental health and substance use in adulthood
(Herrenkohl et al., 2012). These patterns are likely intergenerational, with some work reporting
that conflict in families of origin was strongly associated with current conflict in families of the
proceeding generation (Rothenberg et al., 2016). The spillover hypothesis (Erel & Burman,
1995; Sears et al., 2016), which suggests negative effects from one area of life can transfer from
one family member to another through social interactions, can be applied to family conflict. For
instance, interparental conflict can negatively affect members uninvolved directly in the conflict,
such as youth (Sherrill et al., 2017) through pathways like transmission of increased negative
mood among parents, which can subsequently influence other negative outcomes, such as school
problems (Timmons & Margolin, 2015).
Varying degrees of family conflict are associated with increased risk for maltreatment in
adolescents (Thornberry et al., 2014). Many studies, including one that examined a nationally
8
representative sample, report a strong association between child maltreatment and family
violence and conflict (Hamby et al., 2010; Herrenkohl & Herrenkohl, 2007). It is well
established that family conflict is a common and strong risk factor for child maltreatment
(Finkelhor, 1993; Mollerstrom et al., 1992; Stith et al., 2009). More recent work confirms a
strong correlation and overlap between family conflict and other risk factors involving
problematic relationships and domestic violence between caregivers (Vial et al., 2020).
However, these studies looked at family conflict as a risk factor for child maltreatment, not how
family conflict is experienced following maltreatment.
Many studies have examined the relationships between maltreatment and family conflict
and violence cross-sectionally (e.g., Hamby et al., 2010; Herrenkohl & Herrenkohl, 2007), but
few longitudinally (Campbell et al., 2012; Horwtiz et al., 2011). Cross-sectional approaches limit
our understanding of how family conflict varies across adolescence in maltreated youth and their
families, making longitudinal examination critical to understanding the problem. Two
longitudinal studies using nationally representative data identified risks, such as intimate partner
violence and high family stress faced by youth investigated for suspected maltreatment who
remained at home (Campbell et al., 2012; Horwtiz et al., 2011). Family conflict occurs in
different forms (e.g., arguing between parent-adolescent dyad, aggression between siblings, etc.)
and even at lower levels, can confer risk for negative outcomes such as internalizing and
externalizing problems (Choe et al., 2014; Trentacosta et al., 2011). Levels of family conflict can
increase across adolescence and be associated with consequential outcomes such as depressive
symptoms and violent behavior (Choe et al., 2014); these relationships are difficult to interpret
through a cross-sectional or two-timepoint analysis. Examining the relationship between mental
9
health and family conflict among maltreated youth over time will help to disentangle these
relationships.
Family Conflict and Person-Centered Methodologies
Over the course of adolescent development, levels of family conflict have been shown to
vary. A seminal meta-analysis shows that while some level of family conflict is expected across
adolescent development (Laursen et al., 1998), certain patterns, such as protracted levels of
elevated conflict or increasing levels, can be concerning because they may lead to outcomes such
as internalizing and externalizing problems (e.g., Choe et al. 2014; Trentacosta et al., 2011),
smoking (Yu, 2019), and adjustment problems in emerging adulthood (Castellani et al., 2014).
Particularly, during earlier adolescence, when youth negotiate new roles within families and
work towards autonomy and independence, some families may see elevated levels of conflict
(Branje, 2018), making this an important period for examination.
Methodological advances in person-centered approaches allow for identification and
examination of interindividual differences within unobserved sub-populations that follow
intraindividual changes over time. Similar to latent class analysis, growth mixture modeling
allows for the identification of heterogeneity within sub-populations that is based on longitudinal
changes within each unobserved sub-population in a specific characteristic. This includes factors
like family conflict, which could not otherwise be observed at the mean level (Ram & Grimm,
2009). Further, growth mixture modeling allows for testing associations between these latent
trajectories with anteceding variables – such as demographic covariates or child maltreatment.
There have been a number of studies that have examined heterogeneity in latent
trajectories of family conflict in different populations, ages and developmental phases, with
differing results. The majority of studies have reported either a three- or four-class solution of
10
family conflict trajectories. Common classes of groups include: a 1) persistently (or stable) high
group, 2) moderate or high but decreasing group, 3) moderate or low but increasing, and 4)
persistently low trajectory group (Bi et al., 2015; Castellani et al., 2014; Choe et al., 2014;
Trentacosta et al., 2011; Yu, 2019). But these studies sampled a wide range of families, with
youth ranging from as young as five years old (Trentacosta et al., 2011) to some studies that
included individuals 18 years and above (Bi et al., 2015). Only one study looked at earlier
adolescence (i.e., 4
th
to 8
th
graders; Yu, 2019), but this study was outside of the U.S., limiting
generalizability to U.S. adolescents. These studies also varied by other ethnic groups and family
composition in their sampling. For instance, Choe et al. (2014) examined African American
youth between 14 and 18 years old, while Trentacosta et al. (2011) exclusively examined
mother-son dyads between 5 to 15 years old. Though the measures of family conflict and type of
analyses have also varied (e.g., latent growth curve modeling versus growth mixture modeling),
differences in classes entailed not just characterizations of the levels of conflict, but also the
relative size of each group. For example, most studies have reported the highest percentage of
individuals concentrated in the moderate to low/stable low class, which typically accounted for
about 50% to as high as 90% (Bi et al., 2015; Trentacosta et al., 2011), and the lowest percentage
of individuals in the high/persistently high classes, which ranged from 3% to 15% of samples
analyzed (Choe et al., 2014; Yu, 2019). While these literature address middle-to-late adolescents,
family conflict trajectories may be most pertinent to examine during earlier adolescence given
the increases of family conflict incidents which are more likely at this stage (Branje, 2018).
Understanding this variation among vulnerable youth would better inform our understanding of
this population which would improve targeted prevention and intervention efforts and emphasize
a longitudinal view for service providers and policy makers.
11
Therefore, taken together, these discrepancies, including the number of classes, seem to
indicate that different populations may experience heterogeneous patterns of family conflict, and
in varying proportionality. These variations may be due to age and development of samples but
may also be based on uninvestigated antecedents. That is, populations who have undergone
important experiences of adverse antecedents, like maltreated adolescents, may be at greater
future risk of elevated levels of family conflict, which would be important to understand.
Aims
Since family conflict is a well-known correlate of child maltreatment (Vial et al., 2020),
but less is known about the longitudinal experiences of family conflict in this population, this
study seeks to examine if there is heterogeneity in trajectories of family conflict which differ by
maltreatment. Further, studying family conflict in relation to maltreatment across earlier
adolescence is important because of the modifiable nature of family conflict. This is true even in
more at-risk populations, such as families living in urban, low-income areas, in which family
conflict can be modified through interventions (Fosco et al. 2016). Given that only one study has
examined the salient period of earlier adolescence (Yu, 2019), our study asks: 1) to what extent
is there heterogeneity in trajectories of family conflict across early-to-mid adolescence; and 2)
are maltreated adolescents at greater risk for elevated and or atypical levels of family conflict
across adolescence? Given the discrepancy in trajectories of family conflict from past studies, we
are agnostic about the number of classes, their characteristics (i.e., persistently high, low,
decreasing, increasing, etc.), and the size in our target population. However, based on previous
findings that have identified associations between maltreatment and family conflict, we
hypothesize that maltreated adolescents will likely face elevated or atypical trajectories. In
12
addition, this study uses a comparison group design, from which confounding effects could be
disentangled to better understand the relationship between maltreatment and family conflict.
Methods
Sample
Study recruitment. Data for our analyses comes from the Young Adolescent Project, a
study of the long-term developmental effects of maltreatment on ethnically diverse maltreated
and comparison adolescents from the same urban neighborhoods (Negriff et al., 2020). The
participants in the maltreatment group were recruited from the Los Angeles County Department
of Children and Family Services (LACDCFS) from new cases who met inclusion criteria for the
study. Comparison children were recruited from school lists (for more details on the sample
description and inclusion criteria, see: Negriff et al., 2020). The institutional review boards of the
large private university in California and the LADCFS permitted contact of participants with a
letter regarding the study.
For this study, the first three time points of the study were analyzed, with approximately
12 months between Time 1 (T1) and Time 2 (T2), and 18 months between T2 and Time 3 (T3).
The current study restricted the sample to adolescents who lived with their birth parent(s) and
completed assessments at T1 and at least one assessment at either T2 or T3. The retention rate
from T1 to T2 was 89.9%, and from T2 to T3 was 82.6%; final sample was N = 288.
Participants. Mean age of study participants at T1 was 10.9 years (SD 1.12);
approximately half self-identified as male (49.7%), and 75.4% self-identified as Black or
Hispanic. Average age of parent was 36.5 (SD 6.72), with about 68.6% having completed high
school (or beyond). Detailed information on the sample can be found in Table 1.
Measures
13
Family conflict. We used a subscale of the Family Environment Scale (FES; Moos &
Moos, 1994) to measure family conflict. The FES has been shown to have fair to good test-retest
reliability, internal consistency, and construct and discriminant validity (Moos & Moos, 1994).
The conflict subscale assesses the degree to which aggression, hostility, and violence is endorsed
by the reporter in the family (Moos & Moos, 1994). The scale consists of 14 true(1)/false(0)
agreement items (a full list of items is available in the Appendix), which were calculated as a
sum score ranging from 0 to 14 ( = .85, .85, .89 for T1 to T3, respectively). Minor changes
were made in wording for clarity. For instance, one items reads, “Family members often criticize
each other,” while the original version reads, “household members often criticize each other.”
For this analysis, although we have access to both adolescent and parent reports of family
conflict, and while using both parent and adolescent reports could be useful for capturing a more
comprehensive picture of family conflict (Cummings & Schatz, 2012), the primary analysis used
adolescent self-report of family conflict. This decision was made because a recent study showed
discrepancies between parent and adolescent report of [parent-adolescent] conflict intensity –
indicating differing perspectives of conflict that occur in families during early-to-middle
adolescence (Mastrotheodoros et al., 2020). Also, adolescent perceptions of global family
conflict have been shown to be more closely associated with a third-party outside observer’s than
parents’ report (Laursen & Collins, 2009). To probe results and enhance interpretability, post hoc
analyses examined parent-report means of family conflict. Parent report family conflict also
ranged from 0 to 14 and asked the same questions for adolescents as for adults of the family
environment, with good internal consistency across time ( = .84, .82, .83 for T1 to T3,
respectively).
14
Maltreatment. For this study, we used a dichotomized variable representing
maltreatment (0 comparison/1 maltreatment), with those who were referred to our study by child
protective services included in the maltreatment group and those who were recruited from the
school list sample in the comparison group (for more details see: Negriff et al., 2020). Parents of
comparison adolescents reported no previous experience with child welfare services.
Covariates. Family conflict can differ by various sociodemographic factors (e.g., sex;
Weymouth et al., 2016) and as such, our study included the following as covariates: parental
education (0 did not graduate high school/1 high school graduate and beyond), and adolescent’s
age (centered around mean), and self-identified sex (0 female/1 male).
Analytic Plan
Preliminary analyses examined descriptives and differences in our sample by
maltreatment grouping. Bivariate correlations examined relations between study variables (see
Table 2). We then fit a taxonomy of growth mixture models (GMM) for family conflict across
adolescence via the manual three-step approach (Nylund-Gibson et al., 2014) using Mplus
version 8.0 (Muthé n & Muthé n, 1998-2018). GMM uses an iterative process to cluster
individuals into similar trajectory classes/groups based on intercepts (i.e., starting values) and
slopes (i.e., change process). To determine the optimal number of classes, we first estimated an
unconditional growth model, examining the changes in the sample’s mean levels of adolescent-
reported family conflict across the three timepoints (approximately a year apart between
timepoints). This exploration of our data indicates significant variance existed in both the
intercept and slope and the best fitting model for family conflict followed a linear growth curve.
Accordingly, all models are constrained to linear growth across timepoints.
15
Informed by Nylund et al. (2007), the separate models estimate fitting one to five class
solutions, using the following model fit indices to determine best fit: log likelihood (LL),
negative 2 log likelihood (−2LL), Akaike Information Criteria (AIC), Bayesian Information
Criteria (BIC), and the sample size–adjusted Bayesian Information Criteria (aBIC). The −2LL,
AIC, BIC, and aBIC are all log likelihood measures for which lower values indicate better fit.
Further, we use three likelihood tests: the Vuong-Lo-Mendell-Rubin test (VLMR LRT), the Lo-
Mendell-Rubin adjusted likelihood ratio test (LRT), and the bootstrapped likelihood ratio test
(BLRT), where a p value greater than .05 indicates the k-1 model is a better fit to the data (k
represents number of classes; Berlin et al., 2014). Nylund and colleagues (2007) also suggest
substantive considerations such as existing theory, model parsimony, and previous research to
inform selection of the best fitting model. Selection for class size was also guided by Hipp and
Bauer (2006), who suggest an adequate class size be no less than 5% of the sample.
After determining the optimal number of class trajectories, multinomial logistic
regression models were fit to the data, in which maltreatment, along with covariates (i.e.,
parental education, adolescent age, sex), predicted membership of emergent family conflict
classes. This latent class regression was a multinomial logistic regression. The three-step
approach accommodates the inclusion of auxiliary variables (e.g., covariates) in the growth
model by saving posterior probabilities from the unconditional growth model and creating modal
class assignments; these fixed values (i.e., probabilities) account for measurement error in the
class assignment when covariates are included (Nylund-Gibson et al., 2014). Little’s MCAR test
was not significant (χ
2
= 27.49, df = 20, p = 0.122), so maximum likelihood estimator with
robust standard errors was applied using Mplus to handle missing data, and each participants’
available data were used in full without deletion. To further probe results, post hoc analyses of
16
parent-report family conflict, using modal class assignments and t-tests with Bonferroni
corrections (p < .006) were conducted to examine mean-level differences between adolescent
and parent reports of family conflict across classes and timepoints.
Results
Results from our preliminary analyses are presented in Table 1 (demographics and group
differences by study variables) and Table 2 (bivariate correlations and means). Our sample
significantly differed on parental education between maltreated and comparison samples (χ
2
=
32.87, p < .001), which was included as a covariate in the multinomial regression. Further,
Pearson’s correlations showed family conflict across time was correlated (r = .54, p < .01 from
T1 to T2, r = .44, p < .01 from T2 to T3). Family conflict was not associated with any
demographic variables, except for a small correlation with parental education at T2 (r = .13, p <
.05).
17
Total Sample
(n = 288)
Maltreated
(n = 146)
Comparison
(n = 142)
χ2 /t
M ± SD/% M ± SD/% M ± SD/%
Age of adolescent 10.9 ± 1.12 10.74 ± 1.11 11.06 ± 1.12 2.43
Age of parent 36.49 ± 6.72 36.26 ± 6.27 36.71 ± 7.17 0.46
Sex of adolescent
Male 155 (53.8%) 72 (49.3%) 83 (58.5%) 2.42
Female 133 (46.2%)
74 (50.7%) 59 (41.5%)
Parental education
Did not graduate high school 90 (31.4%) 68 (46.9%) 22 (15.5%) 32.87***
Graduated high school and beyond
197 (68.6%) 77 (53.1%) 120 (84.5%)
Race/ethnicity
Hispanic 101 (35.1%)
56 (38.4%) 71 (50.0%)
5.56
White 30 (10.4%)
16 (11.0%) 14 (9.9%)
Black 127 (44.1%)
60 (41.1%) 41 (28.9%)
Mixed or biracial
30 (10.4%) 14 (9.6%) 16 (11.3%)
T1 Family conflict (adolescent report)
4.06 ± 3.58 4.44 ± 3.69 3.67 ± 3.43 -1.82
T2 Family conflict (adolescent report)
4.11 ± 3.61 4.73 ± 3.58 3.57 ± 3.56 -2.56
T3 Family conflict (adolescent report)
4.97 ± 3.69 5.52 ± 3.74 4.59 ± 3.63 -1.78
*** Indicates significant at the p <.001 level.
Table 1. Demographics and family conflict means by grouping
18
1 2 3 4 5 6 7 8 9
1 Age --
2 Sex .05 --
3 Race/ethnicity -.04 -.04 --
4 Parent age .23
**
.04 .06 --
5 Parental education .03 -.06 -.14
*
.13
*
--
6 Maltreatment -.14
*
-.09 -.12 -.21
**
-.34
**
--
7 T1 Family conflict .02 -.05 -.09 -.01 .06 .11 --
8 T2 Family conflict .04 -.07 -.09 -.02 .13
*
.16
*
.54
**
--
9 T3 Family conflict .12 -.07 -.09 -.03 -.08 .12 .31
**
.44
**
--
* Indicates significant at the p < .05 level, ** p < . 01.
Table 2. Bivariate correlations
19
Table 3. Model fit indices
Notes. AIC = Akaike Information Criterion; BIC = Bayesian Information Criterion; Adj. BIC = Adjusted Bayesian Information
Criterion; VLMRT = Vuong-Lo–Mendell–Rubin Likelihood Ratio Test; LRT = Likelihood Ratio Test; BLRT = Parametric
Bootstrapped Likelihood Ratio Test. Bold line indicates the best fitting, final model.
Log
Likelihood
(LL)
Negative
2 LL
(-2 LL)
AIC BIC
ADJ.
BIC
Entropy
VLMR
LRT
VLMR
LRT p
Value
LRT
LRT
p
Value
BLRT
BLRT
p
Value
1
CLASS
-1934.513 3869.03 3885.03 3914.27 3888.91 -- -- -- -- -- -- --
2
CLASS -1910.475 3820.95 3842.95 3883.17 3848.28 0.788 48.076 0.0001 45.4 0.0002 48.08 0.000
3
CLASS
-1898.142 3796.284 3824.28 3875.47 3831.07 0.725 24.666 0.3379 23.3 0.3581 24.67 0.000
4
CLASS -1884.335 3768.67 3802.67 3864.82 3840.91 0.776 27.613 0.0251 26.1 0.0292 27.61 0.000
5
CLASS
-1873.303 3746.606 3786.61 3859.73 3796.31 0.779 22.064 0.3055 20.8 0.327 22.06 0.000
20
Persistently-low family conflict (Class 4) versus
beta SE OR beta SE OR beta SE OR
LL UL LL UL LL UL
Maltreatment (Comparison reference) 0.22 0.63 1.24 [ 0.00 - 2.48 ] 0.13 0.70 1.13 [ -0.24 - 2.51 ] 1.09 0.70 2.98 [ 1.62 - 4.35 ]
Sex (Female reference) -0.03 0.57 0.97 [ -0.14 - 2.09 ] -0.95 0.61 0.39 [ -0.81 - 1.58 ] 0.64 0.67 1.90 [ 0.59 - 3.20 ]
Age (centered on grand mean) 0.46 0.21 1.58 [ 1.17 - 1.99 ] 0.10 0.26 1.10 [ 0.59 - 1.61 ] 0.11 0.26 1.12 [ 0.61 - 1.63 ]
Parental education (No high school diploma reference) -0.57 0.65 0.56 [ -0.71 - 1.83 ] -0.90 0.79 0.41 [ -1.14 - 1.95 ] -0.05 0.65 0.95 [ -0.32 - 2.23 ]
Low-increasing (Class 1) Persistently-high (Class 2) High-decreasing (Class 3)
95% CI 95% CI 95% CI
Table 4. Odds ratios for multinomial logistic regression with
covariates
21
Table 5. Differences between adolescent-report and parent-report family conflict
Adolescent report
(n = 288)
Parent report
(n = 214)
Class Time Means SD n Means SD n t-test df p-value
1 T1 2.09 1.62 23 5.06 3.86 17 -2.98 25 0.006
T2 5.38 4.26 21 3.25 2.98 16 1.79 34 0.083
T3 11.05 1.86 21 4.47 3.40 15 13.58 30 0.000
2 T1 7.48 1.67 50 5.44 3.54 34 3.13 53 0.003
T2 7.2 3.31 40 4.04 2.71 26 4.24 57 0.000
T3 9.17 2.06 36 5.92 3.03 26 4.74 47 0.000
3 T1 10.56 2.03 32 5.62 3.88 21 5.37 33 0.000
T2 6.48 3.94 27 5.6 4.02 20 0.75 40 0.459
T3 4.09 2.65 22 5.12 4.12 17 -0.90 30 0.377
4 T1 2.21 1.94 181 3.94 3.36 142 -5.46 256 0.000
T2 2.78 2.72 161 3.27 3.11 127 -1.40 261 0.162
T3 2.89 1.88 125 3.85 2.67 102 -3.06 197 0.002
Note. Bonferroni corrections were applied for the number of tests (p <.006).
22
All of our models converged properly, and we fit one- to five-class solutions. Table 3
shows model fit indices including the indices criteria, entropy, and likelihood ratio tests, which
indicates that a four-class solution fit the data best. This four-class solution included four distinct
trajectories of family conflict (see Figure 1): low-increasing (Class 1, n = 23, 8.0%), persistently-
high (Class 2, n = 50, 17.5%), high-decreasing (Class 3, n = 32, 11.2%), and persistently-low
(Class 4, n = 181, 63.3%).
Results of our multinomial logistic regression are found in Table 4. Maltreated
adolescents were more likely than their comparison counterparts to be in the low-increasing and
high-decreasing (Class 1, OR = 1.24, 95% CI [0.00 – 2.48] and Class 3, OR = 2.98, CI [1.62 –
4.35], respectively) conflict classes, relative to the persistently-low family conflict group (Class 4
– referent group). Older age was also associated with elevated and atypical family conflict, and
adolescent males were more likely to be in the low-increasing and high-decreasing class (1 vs. 4,
OR = 1.56, CI [1.17 – 1.99] and 3 vs. 4, OR = 1.12, CI [0.61 – 1.63]).
Post hoc t-tests comparing parent-report and adolescent-report conflict showed
statistically significant differences across classes at various timepoints (see Table 5).
Specifically, in Class 1 (low-increasing), parents reported much lower conflict than adolescents
at T3 (t = 13.58, p < .001), while in Class 3 (high-decreasing), parents reported much lower
conflict at T1 (t = 5.37, p < .001) than adolescents. In Class 2 (persistently-high), parents
reported lower conflict across time relative to adolescents (t = 3.13, p = .003 at T1, t = 4.24, p <
.001 at T2, t = 4.74, p < .001 at T3). And in Class 4 (persistently-low), parents reported slightly
higher conflict, only at T3 (t = -3.06, p = .002) than adolescents.
23
Figure 1. Family conflict classes
Note. Adolescent-reported family conflict means by classes across time based on total sample
24
Discussion
This study was the first known to explore and clarify differences in the longitudinal
experiences of family conflict across early-to-mid adolescence for maltreated and comparison
adolescents. The large numbers of maltreated youth who remain at home following substantiated
maltreatment, and the known family-level risks for this population, prompted an investigation of
experiences of family conflict during adolescence. Results highlight varying levels of conflict
trajectories across early-to-mid adolescence, where maltreated adolescents are at greater risk for
experiencing elevating and atypical levels of family conflict, which contributes to our
understanding of experiences of family conflict faced by maltreated adolescents. The study of
heterogeneity in family conflict trajectories during adolescence, especially for at-risk populations
such as maltreated adolescents, can have important implications for studying heterogeneity in
associated outcomes, including psychopathology (Choe et al., 2014).
Family conflict during earlier adolescence
Addressing our first aim, we found four emergent classes of family conflict: low-
increasing, persistently-high, high-decreasing, and persistently-low. This finding lends support
that trajectories of family conflict vary by period of development and demographics. For
instance, the three studies of family conflict trajectories with older populations (i.e., Bi et al.,
2015; Castellani et al., 2014; Yu, 2019) found only three classes of family conflict, whereas
studies using younger samples found four (e.g., Choe et al., 2014; Trentacosta et al., 2011). This
is in line with previous findings that describe, on average, family conflict tends to be more
frequent during the earlier years of adolescence, but lower during later adolescent years (Branje,
2018), which may be what the additional class is representative of – more variation and
heterogeneity. This reinforces the importance of honing in on periods of adolescence in which
25
family conflict may be prone to be particularly elevated, and clarifying for which populations
this may be more likely for. This knowledge can help researchers and practitioners more
effectively tailor their approach to studying and targeting family conflict for interventions – both
by population (i.e., urban adolescents) and period of development (i.e., early versus late
adolescence).
Further, the results in our sample most closely mirrored Yu’s study (2019), which also
found a four-class solution in his younger cohort of Korean youth (ages 10 to 14), which helps
add clarity to the types of family conflict trajectories we might expect to see during early-to-mid
adolescence. Though the samples differed (e.g., country, sample size, etc.), as well as the
measure of family conflict, both groups were around 10 years old at T1, and used adolescent-
reported measures of family conflict. Perhaps given this, similar classes and proportions of the
classes on a percentage basis aligned. Regarding proportions, 17.5% of our sample were in the
persistently-high conflict class, which is a larger proportion than other studies with similar
classes (e.g., 9.4% in Trentacosta et al., 2011; 3% in Choe et al., 2014; 7.9% in Yu, 2019). This
may suggest that urban adolescents (regardless of maltreatment) may face greater persistently
elevated levels of family conflict than other populations that have been studied. This may also
reflect greater instability in the family environments of our sample, half of whom were
maltreated adolescents facing a multitude of other risks, such as greater personal and parental
mental and behavioral health symptoms (Mennen et al., 2018; Negriff et al., 2020), which could
contribute to increases in family conflict across time (Lee et al., under review). Our findings in
this area underscore the importance of accounting for family conflict trajectories in populations
who may be at greater risk for conflict (e.g., those with greater proportions in persistently-high or
26
low-increasing classes), as it highlights a stronger need for family-wide support in these families
that may persist over time.
Our substantive analysis focused on adolescent-reported family conflict, due to previous
research which indicated adolescent-report, but not parent-report of family conflict was
associated with cross-validating outcomes, such as youth depressive symptoms and wellbeing
(Lee et al., under review; Xu et al., 2017). However, post hoc analyses comparing adolescent-
report to parent-report family conflict were performed to contextualize the emergent classes,
which were based on adolescent-report family conflict. We found that parents generally seemed
to report lower family conflict across classes relative to adolescent reports, except for the
persistently-low class – which had parents reporting slightly higher means across time. The lower
average parent-report family conflict may reflect the effect of social bias, which Lau et al. (2006)
posit may be especially relevant for abusive parents. More recently, scholars have theorized
regarding the utility of discrepancies in family relations (De Los Reyes et al., 2019), and they
posit some alternative explanations of why parent and adolescent reports of family relations
differ. These explanations may aid in contextualizing the discrepancies in adolescent- and
parent-report family conflict in the low-increasing (Class 1) and high-decreasing (Class 3)
classes and parent-report means of family conflict. Parent-reported means of the low-increasing
(Class 1) class indicated that at T3, parents reported far lower conflict than adolescents. One
reason, per De Los Reyes et al. (2019), may be increasingly poor communication or a lack of
awareness by parent(s) on important difficult elements of an adolescents’ life that transpire
across time, which may lead adolescent(s) to report higher conflict than parent(s).
The parent-reported means of the high-decreasing (Class 3) class reflect a pattern more in
line with a persistently-high conflict class, with statistically significant difference between parent
27
and adolescent reports at T1. One explanation for this discrepancy between parent- and
adolescent-reported family conflict in the high-decreasing class may be due to a normative
increased push for independence and autonomy by adolescents (De Los Reyes et al., 2019).
Another explanation may be adolescent disengagement from the family; that is, adolescents who
initially report elevated levels of family conflict may be more likely to disengage or withdraw
from their families over time – possibly being blind to conflicts in the family system as a means
to cope with stress (Roubinov & Luecken, 2013). This, in turn might further parental
frustration/stress and increase the likelihood for family conflict (Kelly et al., 2016).
Alternatively, these adolescents may also be experiencing higher levels of externalizing
problems, such that they spend more time in delinquent activities (outside of the home), which
may contribute to greater reports of family conflict at home between parents/caregivers, but
lower/decreasing reports among adolescents. This is partially supported by Yu’s (2019) findings
regarding elevated conflict in the home being associated with higher baseline aggression in both
fourth and eighth grader cohorts. While difficult to fully interpret without further context, these
possibilities encourage both researchers and practitioners to more closely assess discrepancies
that may arise in varying situations of family conflict to understand their meaning.
Maltreatment and Family Conflict
Our second aim sought to examine whether child maltreatment was associated with
emergent family conflict trajectories. We hypothesized that maltreatment would be associated
with trajectories that reflected more elevated family conflict. Indeed, we found that maltreatment
was associated with two classes of family conflict during adolescence that reflected
elevated/elevating levels (i.e., Classes 1 low-increasing and 3 high-decreasing). However,
maltreated adolescents were not at greater risk for being in the persistently-high family conflict
28
class. This is further supported in the higher percentages of adolescents in the persistently-high
class. This may be a reflection of our sample as the comparison group came from the same
neighborhoods with high rates of poverty and community violence (Stevens et al., 2015).
In line with other studies that have found family-level risk factors, such as intimate
partner violence, for maltreated youth who remain at home (Campbell et al., 2012), our results
indicate that maltreated adolescents are more likely than comparison adolescents to belong to
two distinct classes of family conflict: low-increasing (Class 1) and high-decreasing (Class 3).
Given our comparison group and longitudinal design, the greater odds for maltreated adolescents
of belonging to the low-increasing class might suggest that maltreatment has a causative
influence on family conflict for some adolescents. In other words, a subset of the population who
experience maltreatment might have deteriorating familial relationships such that during and
across adolescence, even those with lower levels of family conflict experience escalating tension,
hostility, anger, and or aggression. As a caveat, a possible confounder in this relationship within
our results is age; older age was significantly associated with all the elevated conflict classes,
including the low-increasing class relative to the persistently-low conflict class, which may
reflect what others have found about increasing turbulent relationships with parents (low
support/high conflict) toward middle adolescence (around age 16; Hadiwijaya et al., 2017).
Another focus of our study was on population characteristics in relation to the class
solutions found. With regard to similarity in socioeconomic and racial/ethnic backgrounds, Choe
et al. (2014) and Trentacosta et al. (2011) had similar samples of U.S. adolescents of ethnically
diverse lower socioeconomic backgrounds. Trentacosta et al. (2011) examined conflict in a
sample of ethnically diverse low-income mother-son dyads, while Choe et al. (2014) examined
urban “at-risk” African American high school students; their numbers of youths who belonged to
29
the persistently high conflict group seemed much lower (9.4% and 3%, respectively versus our
17.5%). This underscores two related points. First, as Branje (2018) mentions, past relationship
history affects how parents and youth navigate conflicts during adolescence, such that those with
troubled relationships experience greater relational difficulties – like the majority of the
maltreated adolescents in our sample who were maltreated by either their mothers (or another
family member). Second, this result may indicate that our overall sample of maltreated and
comparison adolescents had a higher concentration of adolescents with familial difficulties than
the general population, based on the high percentage of those who belonged to the persistently-
high conflict class. The lack of association with maltreatment in the persistently-high class may
have to do with the at-risk nature of our sample; alternatively, it may imply that beyond
maltreatment, other weightier factors may be associated with persistently high family conflict
(e.g., parental mental health or substance abuse; Herrenkohl et al., 2012).
While at first glance, the subset of high-decreasing conflict class making up more than
10% of our sample presents a possibly optimistic narrative; however, it bears a deeper look. An
interpretation of this result as a hopeful sign that child welfare service involvement may decrease
family conflict levels for a subset of maltreated adolescents who start off with high levels of
family conflict seems dispelled by the post hoc t-tests. A combination of factors together may
indicate, for the high-decreasing conflict class, adolescents may be underreporting family
conflict as time passes, particularly if adolescents are experiencing familial difficulties (Li et al.,
2017). For instance, relative to the persistently-low class, this class had increased odds for
membership for older maltreated males. Considering the well-established link between
maltreated adolescent males and externalizing symptoms (Li et al., 2017; Villodas et al., 2015),
together with parent-reported mean levels of conflict along the three timepoints for this class
30
(showing trends similar to that of the persistently-high family conflict class), indicate a
discrepancy that points towards adolescent disengagement from the family. That is, this class
seems more likely to be older maltreated male adolescents who perhaps are spending less time at
home/with family, and possibly experiencing other challenges, such as externalizing problems
(e.g., Mennen et al. 2018), which are associated with interparental conflicts (Li et al., 2017).
Scenarios such as these are not uncommon and are supported in works such as Sears et al.
(2016), who found youth externalizing behaviors had spillover effects on family conflict.
Limitations
Our study has a number of strengths, including the longitudinal design and sampling from
an ethnically diverse, urban, lower socioeconomic community, which allows better
understanding of the experiences of a salient subset of maltreated adolescents. Other strengths
include a comparison group from the same neighborhoods, which helps provide controls for
potential confounders, such as neighborhood violence and gang involvement. However, this is
not a representative sample. Other limitations include a single source indicator of family conflict
for trajectories. Still, while our study was limited in our main analysis to the adolescent-report of
family conflict, post hoc analysis of parent-report of family conflict helped to bolster
interpretation of family conflict classes, albeit limited in its ability to use logit probabilities for
modal assignments. This underscores the importance of taking into account both parent- and
adolescent-report of conflict.
Our findings suggest that across earlier adolescence, particularly within urban and
ethnically/racially diverse families, adolescent-reported family conflict varies. Further,
maltreated adolescents seem to be at greater risk of experiencing elevating and atypical levels of
family conflict relative to comparison adolescents, which may put them at risk of other adverse
31
outcomes associated with family conflict. These findings highlight the importance of identifying
and addressing family conflict during and throughout this crucial period of development;
especially considering the recent promotion of preventive family services to keep maltreated
youth with families (see Family First Act of 2018; Lindell et al., 2020). Family conflict is an
important target for child welfare workers to assess and address in order to mitigate future
adverse outcomes faced by adolescents as well as their families, and this study suggests that
policy makers and practitioners take a longitudinal view that incorporates the knowledge of
increasing family conflict trajectories for a subset of maltreated adolescents into policy and
practice (e.g., by assessing family conflict periodically through adolescents as well as parents, at
different points beyond baseline intervention/investigations).
32
Chapter 3: Externalizing symptoms, family conflict, and family cohesion: Testing
competing models of between and within-person associations across adolescence for
maltreated youth
Introduction
Child maltreatment is a serious public health problem associated with a host of long-term
mental health disorders (McLaughlin et al., 2010) and familial challenges (Denholm et al., 2013).
These issues are relevant to both maltreated youth who are removed from the home, and those
who remain at home. The latter represent the majority of maltreated youth (U.S. Department of
Health and Human Services [USDHHS], 2021), and approximately 36% of these youth
experience clinical levels of externalizing symptoms, compared to an estimated 16.5% in the
general population (Campbell et al.,2012). Further, Horwitz, Hurlburt, and colleagues (2011)
report that a sizable number experience family-level risks, such as high levels of stress on the
family (49.2%), family violence (22.1%), and excessive use of physical discipline (15.1%).
Addressing mental health and family issues in these youth and their families is of critical
importance if we are to address the residual effects of child maltreatment. The focus of this paper
is on the relationships between externalizing problems, family conflict, and family cohesion in a
sample of maltreated and comparison adolescents from the same urban neighborhoods.
While many studies have examined the reciprocal relations between externalizing
symptoms and family-level challenges, few have investigated how protective factors may
mitigate downstream risk for externalizing problems and family conflict into their models, and
fewer still have addressed it in maltreated youth and their families. These relations are important
to investigate in adolescence, because of the increased risk for externalizing problems – which
includes aggression, rule breaking, defiance, and disruptiveness, (Brumley & Jaffee, 2016;
33
Hinshaw 1992). Externalizing problems have been shown to lead to worse mental health,
behavioral health, and socioeconomic outcomes in adulthood (Carter, 2019; Lee et al., 2014),
particularly for maltreated adolescents (Allen et al., 2021; Olson et al., 2021). Further, as
adolescents grow in desire for more autonomy and independence from their family, so too might
increases in incidents of general family conflict – defined as anger, tension, criticism, and
hostility among family members (Branje, 2018; Fosco, Van Ryzin, Connell, et al., 2016). Higher
levels of family conflict are known to increase risk for downstream externalizing problems
(Choe et al., 2014). In contrast, general family conflicts (i.e., not dyad-specific) have also been
shown to precede externalizing problems during adolescence (Benson & Buehler, 2012), and can
have reciprocal relations over time with externalizing symptoms such as violence (Choe &
Zimmerman, 2014). However, family cohesion – defined as feelings of emotional connectedness
and bonding that family members have toward one another (Moos & Moos, 1994; Olson et al.,
2019) – may mitigate these risks and protect adolescents against externalizing problems (Taylor
et al., 2016). The direction of influence and the temporal relationships across adolescence for
family conflict, family cohesion, and externalizing symptoms is important to clarify, as this
knowledge would help focus intervention efforts and advance our understanding of drivers of
risk and protection. In addition, despite the risks maltreated children face (Campbell et al., 2012),
no identified study has examined these relationships within a sample who remain at home. To
better understand these relationships, we simultaneously test three competing conceptual models
(i.e., interpersonal risk, symptom driven, and transactional models) to clarify directionality of
influence and reciprocal associations through examining a sample of urban adolescents – about
half of whom were maltreated and remained at home.
Externalizing problems and family conflict: Competing conceptual models
34
Family conflict is a well-documented risk factor for externalizing problems. Multiple
studies have shown increases in family conflict predicting subsequent higher externalizing
problems (Lucia & Breslau, 2006), including aggression (Karriker-Jaffe et al., 2013). Though
longitudinal studies of ethnic/racial minorities are less represented in the literature, a longitudinal
study of family conflict in African American adolescents living in a setting with twice the
homicides rates of national averages showed an association between elevated levels of family
conflict across later adolescence with future externalizing symptoms (Choe et al., 2014).
This view is represented by interpersonal risk models which posit stressors or poorly
managed social interactions, such as family conflict, antecede psychopathology for youth,
including externalizing symptoms (Patterson et al., 1989; Fosco, Van Ryzin, Xia, et al., 2016).
For instance, social learning theory (Bandura & McClelland, 1977), can also be used to explain
that adolescents learn a style of relating through conflict in their families and continue that style
outside the family, leading to bullying, aggression, or rule breaking in school settings (Lereya et
al., 2013). Further, Benson & Buehler (2012) found that hostility within the family in early-to-
mid adolescence led to externalizing aggression in subsequent years Maltreated youth may be
even more susceptible to elevated family conflict during adolescence (see Chapter 2/Lee et al.,
under review), with physical abuse and neglect in late childhood being associated with increased
aggression and rule-breaking at age 12 (Villodas et al., 2015).
On the other hand, symptom driven models situate externalizing problems (i.e.,
symptomatology) as occurring prior to future interpersonal risk outcomes, such as elevated
family conflict (Steeger & Gondoli, 2013). Zemp et al. (2018) found, adolescents’ externalizing
problems at around age 10 predicted future conflict between parents, but conflict between
parents did not predict future externalizing problems for adolescents. A study by Steeger and
35
Gondoli (2013) showed prior adolescent aggression (in 6
th
grade) was associated with future
parent-adolescent conflicts for adolescents. These studies mostly examined middle-income
European two-parent families and mother-adolescent dyads, so it is unclear whether these results
would extend to urban adolescents who have experienced maltreatment. Allen et al. (2021),
using a cross-lagged model on a large group of ethnically diverse adolescents (about half
maltreated), found that earlier maltreatment was associated with greater externalizing problems,
which in turn had cross-lagged and indirect effects with less engagement in prosocial activities
with peers. This lends some support for the symptom driven models’ applicability to vulnerable
adolescents and their interpersonal relationships, as it relates to externalizing problems driving
interpersonal risks.
Transactional models (Sameroff, 1983; Selig & Little, 2012), in contrast to the symptom
driven and interpersonal risk models, are represented in the literature through various
frameworks. For example, family systems theory posits individuals within a family not only have
a direct influence but can also have reciprocal influences on one another (Cox & Paley, 1997).
Through processes such as the bidirectional spillover effect, affective changes in, for example a
parent may influence their behaviors toward other members in the family (e.g., conflict with
adolescents); in turn, this can prompt adolescents to reciprocate similar negative mood toward
their parents (Sears et al., 2016). These reciprocal associations within families, represented by
transactional models, are capable of testing whether externalizing problems and family conflict
might have transactional effects with each other across time. Choe & Zimmerman (2014) show
these reciprocating effects between adolescent violence and family conflict in African American
youth living in urban settings with high homicide rates. Further, Cui and colleagues (2007) also
found support for the transactional model, in their study of early adolescents and their two-parent
36
families, where interparental conflict over childrearing predicted adolescent delinquency and in
turn, adolescent problems exacerbated interparental conflict over childrearing. While Choe &
Zimmerman (2014) suggest transactional relations, their measure of externalizing was explicit
violence, which only captures one kind of externalizing problem, and misses others such as
aggression and rule breaking, which do not always lead to violence.
Family cohesion
Previous research investigating longitudinal associations between family conflict, family
cohesion, and externalizing behaviors – either together or separately – provide support for each
of the three models listed above, but the majority of studies have failed to address important
limitations. First, no identified study has examined longitudinal associations between family
conflict and externalizing behaviors together with a key protective factor – family cohesion.
Family cohesion has been shown to have a buffering effect against externalizing behavior
problems in adolescence (Lucia & Breslau, 2006; Taylor et al., 2016). For adolescents who
report posttraumatic stress symptoms, moderate to high levels of adolescent-perceived family
cohesion have been shown to buffer the risk of experiencing externalizing symptoms a year later
(Deane et al., 2018). Family cohesion may mitigate risk for externalizing problems through
theories such as social control theory (Hirschi, 1969), which asserts that adolescents may be less
prone to engaging in antisocial behaviors when they feel more connected to their families. This is
seen empirically in ethnically/racially diverse populations, where improvements in short-term
parent-adolescent communication (Molleda et al., 2017) and family cohesion (Marsiglia et al.,
2009) in Hispanic families are associated with less future conduct problems in adolescents. In
contrast, low family cohesion has been associated with adverse outcomes downstream. In a
national sample of American adolescents, low levels of family cohesion were shown to be
37
associated with 1.84 times the odds of greater delinquency outcomes approximately a year later
(Barr et al., 2012). This study also showed that low levels of family cohesion, together with
witnessing community violence, was associated with future delinquency. These represent the
interpersonal risk model as it relates to the protection and risk associated with family cohesion
and externalizing problems.
In contrast, externalizing problems may erode family cohesion, which represents the
symptom driven model. For instance, in an international sample of European adolescents from
varying socioeconomic levels, Mastrotheodoros et al. (2020) found adolescents with higher
externalizing problems, a year later experienced worse family functioning, including lower
family cohesion. This study’s finding aligns with a symptom driven perspective, but
Mastrotheodoros and colleagues (2020), who conducted one of the few studies that explicitly
tested for transactional effects, did not find worse family functioning to be longitudinally
associated with worse externalizing problems in their sample of Greek adolescents. Overall,
transactional models were less represented in the literature, and we could identify no studies that
examined family conflict, family cohesion, and externalizing problems together. Also
underrepresented in the literature were studies that examined temporal relations between family
cohesion and family conflict, which are typically inversely correlated at moderate to high levels
(e.g., Fosco & Lydon-Staley, 2020), suggesting a higher-lower (vs. higher-higher) relationship
which presumably reciprocates concurrently and across time.
Between- and within-person effects
With some exceptions, the second important limitation of previous work is that most have
used models such as the Autoregressive Cross-Lagged Model (ARCL; Elam et al., 2018)/Cross-
Lagged Panel Model (CLPM; Mastrotheodoros et al., 2020), which have been the subject of
38
recent methodological scrutiny as it relates to their ability to disentangle and interpret resulting
effect sizes (Berry & Willoughby, 2017). Specifically, traditional ARCL/CLPM models produce
effect sizes that conflate between- and within-person variance into a single estimate, making it
difficult to understand the true underlying relationships. Curran and Bauer (2011) have suggested
analytic approaches need to parse out these between- and within-family variances to better
understand how changes within a family system are truly related. To illustrate, within-person
effects capture variation around an individual mean, which allows us to understand how
increases in exposure to family conflict at lower levels than one’s own typical levels (i.e., one’s
own mean over time) are associated with levels of externalizing problems. In contrast, between-
person effects capture variation from the overall average of the sample and helps explain, for
instance, how higher values of family cohesion relative to the entire sample are associated with
externalizing problems. A more appropriate type of analysis to investigate changes at both the
between- and within-person level would be the Autoregressive Latent Trajectory model with
Structured Residuals (ALT-SR; Curran et al., 2014), which disaggregates variance at these levels
to improve our understanding of cross-lagged and or reciprocal relationships across time (Berry
& Willoughby, 206; Davis et al. 2019). The ALT-SR model calculates between-person
relationships (e.g., mean levels and change/growth rates), while concurrently estimating
transactional relationships between variables at the within-person level over time (see Curran et
al., 2014 for more details).
A few current works related to externalizing problems and family functioning have tried
addressing this issue by using the Random Intercept-Cross Lagged Panel Model (RI-CLPM),
which also aims to disaggregate between- and within-person/family variance. For instance, in a
national sample of German families with adolescents (mean age 10), at the within-family level,
39
Zemp et al. (2018) found that higher than typical externalizing problems seemed to predict future
conflict among parents (regarding parenting), but conflict among parents did not predict future
externalizing problems. At the between-family level, results showed higher than average
coparenting conflict was associated with future externalizing problems in youth (Zemp et al.,
2018). On the other hand, Mastrotheodoros and colleagues (2020) found no significant
associations between family functioning (i.e., cohesion, communication and flexibility) and
aggression for Greek adolescents using the RI-CLPM; however, using a standard CLPM, they
found that externalizing problems lowered family cohesion downstream. In another study of
married Army households with children (ages 4 to 18), Knopp et al. (2017) found that at the
within-family level, overt parental conflict was associated with future parent-reported
externalizing problems in youth. While these studies aimed to disentangle between- and within-
family level differences, it is hard to discern a pattern given inconsistencies around measures and
conceptualizations of constructs such as family conflict. While dyadic measures of family
conflict (e.g., between mother-adolescent; Steeger & Gondoli, 2013) may help deepen
understandings and improve targets for intervention, they miss adolescents’ exposure to general
family conflict across multiple members in the family, which can influence adolescent
externalizing problems (Horwitz, Ganiban et al., 2011). Moreover, the generalizability of past
studies is limited given most looked at two-parent households from predominantly European and
European-American racial/ethnic backgrounds (Knopp et al., 2017; Mastrotheodoros et al., 2020;
Zemp et al., 2018). To date, we could not identify any studies parsing between- and within-
person/family variance with more vulnerable families of color, in which patterns of relationships
between family risk and protective factors, and adolescent externalizing problems may differ
(e.g., Barr et al., 2012; Choe et al., 2014; Juang & Alvarez, 2010).
40
Current study
To address these limitations and extend previous research, the aim of this study is to
clarify the between- and within-person level of longitudinal relations between family conflict,
externalizing symptoms, and family cohesion by testing three competing conceptual frameworks:
interpersonal risk, symptom driven, and transactional models. Further, given the potential
family-level and symptom driven risks faced by maltreated adolescents who remain at home
(Campbell et al., 2012), this study examines a sample of ethnically/racially diverse adolescents –
about half of whom were maltreated – to examine whether different pathways of associations
emerge. Three distinct research questions guide this study. Question 1: what are the longitudinal
between-person associations for family conflict, family cohesion, and externalizing problems
across adolescence? At the between-person level, we hypothesize higher externalizing problems
would be associated with higher baseline levels and change processes of family conflict but
lower baseline levels and change processes of family cohesion. Further, higher family conflict
would be associated with higher baseline levels and change processes for family cohesion (Fosco
& Lydon-Staley, 2020). Higher family cohesion will have a protective effect on externalizing
symptoms at the between-person level (e.g., Barr et al., 2012), but not at the within-person level
as others have found (Mastrotheodoros et al., 2020). Question 2: which conceptual model best
explains the within-person associations for family conflict, family cohesion, and externalizing
problems across adolescence? We propose that a symptom driven model in which externalizing
problems predict subsequent family conflict will be the most likely finding at the within-person
level, following Zemp et al. (2018). Question 3: do maltreated adolescents who remain home
have different pathways of association than non-maltreated adolescents? Due to the increased
likelihood of experiencing externalizing problems in maltreated adolescents (Villodas et al.,
41
2015), we hypothesize that externalizing problems will be the driver for maltreated adolescents,
while for non-maltreated adolescents we hypothesize an interpersonal risk model will be more
likely to explain within-family associations, whereby greater than one’s typical family conflict
drives future externalizing outcomes (e.g., Fosco & Lydon-Staley, 2020).
Methods
Sample
Study recruitment. Our data comes from a Young Adolescent Project, longitudinal
study of the developmental effects of maltreatment on a group of maltreated and comparison
adolescents from the same urban neighborhoods. The maltreated sample was recruited from the
Los Angeles County Department of Children and Family Services (LACDCFS) from new cases
who met criteria for the study. The comparison sample was recruited from school lists (more
details can be found in Negriff et al., 2020). The institutional review board of a private university
in Southern California, the Los Angeles County Department of Children and Family Services,
and the Los Angeles County Juvenile Court system approved contact of participants.
Participants. Four time points of the study were analyzed, with approximately a
year between Time 1 (T1) and Time 2 (T2), a year and a half between T2 and Time 3 (T3),
and approximately five and a half years between T3 and Time 4 (T4). The current study
restricted the sample to adolescents who lived with their birth parent(s) and completed
assessments at T1 and at least one assessment from T2, T3, and T4. The retention rate from
T1 to T2 was 89.9%, from T2 to T3 was 82.6%. From T1 to T4, retention rate was 79.9%.
Our final sample was 288 adolescents, with mean age of adolescents being 10.95 years old
(SD = 1.12) at entry into the study, 46.2% female, and predominantly identifying as Black
42
or Hispanic (79.2%; see Table 6). Mean age of parent was 36.49 (SD = 6.72) at T1, and
approximately 68.6% had graduated high school or beyond in their education.
Measures
Family conflict and family cohesion. The Family Environment Scale (FES; Moos &
Moos, 1994) is a 90-item self-report scale consisting of true or false agreement items. It has
been translated and adapted for cross-cultural research, with adaption in over 18 languages
(Moos & Moos, 1994). Two subscales of the FES were used for this study: conflict and
cohesion. These subscales have been shown to have fair construct validity, internal
consistency, good test-retest reliability over time, and moderate long-term stability (Moos &
Moos, 1994). The conflict subscale assesses the degree to which aggression, violence, and
conflict plays out in the family, while the cohesion subscale, assesses the degree to which
expressiveness and communication plays out in the family (Moos & Moos, 1994). These
consist of true(1)/false(0) agreement items, calculated as a sum score ranging from 0 to 14 for
general family conflict and 0 to 13 for family cohesion, with higher scores reflecting more
conflict and more cohesion. Our sample’s Cronbach’s = .85-89 (Malpha = .87) from T1-4 for
adolescent-reported conflict and = .80-86 (Malpha = .84) from T1-4 for adolescent-reported
family cohesion. Minor changes were made in wording for clarity; for example, one of our
conflict items reads, “Family members often criticize each other,” while the original reads,
“household members often criticize each other.” One of our cohesion items reads, “Family
members really stick up for each other,” whereas the original reads, “Household members
really back each other up.”
Adolescent self-report for family conflict and family cohesion was used for this study
because studies have shown adolescent and parent report of family environment/relations tap
43
into differing perspectives (De Los Reyes et al., 2019; Mastrotheodoros et al., 2020).
Adolescent reports of family conflict, however, have been shown to be more closely aligned
with a third-party outside observer’s reports than parents’ reports were (Laursen & Collins,
2009). In addition, family conflict and family cohesion as reported by adolescents have been
more closely associated with clinician’s and adolescent’s ratings of wellbeing and mood for
adolescent outcomes relative to parents’ reports (Fosco & Lydon-Staley, 2020; Xu et al.,
2017). The Intraclass Correlation Coefficient (ICC) was 0.68 for conflict and 0.67 for
cohesion, indicating that 68% and 67% of the variance, respectively for conflict and cohesion,
was due to stable differences between-individuals, while the remaining 42% and 43% can be
contributed to changes across time or variance within-individuals.
Youth externalizing symptoms. The Youth Self Report (YSR) is a self-report
instrument that has been widely used in adolescent research. It measures externalizing
behavior problems through two subscales: aggressive behavior and rule breaking
(Achenbach, 1987). The YSR has been used in many studies and has been validated in a
diverse range of populations, with good reliability and internal consistency (Achenbach &
Rescorla, 2003; O’Keefe et al., 2006). There are 14 and 10 items on these aggression and
rule breaking subscales, which are scored from a 0 to 2 scale (“not at all” to “a lot”) for
each item; total scores range from 0-28 and 0-20 respectively, with higher scores indicating
more problematic functioning. An example item from the aggression subscale is, “I threaten
to hurt people,” and from rule-breaking, “I lie or cheat.” At T3 and T4, three items for
aggressive behaviors and four items for rule breaking were not administered due to the age-
inappropriate nature of the questions; after deletion, this improved Cronbach’s in our total
sample (Peckins et al., 2018). For this study, the sample’s Cronbach’s = .81-.86 (Malpha =
44
.83) from T1 to T4 for aggression and = .73-.81 (Malpha = .77) from T1 to 4 for rule
breaking, indicating adequate to good internal consistency. The ICC for aggression was
0.73 and rule breaking, 0.46, suggesting 73% and 46% of variance was accounted for
between-person, while 26% and 54% was accounted for at the within-person level.
Covariates. Based on other works related to the population of maltreated youth
within the family studies literature and differences by demographics for family conflict and
family cohesion (Campbell et al., 2012; Choe et al., 2014; Elam et al., 2018; Lucia &
Breslau, 2006) covariates that could significantly influence the relationship between family
factors and adolescent externalizing were included. Adolescent’s age (continuous variable),
sex (male/female), parental education (did not graduate High School/HS graduate and
beyond), and maltreatment grouping (comparison/maltreated) were included.
Analytic Plan
Preliminary analyses, including descriptive data, t-tests and chi-square tests, intraclass
correlation coefficients, and Pearson’s correlations were conducted. To examine between- and
within-person effects of family conflict, family cohesion, and externalizing behavior symptoms
(i.e., aggression and delinquency separately) among our sample of urban adolescents, we used
Mplus version 8 (Muthén & Muthén, 2012) to fit a taxonomy of autoregressive latent trajectory
models with structured residuals (ALT-SR; see Figure 1) using the procedures outlined
elsewhere (Berry & Willoughby, 2016; Curran et al., 2014). This procedure allows for separation
of variance on different levels: between-person (i.e., the latent intercepts and slopes), and within-
person (i.e., auto-regressive and cross-lagged effects). The between-person effects are estimated
by correlating latent intercepts (means) and latent slopes specified as linear functions (growth
parameters/trajectories represented by blinear below), and these effects are represented as
45
φ
standardized
below. Correlating these latent intercept and slopes to capture between-person
variance then drives the within-person variance into the structured residual level of the model
represented by within-person autoregressive and cross-lagged effects.
Our model building process included examining whether random slopes were needed for
family conflict, family cohesion, and externalizing symptoms (or fixed to zero – only family
conflict in the aggression model was fixed), and if quadratic effects improved model fit (they did
not). These model comparisons were done by constraining and freeing slopes and using
likelihood ratio tests to compare fit. A measurement model with a good fit was established,
examining within-person autoregressive associations (Model 1). This was followed by
examining cross-lagged and reciprocal associations between family conflict, family cohesion,
and externalizing symptoms -- in two models for aggression and rule breaking (Model 2). Then,
using model constraint tests, we examined if constraining within-person cross-lagged effects
significantly degraded model fit relative to freely estimated models. Our model building process
indicated models in which, for the aggression and rule breaking models, all autoregressive, cross-
lagged, and within-time correlations were constrained (fixed) to be equal across time. Lastly, we
investigated how associations between family conflict, family cohesion, and externalizing
symptoms differed by maltreatment/comparison status (Model 3) via a multi-group approach. To
assess overall model fit, a number of other fit statistics were applied including the Comparative
Fit Index (CFI = .95 or greater indicates good fit), Root Mean Square Error of Approximation
(RMSEA = .05 or less indicates good fit), and Standardized Root Mean Square Residual (SRMR
= .08 or less indicates good fit). In our model building process, we included covariates but in our
final models, only significant covariates (i.e., age) were included for parsimony and power
considerations.
46
Little’s MCAR test was conducted to test for patterns of missingness. Under the missing
completely at random (MCAR) assumption, we addressed missing data (ranged from 0% to 29%
across four timepoints), utilizing the full information maximum likelihood (FIML) estimator in
Mplus, which treats all observed predictors as a single-item latent variable and allows each
individual to contribute all the data they have available to the likelihood function without
removal of individuals via list-wise deletion. We also conducted tests of differences by
maltreatment/comparison group status on demographic and study variables at baseline (i.e.,
initial levels; see Table 6).
Results
Preliminary analyses showed maltreated and comparison groups only significantly
differed on parental education at baseline (χ
2
= 32.87, p < .001), and it was included as a
covariate in our initial full models. Little’s MCAR indicated data were missing completely at
random (χ
2
= 321.79, df=317, p = .415). Pearson’s correlations between primary study variables
can be found in Table 7, with results showing significant autoregressive correlations for all
variables across time (r’s range from .27 to .60), as well as study variable means and standard
deviations (SD). Our model results are reported with unstandardized parameter estimates and
standard errors (SE) in Table 8 and Figures 1-6. Below, we report unstandardized estimates (b)
as well as standardized (ß; that are not found in the tables or figures) estimates. Our between-
person correlations are represented by φ
standardized below.
Overall Growth Trajectories
Overall mean trajectories (slopes) showed significant small to moderate increases in rule
breaking and family conflict (blinear = .41, SE = .04, p < .001 and blinear = .18, SE = .05, p < .001,
respectively), and significant small to moderate decreases in aggression and family cohesion
47
(blinear = -.12, SE = .06, p < .05 and blinear = -.35, SE = .04, p < .001, respectively). Our final
models resulted in adequate and good model fit for rule breaking and aggression, respectively
(CFI = .92, RMSEA = .07, 90% CI [.04-.07], and SRMR = .09 and CFI = .95, RMSEA = .06,
90% CI [.04-.07], and SRMR = .07). These are represented in our final model with the total
sample in Table 8.
48
Table 6. Baseline demographics
Baseline
Total Sample
(n = 288)
Maltreated
(n = 146)
Comparison
(n = 142)
χ
2
/ t
M ± SD/% M ± SD/% M ± SD/%
Age of Adolescent 10.95 ± 1.09 10.74 ± 1.10 11.06 ± 1.12 2.43
Age of Parent 36.49 ± 6.72 36.26 ± 6.27 36.71 ± 7.17 0.46
Sex of adolescent
Male 155 (53.8%) 72 (49.3%) 83 (58.5%) 2.42
Female 133 (46.2%) 74 (50.7%) 59 (41.5%)
Parental education
Did not graduate high
school
90 (31.4%) 68 (46.9%) 22 (15.5%) 32.87***
Graduated high school
and beyond
197 (68.6%) 77 (53.1%) 120 (84.5%)
Race/ethnicity
Hispanic 101 (35.1%) 56 (38.4%) 71 (50.0%) 5.56
White 30 (10.4%) 16 (11.0%) 14 (9.9%)
Black 127 (44.1%) 60 (41.1%) 41 (28.9%)
Mixed or biracial 30 (10.4%) 14 (9.6%) 16 (11.3%)
*** Indicates p is significant at < .001.
49
Correlations 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
1 T1 Agg --
2 T2 Agg .46
**
--
3 T3 Agg .38
**
60
**
--
4 T4 Agg .32
**
.32
**
.40
**
--
5 T1 RB .69
**
.26
**
.18
*
.18
**
--
6 T2 RB .28
**
.65
**
.36
**
.12 .27
**
--
7 T3 RB .28
**
.50
**
.70
**
.20
**
.19
**
.45
**
--
8 T4 RB .22
**
.24
**
.34
**
.67
**
.19
**
.13 .31
**
--
9
T1
Conflict
.40
**
.24
**
.21
**
.27
**
.35
**
.15
*
.11 .15
*
--
10
T2
Conflict
.31
**
.42
**
.30
**
.23
**
.23
**
.22
**
.26
**
.15
*
.54
**
--
11
T3
Conflict
.07 .31
**
.44
**
.24
**
.09 .25
**
.37
**
.32
**
.31
**
.44
**
--
12
T4
Conflict
.17
**
.24
**
.33
**
.50
**
.06 .15
*
.16
*
.41
**
.32
**
.23
**
.34
**
--
13
T1
Cohesion
-.34
**
-
.21
**
-
.23
**
-.10
-
.35
**
-.15
*
-.16
*
-.12
-
.50
**
-
.42
**
-
.31
**
-.15
*
--
14
T2
Cohesion
-.23
**
-
.32
**
-
.30
**
-.16
*
-
.21
**
-
.25
**
-
.21
**
-.17
*
-
.22
**
-
.48
**
-
.27
**
-.10 .57
**
--
15
T3
Cohesion
-.16
*
-
.21
**
-
.31
**
-.10
-
.27
**
-
.24
**
-
.26
**
-.17
*
-
.22
**
-
.28
**
-
.50
**
-
.19
**
.48
**
.49
**
--
16
T4
Cohesion
-.24
**
-
.19
**
-
.29
**
-
.37
**
-
.17
**
-.11
-
.22
**
-
.37
**
-
.17
**
-
.18
**
-
.28
**
-
.55
**
.25
**
.22
**
.36
**
--
MEAN 5.26 5.04 5.49 4.59 1.49 1.70 2.48 3.97 4.06 4.11 4.97 5.17 11.1 11.0 10.3 8.9
SD 4.51 4.42 4.60 4.60 2.51 2.30 2.77 3.30 3.58 3.61 3.69 4.07 2.48 2.65 3.09 3.6
Table 7. Correlations, means, and standard deviations
* Indicates p is significant at < .05, ** p < .01, *** p < .001
50
Table 8. Full model and multi-group model results
51
Unstandardized estimates reported.
* p<.05, ** p<.01, *** p<.001
a Negative two log-likelihood
b Akaike Information Criteria
c Bayesian Information Criteria
d Root Mean Square Error indices below .05 are considered to be representative of good model fit
e Standardized Root Mean Square Residual indices below .08 are considered to be representative of good model fit
f Comparative Fit Index scores above .90 are indicative of good model fit
Table 8. (continued)
52
Question 1: Between-person associations
In the final model, we also found small to moderate associations between family conflict,
family cohesion, and externalizing symptoms -- see Table 8 under “(Co) Variances (Between-
person)”. Between-person results indicate on average, adolescents who reported higher initial
levels of family conflict reported higher initial levels of aggression (φ
standardized
= .43, SE = 13. p
= .001) but a less steep conflict slope (φ
standardized
= -.49, SE = .15 p = .001). Further, on average,
those who reported higher initial levels of family cohesion reported lower initial levels of family
conflict (φ
standardized
= -.61, SE = .09 p < .001), and externalizing symptoms (φ
standardized
= -.54, SE
= .17, p = .002 for rule breaking, and φ
standardized
= -.42, SE = .12 p < .001 for aggression). Higher
initial levels of family cohesion were also associated with a less negative cohesion slope
(φ
standardized
= -.39, SE = .09, p < .001 for rule breaking, and φ
standardized
= -.30, SE = .11 p = .007
for aggression).
53
Figure 2. Total sample models
Notes. Final models reflected above. Bolded lines indicate significant pathways, whereas grey
lines indicate non-significant paths.
RB = Rule breaking symptoms (top), Agg = aggression symptoms (bottom), Con = family
conflict, Coh = family cohesion
TOTAL
TOTAL
54
Question 2: Within-person associations
For the final models for the total sample (i.e., aggression and rule breaking), the effect
sizes for within-person cross-lagged and autoregressive effects are presented in Table 8 (under
“within-person cross-lags” and “auto-regressive”) and in Figure 2. The results of the cross-
lagged pathways can be interpreted such that variables on the left of the “on” statement represent
the dependent variable. For example, “externalizing t+1 on cohesion t” translates to, the effect of
family cohesion at time t on externalizing symptoms at time t+1.
In the rule breaking model with the total sample, our within-person results reflect a
transactional model, where adolescents who reported higher than their typical levels of family
conflict reported greater rule breaking symptoms (b = .21, 95% CI [.05, .37]; ß = .18)
subsequently. Adolescents who reported higher than their typical levels of rule breaking
symptoms also reported more family conflict at subsequent timepoints (b = .16, 95% CI [.06,
.26]; ß = .18). Further, adolescents reporting higher than their typical levels of rule breaking
symptoms reported greater rule breaking symptoms (b = .27, 95% CI [.13, .41]; ß = .26), and
decreased family cohesion (b = -.13, 95% CI [-.25, -.01]; ß = -.15) at subsequent time points –
which partially represents a symptom driven model. Post hoc mediation analyses did not show
significant mediational effects through rule breaking.
In the aggression model with the total sample, our within-person results reflect a
transactional model, aggression symptoms were also associated with future aggression
symptoms (b = .19, 95% CI [0.03, 0.35]; ß = .19), and future family conflict (b = .14, 95% CI
[.04, .24]; ß = .15). Further, family conflict was associated with increases in future family
conflict (b = .18, 95% CI [.04, .32]; ß = .18).
55
Figure 3. Rule breaking by maltreatment and comparison groups
Notes. Final models reflected above. Bolded lines indicate significant pathways, whereas grey
lines indicate non-significant paths.
RB = Rule breaking symptoms, Con = family conflict, Coh = family cohesion (comparison
above, maltreated below)
COMPARISON
MALTREATED
56
Figure 4. Aggression models by maltreatment and comparison groups
Notes. Final models reflected above. Bolded lines indicate significant pathways, whereas grey
lines indicate non-significant paths.
Agg = Aggression symptoms, Con = family conflict, Coh = family cohesion (comparison above,
maltreated below)
COMPARISON
MALTREATED
57
Question 3: Multi-group: Maltreated and comparison adolescents
For estimates and figures of these models, see Table 8 and Figure 3-4.
Between-person associations showed higher initial levels of family cohesion was
correlated with less steeply declining levels of family cohesion applied across maltreated and
comparison groups (φ
standardized
= -.41, SE = .16, p = .008, φ
standardized
= -.38, SE = .12, p = .002,
respectively). Within-person associations for maltreated adolescents support a symptom driven
model in which those who reported higher than their typical levels of rule breaking reported
more rule breaking (b = .26, 95% CI [.06, .46]; ß = .26), more family conflict (b = .36, 95% CI
[.12, .60]; ß = .29), and less family cohesion (b = -.29, 95% CI [-.49, -.09]; ß = -.30) at
subsequent time points. Maltreated adolescents who reported higher than their typical levels of
aggression reported more aggression (b = .28, 95% CI [.03, .53]; ß = .28), and more family
conflict (b = .19, 95% CI [.05, .33]; ß = .23) at subsequent time points.
In contrast, within-person associations for comparison adolescents support an
interpersonal risk model, in which comparison adolescents who reported higher than their typical
levels of family conflict reported more rule breaking (b = .22, 95% CI [.08, .36]; ß = .24), and
adolescents who reported higher than their typical levels of rule breaking reported more rule
breaking (b = .29, 95% CI [.13, .45]; ß = .29) across subsequent time points. Comparison
adolescents who reported higher than their typical levels of family conflict reported more family
conflict (b = .29, 95% CI [.09, .49]; ß = .29) across subsequent time points.
Discussion
This study advances our understanding of longitudinal relationships between family
conflict, family cohesion, and externalizing symptoms by examining three competing conceptual
models, and how maltreatment impacted these pathways of association. Our results indicate that,
58
family cohesion is inversely associated with both family conflict and externalizing symptoms at
the between-person level, as previous studies have found (Barr et al., 2012; Lucia & Breslau,
2006). We found that the relationship between aggression, family conflict, and family cohesion
is best characterized by a symptom driven model at the within-person level, whereby higher than
one’s own typical levels of aggression drive future aggression and family conflict. For rule
breaking, we see evidence for a transactional model at the within-person level whereby higher
than one’s own typical levels of rule breaking for adolescents, and their levels of family conflict
share a reciprocal association across time. At the within-person level, family cohesion did not
buffer the effects of externalizing symptoms or family conflict; instead, higher than typical rule
breaking (but not aggression) was associated with lower cohesion at future timepoints. These
patterns became more clearly disaggregated in the multi-group models by maltreatment, which
show that maltreated adolescent models were more reflective of the symptom driven models –
indicating the importance of investigating between- and within-person effects for this population.
This supports findings of past studies (e.g., Elam et al., 2018; Steeger & Gondoli, 2013), but
extends them to the within-family level with a vulnerable group of maltreated adolescents who
remain at home with practice implications and suggested directions for future research.
Between-person level associations (Question 1)
As hypothesized, we see significant correlations between initial levels of externalizing
symptoms, family conflict, and family cohesion in expected ways (Barr et al., 2012; Benson &
Buehler, 2012). Specifically, higher initial levels of family conflict are associated with higher
initial levels of rule breaking, and aggression, as well as lower initial levels of family cohesion,
which is similar to what Fosco & Lydon-Staley (2020) found in a sample of two-parent
households with adolescents. Family cohesion, at the between-person level (but not at the within-
59
person level), does seem to play a protective role in its relations to initial levels of family conflict
and externalizing behaviors, which affirms past findings (Molleda et al., 2017; Taylor et al.,
2016). Unlike Marsiglia et al. (2009), who examined a group of Hispanic adolescents and found
greater family cohesion was associated with lower rule breaking, but not aggressive symptoms,
our findings mirror more the results of Molleda et al. (2017) and Taylor et al. (2016). They
examined family communication, respectively as buffers against conduct problems and
aggression in adolescents and found that higher levels of family communication and cohesion
were associated with decreased externalizing problems. Though our findings partially reinforce
and clarify past findings, they raise questions about whether past studies of family cohesion were
also tapping this between-person/family variance rather than within-person.
As for change processes, we found adolescents’ baseline levels of family cohesion were
associated with a less negative slope for cohesion, whereas their baseline levels of family
conflict were associated with a less steep inclined slope. In other words, those who started with
higher levels of family cohesion had slower decreasing trajectories over time, and those who
started with higher levels of conflict had an inclined slope but a less steep increasing trajectory.
This is similar to past findings that have suggested that there is a normative decline in certain
aspects of family relationships, including cohesion, as adolescents grow older (Tsai et al., 2013),
and points to a general pattern of “desisting” from externalizing patterns across time, which a
subgroup of adolescents do (Jennings & Reingle, 2012). We also found, as others have in
bivariate correlations (e.g., Fosco & Lydon-Staley, 2020) that slopes of family conflict and
family cohesion were inversely correlated. Our findings largely mirror the literature in this
regard.
Within-person level association (Question 2)
60
As hypothesized, our within-person findings indicate that symptom driven models
characterize the pattern of relationships between family conflict, family cohesion, and
externalizing problems in our adolescents. This supports and extends past findings: for instance,
Zemp and colleagues (2018), in a national sample of European adolescents found that at the
within-person level, externalizing problems in early adolescence were associated with more
interparental conflicts about coparenting at subsequent times. Further, a symptom driven model
during earlier adolescence seems to be in line with other works which posits adolescence as a
period when externalizing problems begin to rise – particularly for at-risk youth such as those
with maltreatment histories (Mills et al., 2013; Villodas et al., 2015). As such, it makes sense that
greater than one’s own typical levels of externalizing symptoms would drive greater family
conflict and lower family cohesion, as others have found (Elam et al., 2018; Steeger & Gondoli,
2013). These studies did not examine within-person effects, so our findings extend and
strengthen their findings, adding to the body of literature which empirically shows that
externalizing problems such as greater-than-typical aggression and rule breaking during
adolescence drives greater adverse interpersonal outcomes, such as greater family conflict and
lower family cohesion.
Second, our rule breaking model supported a transactional model, and our aggression
model supported an interpersonal risk model. In the total sample model for aggression,
significant autoregressive within-person effects of family conflict and aggression suggest that
adolescents who report greater than their typical levels of family conflict and aggression
continue to experience these across adolescence. We do not see autoregressive effects for family
cohesion and conflict (for rule breaking model). One reason may have to do with the timeframe
in which our data were collected. Others have noted, it might be more appropriate to measure
61
family conflict and family cohesion on a week-by-week or even day-by-day scale (e.g., Fosco &
Lydon-Staley, 2020), depending on timeframe and length of data collection, to observe
variability at tighter intervals, which may be more meaningful in relation to parsing out relations
with externalizing symptoms. For example, Fosco & Lydon-Staley (2020) found that daily
variability in family conflict was associated with adolescent anger.
In the rule breaking model of our total sample, we found transactional effects across time
between family conflict and rule breaking, indicating that likely higher levels of one feed higher
levels of the other. While this finding is unique and no other studies to date have found this at the
within-person level – we save an extended discussion of this for Question 3. However, in the rule
breaking model, we also saw that rule breaking drove lower family cohesion at subsequent
timepoints. This allowed for an opportunity to test whether there was a mediational effect from
family conflict that leads to rule breaking, which leads to lower family cohesion, but the indirect
effect of these paths was not significant, indicating no mediation effect. Surprisingly, this was
the only association with family cohesion at the within-person level, and it adds clarity to the
relationship between family cohesion and externalizing problems, as others have found that
family cohesion is not associated with externalizing (or internalizing) symptoms at the within-
person level (Mastrotheodoros et al., 2020).
For the aggression model, while greater aggression seems to drive future aggression, it
also seems to drive more family conflict at subsequent timepoints. Within-family level results
showed family conflict was associated with future conflict for the total sample. This is in line
with past studies that have posited family conflict may reinforce patterns of interacting within
families that perpetuate more family conflict – which could persist even across generations
(Rothenberg et al., 2016). Further, despite our initial thinking that family cohesion would buffer
62
against negative effects of family conflict and externalizing symptoms, we did not find this to be
the case. Similar to the findings of Mastrotheodoros and colleagues (2020), our within-person
autoregressive (stability) paths for family cohesion were not significant, indicating that the
protective effects of family cohesion may only be at the between-person/family level.
Multi-group models: maltreatment and comparison (Question 3)
The multi-group models further clarified patterns of within-person associations, in
maltreated adolescents, a clear symptom driven model is seen in both the aggression and rule
breaking models, but in the comparison group, an interpersonal risk model is indicated only for
rule breaking, generally confirming our hypothesis. For family conflict, family cohesion, and
externalizing symptoms, we see different conceptual models explaining the variance for
maltreated and comparison samples at the within-person level. For within-person aggression in
the comparison group, typical family conflict drove future conflict, but no other within-person
associations were apparent across time. But at the within-person level for maltreated individuals,
aggression drove future aggression as well as future conflict, indicating a symptom driven
model. For rule breaking, the overall model showed a transactional model best fit the data;
however, when examining the comparison group separately, only conflict and rule breaking led
to future rule breaking, whereas in the maltreated sample, rule breaking seemed to drive future
conflict and lower cohesion.
In comparison adolescents, variance between study variables is mostly explained by
between-person differences in initial levels of aggression with aggression slope and initial levels
of family cohesion. At the within-person level, we see family conflict continues to drive higher
future family conflict for our comparison adolescents – who may be at higher risk for family
conflict relative to non-urban adolescents (Lee et al., under review). However, for maltreated
63
adolescents, our findings suggest that a symptom driven model best explains the unique
experiences of maltreated adolescents. More specifically, we see within-person aggression
driving both future aggression as well as family conflict. This fits with research that reports
adolescents experience greater externalizing symptoms following maltreatment (Allen et al.,
2021; Villodas et al., 2015), which further drives interpersonal risks, including higher family
conflict and lower family cohesion.
Moreover, maltreated adolescents are at greater risk for externalizing problems (Mills et
al., 2013), such as aggression, through the mechanism of emotional reactivity and maladaptive
response to distress (Heleniak et al., 2016). These adolescents, who remain at home, may face
higher family stress and violence (Campbell et al., 2012), which taken together may mean that
their aggressiveness may lead to more family conflicts – especially if maltreated adolescents
experience negative emotion reactivity to stressors that are common in adolescence (e.g., role
negotiations with parents or vying for autonomy; Branje, 2018) and respond with aggression.
Evidence from neuroscience seems to lend further support to this hypothesis: maltreated
adolescents who have had early life stressors, such as physical abuse, may have smaller
amygdala and hippocampal volumes which play a role in socioemotional processing and
regulation (Hanson et al., 2015). The symptom driven model for maltreated adolescents points to
the need for greater wraparound services for those who remain with their families – particularly
those who are experiencing externalizing symptoms because these may further erode family
functioning. Since maltreated adolescents who show greater aggressiveness may drive their own
future aggression and family conflict, intervention efforts targeting aggression through programs
such as the Mindful Coping Power program may be indicated. This intervention targets
mechanisms such as reactive aggression and emotional dysregulation (Miller et al., 2020).
64
Incorporating mindfulness into existing interventions with both high-risk youth and their parents
may maximize effects as it may help potentially disrupt the aggression-family conflict links that
have been found. Moreover, these findings point to the need for prevention efforts at addressing
externalizing symptoms earlier on for maltreated adolescents.
Limitations and conclusion
Though our study contributed to examining simultaneous relationships between family
conflict, family cohesion, and externalizing symptoms in an analytic framework which
disaggregated variance explained by between- and within-person levels and examined pathways
by maltreatment, some limitations should be noted. First, our study used only adolescent self-
reported data, which may be prone to biases and shared method variance. Further, our measure of
family conflict and family cohesion captures general family conflict and cohesion as perceived
and reported by adolescents, which has strengths and limitations. Some scholars have suggested
family conflict must be viewed from multiple family members in order to most accurately gauge
impact (Rothenberg et al., 2017). Conversely, measuring general family conflict and family
cohesion can tell us about the family environment as a whole and not miss the impact of any
conflict within the family (e.g., conflict between separate sets of dyads like mother-adolescent,
father-adolescent, adolescent-sibling, etc.) which may impact individuals in the family through
spillover effects (Sears et al., 2016). We chose adolescent reports of family relations because
recent studies have found that adolescent self-reported family conflict and family cohesion are
more closely associated with adolescent outcomes (Fosco & Lydon-Staley, 2020; Xu et al.,
2017). Lastly, our sample size may have limited our ability to detect small effect sizes that did
not meet threshold for statistical significance, so replication with a larger sample may be needed.
65
Overall, this study clarifies conceptual frameworks linking family conflict, family
cohesion, and externalizing symptoms for at-risk adolescents. Further, it opens up avenues for
research for between- and within-family effects for maltreated adolescents and provides evidence
for symptom driven and interpersonal risk models in the relationships between family conflict,
family cohesion, and externalizing symptoms that differ in maltreated and comparison
adolescents living in urban settings. Our findings provide important insight into prevention and
intervention efforts in choosing intervention targets.
66
Chapter 4: Parenting attitudes and adolescent depressive symptoms in maltreated youth:
Are family conflict and cohesion mediators?
Introduction
Parenting and family environment are important elements in the development and
outcomes of youth depression (Stark et al., 2012; Yap et al., 2014). The Social Information
Processing model (SIP model), a socio-cognitive model of parenting (Milner, 1993; 2003), posits
that parenting attitudes toward childrearing have been linked to risk factors including child
maltreatment (Camilo et al., 2020a) and depression (Weed et al., 2013). Though individual-level
factors such as negative cognitions and decreasing self-esteem have been shown to mediate the
relationship between parenting attitudes and youth depressive symptoms (Park et al., 2016), less
work has examined mediating mechanisms on the family level. Family-level factors are
particularly important to consider since healthy families provide the scaffolding for adolescent
development and well-being (Branje, 2018; Xu et al., 2017). Both family conflict and family
cohesion have been linked to adolescent depressive outcomes (Weymouth et al., 2016; White et
al., 2014) and theoretically, may link parenting attitudes to adolescent depression. This study
seeks to empirically examine the mediating role of family conflict and cohesion in the
relationship between parenting attitudes and adolescent depressive symptoms. These
relationships are relevant because adolescent depression can be modified by interventions that
address family conflict, family cohesion (Fosco et al., 2016), and parenting attitudes (Clark et al.,
2013). Knowing the relationship between parenting attitudes, family conflict, cohesion, and
adolescent depression can help target interventions. In addition, given the differences in
depressive outcomes as well as the differences between parenting and family environment by
child maltreatment (Camilo et al., 2020a; Campbell et al., 2012; Norman et al., 2012) and sex
67
(Lewis et al., 2015; Kaferly et al., 2020; Salk et al., 2017), this study examines differences in
pathways by maltreatment and sex.
Parenting attitudes and adolescent depression
Adolescent depression has been on a troubling rise over the past two decades (Keyes et
al., 2019), with a recent meta-analysis estimating a prevalence rate between 10.5% to 18.6%
(Barker et al., 2019). Both being a girl and a history of maltreatment confer greater risk. A
national sample of American youth found girls had two to three times higher rates of depression
than boys (Salk et al. 2017), while maltreated youth faced one and a half to three times the rate
of depression relative to their non-maltreated peers (Norman et al., 2012). Depression can have
lasting effects. In one longitudinal study, onset of depression in adolescence was linked to a 54%
chance of recurrence within three years (Curry et al., 2011). Adolescent depression is linked to
other adverse outcomes such as failure to complete high school, unemployment in adulthood
(Clayborne et al., 2018), and heavy alcohol use (Chan et al., 2013).
The Social Information Processing model (Azar et al., 2013; Milner, 1993, 2003) posits
that parents have pre-existing schemas about parenting (i.e., attitudes), which serve as a lens
through which their caregiving environment is interpreted, particularly as it relates to
expectations of children, and their role as a caregiver (e.g., Azar et al., 2008); in turn, these are
likely to influence interpretations of children’s behaviors and needs, which then inform how a
parent responds (Milner, 2003). For certain parents, this may lead to negative attributions of
children’s behaviors and intents, placing youth at greater risk for inappropriate parenting,
including maltreatment (Azar et al., 2013; Camilo et al., 2020b). In the SIP framework, attitudes
are situated as precursors to behaviors which influence various outcomes such as adolescent
depressive symptoms – a relationship which could differ by sex (Park et al., 2016). In younger
68
children, Barnett et al. (2010) report that high and average levels of inappropriate beliefs
regarding spoiling a baby (e.g., paying too much attention) were associated with elevated
internalizing symptoms in 30-month-olds. In adolescents, a longitudinal study found rigid and
inappropriate parenting attitudes, measured using the Child Abuse Potential Inventory when a
child was 8 years old, predicted elevated depressive symptoms for youth at age 14, albeit via a
small-sized effect (Weed et al., 2013). Parenting attitudes have been shown to influence
adolescent depressive symptoms through individual level mechanisms such as increasing
negative cognitions and decreasing self-esteem (Park et al., 2016).
Adolescent depression and the family environment
Adolescent depressive symptoms are also influenced by familial and environmental
factors (Stark et al., 2012). For instance, on the family level, a lack of empathy or the use of
corporal punishment can produce changes in the family environment, such as in levels of conflict
or cohesion (Bornstein et al., 2011). While the SIP model as applied to maltreatment deals with
the parent-level antecedents of outcomes, Family systems theory posits that a family is a system
made up of individual subsystems that interact with one another and can influence change in
members through mechanisms such as the spillover and crossover effects (Cox & Paley, 1997;
Sears et al., 2016). This theoretical framework supports the notion that parenting could influence
family conflict or cohesion, which in turn can impact adolescent depressive symptoms.
Two aspects of the family environment that have been found to be important in
relationship to the development of adolescent depression are family conflict and family cohesion
(Fosco et al., 2016). Adolescence is an important period for the examination of these
relationships because families undergo salient changes that can fundamentally alter how children
and parents relate to one another. As youth undergo puberty and strive for autonomy from their
69
families, negotiations over evolving family roles increase conflicts (Branje, 2018). This may
happen in parallel with increasing rates of depression which can further be elevated by the
presence of family conflict – a process which differs for male and female adolescents
(Weymouth et al., 2016). Adolescent girls who reported elevated family conflict also reported
greater depressive symptoms relative to boys (Chan et al., 2013; Lewis et al., 2015). In contrast,
family cohesion is a source of protection for adolescents, as increases in cohesion levels have
been associated with lower depressive symptoms (Moreira & Telzer, 2015), and less global
impairment – even in clinical samples (Xu et al., 2017). A national longitudinal study also
showed adolescent females who self-reported (versus parent-reported) higher levels of cohesion
at age 12 had lower levels of depressive symptoms at age 17 (White et al., 2014). Meanwhile,
low levels of cohesion were associated with greater depressed mood in adolescents (Fosco &
Lydon-Staley, 2020). These studies suggest, high levels of conflict/low levels of cohesion can
present risk while low levels of conflict/high levels of cohesion can present protection against
adolescent depressive symptoms.
These finding suggest that more inappropriate parenting attitudes could be associated
with greater risk for increased family conflicts and decreased cohesion. Certain parenting
attitudes, such as endorsement of parental lack of empathy or more negative attribution of child’s
behaviors, or more inappropriate expectations of their youths’ ability (i.e., more inappropriate
parenting attitudes) may act in ways that increase the possibility of aggressive conflict or
decrease levels of cohesion during adolescence (Rodriguez, 2013; Rodriguez et al., 2019) which
could lead to greater risk for adolescent depressive symptoms (Yap et al., 2014). Another
example is the use of corporal punishment which may lead to parent-youth aggression or
physical conflicts (Rodriguez, 2010). Conflict and cohesion could theoretically serve as a
70
mediating link between parenting attitudes and adolescent depressive symptoms. This is relevant
within the SIP model because parenting attitudes may help shape the environments for family
interactions.
Differences by maltreatment and sex
For adolescents with maltreatment histories, these individual- and family-level risks may
be compounded given the associations between maltreatment and parenting attitudes (Camilo et
al., 2020a), family conflict (Sousa et al., 2018), cohesion (Stith et al., 2009), and adolescent
depression (Thornberry et al., 2014). The relationship between maltreatment and parenting
attitudes has been long studied (e.g., Bavolek et al., 1979), with recent reports continuing to
show maltreating caregivers having more inappropriate parenting attitudes relative to non-
maltreating caregivers (Camilo et al., 2020a; Wamser-Nanney & Campbell, 2020). Further, some
evidence suggests, parents who retain care of their maltreated youth may continue to hold
inappropriate parenting attitudes (Mennen & Trickett, 2011), which points to the need to support
efforts focused on improving parenting outcomes in the child welfare system.
After a report of maltreatment, the vast majority of the children remain with their parent.
In 2018 in the U.S, of the 656,000 youth with reports of substantiated maltreatment only 22.9%
of youth were removed from the home (USDHHS, 2021). These maltreated youth who remain at
home have high rates of mental health problems (Campbell et al., 2012). Campbell et al. (2012)
and Horwitz et al. (2011) report that a sizable number of their national samples of maltreated
youth had family-level risks at baseline: poor social support (36.3%), a lack of a supportive
caregiver in the home (53.1%), intimate partner violence (22.1%), primary caregivers with poor
parenting skills (27%), unrealistic expectations of the child (13.5%), excessive/inappropriate use
of discipline (15.1%), and high stress on the family (49.2%). This suggests that maltreated youth
71
who remain at home may be at greater risk for inappropriate parenting, family conflict, lack of
family cohesion, and depressive symptoms relative to their non-maltreated counterparts.
The research indicates family and parenting attitude risks have differential impacts on
males and females. Weymouth et al. (2016) report in their meta-analysis a greater positive
association with parent-child conflict and cohesion in adolescent female samples than adolescent
male samples. However, the literature is equivocal. In an Australian population study of 10- to
14-year-olds, Lewis et al. (2015) found that while higher levels of family conflict were
associated with more adolescent depressive symptoms, there were no significant moderation
effects by gender for family conflict. Instead, they found that adolescent females reporting lower
levels of closeness to their parents (i.e., cohesion) were 2.3 times more likely to report higher
depressive symptoms than females reporting high levels of cohesion. With regard to parenting
attitudes, in a study examining caregivers and their children (ages 0-18), parents of boys had
double the odds for endorsement of physical punishment, and almost four times lower odds for
appropriate parent-child roles (Kaferly et al. 2020). These findings seem to suggest the
relationship between adolescent sex and parenting attitudes present greater risk for inappropriate
parenting for adolescent males, while the relationship between adolescent sex and depressive
symptoms presents greater risk for females than males, particularly in the presence of family
conflict and low levels of cohesion (Lewis et al., 2015; Park et al., 2016).
Current study
The current study seeks to expand the SIP model to address parenting attitudes as it
relates to adolescent depressive outcomes by including family-level variables. Specifically, we
examine whether family conflict and family cohesion mediate the relationship between parenting
attitudes and adolescent depressive symptoms. Further, given the differences in maltreated and
72
non-maltreated, and in male and female adolescents on depressive outcomes, parenting, and
family-level risk factors, we investigate the differences by maltreatment and sex. This study
focuses on a sample of urban maltreated adolescents who remain at home following
maltreatment and peers from the same communities to help disentangle the effect of
maltreatment from other co-occurring risks such as community violence. We ask: 1) What is the
relationship between parenting attitudes on adolescent depressive symptoms? 2) Do family
conflict and family cohesion mediate the relationship between parenting attitudes and adolescent
depressive symptoms? 3) Are there differences in these relationships by maltreatment and sex?
We hypothesize that inappropriate parenting attitudes will be associated with increase in
future depressive symptoms in adolescents. Further, we hypothesize that more appropriate
parenting attitudes will be associated with greater cohesion, while more inappropriate parenting
attitudes will be associated with conflict, and future depressive symptoms (i.e., higher symptom
levels for conflict and lower for cohesion). Specifically, for maltreated adolescents and
adolescent males, we hypothesize they will face greater inappropriate parenting attitudes, which
will be associated with increases in family conflict/decreases in family cohesion; in turn these
will be associated with higher depressive symptom levels. This study has two aims: first, to
contribute to the expansion of the SIP model in the context of maltreatment to the family level by
examining the links between parenting attitudes and adolescent depressive outcomes through
family conflict and family cohesion. Our second aim is to clarify the temporal ordering of these
relations and illuminate targets to improve adolescent depressive symptoms – given the
modifiable nature of these factors through intervention (Estefan et al., 2013; Fosco et al., 2016).
73
Methods
Sample
Study recruitment. The data for this study comes from the Young Adolescent Project
(YAP), a longitudinal study on the developmental effects of maltreatment on adolescents with a
comparison sample from the same neighborhoods. The maltreated sample was recruited from the
Los Angeles County Department of Children and Family Services based on new cases who met
inclusion criteria for the study. The comparison sample was drawn from school lists. More
details on the study enrolment can be found in Negriff et al. (2020). The institutional review
board of the University of Southern California and the Los Angeles County Department of
Children and Family Services, and the Los Angeles County Juvenile Court approved contact and
data collection. Three timepoints were analyzed, with approximately a year to a year and a half
between Time 1 (T1), 2 (T2), and 3 (T3). The sample was restricted to adolescents who lived
with their birth parent(s). The parent(s) and youth completed T1 assessment and at least one T2
or T3 adolescent assessment was required for inclusion. The retention rate from T1 to T2 was
90.0%, and from T2 to T3 was 84.3%. Descriptives showed statistically significant differences
between maltreated and comparison groups (Table 9). Male and comparison adolescents were on
average about 0.3 year older. Comparison adolescents also had parents who were more educated
and older, by an average of 3 years. Our final sample was 220.
73
Demographics Total Sample
Maltreated Comparison
t/χ² tests
Male Female
t/χ² tests
N/Mean (%/SD) (n=220) (n=110) (n=110) (n=121) (n=99)
Group 3.1
Maltreated 110 (50%) -- -- 54 (44.6%) 56 (56.6%)
Comparison 110 (50%) -- -- 67 (55.4%) 43 (43.4%)
Sex 3.1
Male 121 (55.0%) 54 (49.1%) 67 (60.9%)
-- --
Female 99 (45.0%) 56 (50.9%) 43 (39.1%)
-- --
Adolescent age 10.95 (1.10) 10.79 (1.09) 11.12 (1.09) 2.25* 11.09 (1.09) 10.78 (1.10) 2.06*
Ethnicity/Race 1.51 0.26
Black 77 (35.0%) 44 (40.0%) 33 (30.0%)
42 (34.7%) 35 (35.4%)
White 24 (10.9%) 12 (10.9%) 12 (10.9%)
15 (12.4%) 9 (9.1%)
Biracial or mixed 26 (11.8%) 12 (10.9%) 14 (12.7%)
11 (9.1%) 15 (15.2%)
Hispanic 93 (42.3%) 42 (38.2%) 51 (46.4%)
53 (43.8%) 40 (40.4%)
Parental education 25.1*** 0.14
Did not grad high school
68 (30.9%) 51 (46.8%) 17 (15.5%)
36 (30.0%) 32 (32.3%)
High school grad/beyond 151 (68.6%) 58 (53.2%) 93 (84.5%)
84 (70.0%) 67 (67.7%)
Parent age 37.1 (7.24) 35.7 (7.22) 38.6 (6.99) 3.05** 37.7 (7.45) 36.4 (6.95) 1.29
MEANS (SD)
Parenting Attitudes 89.7 (18.5) 86.8 (19.7) 92.5 (16.8) 2.23* 88.1 (19.1) 91.7 (17.6) -1.38
Conflict T1
4.05 (3.58)
4.63 (3.75) 3.48 (3.33) -2.39* 4.00 (3.69) 4.11 (3.46) -0.23
Conflict T2
3.99 (3.45)
4.66 (3.36) 3.41 (3.44) -2.52** 3.88 (3.30) 4.11 (3.62) -0.46
Cohesion T1
11.10 (2.55)
11.23 (2.54) 10.96 (2.57) -0.77 11.03 (2.80) 11.18 (2.22) -0.41
Cohesion T2
11.02 (2.65)
11.00 (2.72) 11.03 (2.60) 0.08 11.21 (2.50) 10.8 (2.81) 1.07
Depression T1
8.96 (6.88)
9.89 (7.95) 8.05 (5.54) -1.98* 8.93 (7.26) 9.00 (6.43) -0.08
Depression T2
8.01 (6.08)
9.17 (6.57) 6.99 (5.44) -2.50** 8.11 (6.13) 7.89 (6.05) 0.25
Depression T3
7.82 (6.32)
8.47 (7.02) 7.34 (5.76) -1.11 6.90 (5.37) 8.79 (7.11) -1.89
Note. *p < 0.05, ** p < 0.01, *** p <0.001
Table 9. Demographics, means, and t/χ² tests for differences in study variables by maltreatment and sex
74
Measures
Parenting attitudes. Parenting attitudes were evaluated using the Adult-Adolescent
Parenting Inventory-2 Form A (AAPI-2; Bavolek & Keene, 1999). The AAPI-2 has been widely
used to evaluate parenting attitudes and as others have recently done (e.g., Camilo et al., 2020a),
we used a global assessment by taking the total sum score of all 40 items. The scale was normed
based on 1,427 adults and adolescents from a diverse population, recruited from a variety of
settings across 23 U.S. states. The measure has shown high internal consistency and discriminant
validity (Bavolek & Keene, 1999). Sample items include “children should do what they’re told to
do,” and “good children always obey parents,” and are rated on a 5-point Likert scale ranging
from strongly disagree to strongly agree (coded 0 to 4, respectively) to capture the full
continuum of attitudinal responses. Total scores can range from 0 to 160 with lower scores
reflecting more inappropriate parenting attitudes and higher scores reflecting more appropriate
parenting. (our sample’s = .88).
Family conflict and cohesion. We used the conflict and cohesion subscales of the Family
Environment Scale (FES; Moos & Moos, 1994) to measure family conflict and cohesion. These
subscales measure global family conflict and cohesion (i.e., within the entire family environment
– not in specific dyads) and have demonstrated good internal consistency as well as fair test-
retest reliability, construct, and discriminant validity (Moos & Moos, 1994). The conflict
subscale assesses the degree to which aggression, hostility, and violence is endorsed by the
reporter in the family, while the cohesion subscale assesses the degree to which the reporter sees
family members expressing emotional support and bonding (Moos & Moos, 1994). Each
subscale consists of 14 true(1)/false(0) agreement items, which are summed for a score ranging
from 0 to 14 (our sample’s = .85 at T1 and .83 at T2 for conflict, and = .82 at T1 and .83 at
75
T2 for cohesion). Minor changes were made in wording for clarity; for example, one of our items
reads, “Family members often criticize each other,” while the original reads, “household
members often criticize each other.” We used adolescent report for family conflict because
adolescent and parent report of family conflict seem to indicate differing viewpoints (De Los
Reyes et al., 2019; Mastrotheodoros et al., 2020), with adolescent perceptions of global family
conflict more closely associated with a third-party outside observer’s report (Laursen & Collins,
2009).
Depressive symptoms. The Children’s Depression Inventory (CDI; Kovacs, 1992) was
used to measure depressive symptoms. The CDI is a well-validated 27-item self-report symptom-
oriented scale. Symptoms are rated on a 0 to 2 scale, and items rate symptoms from the past two
weeks. An example item is, “I am sad all the time.” Standardization studies for the CDI have
shown good internal consistency with Cronbach’s alpha and fair test-retest reliability (Finch et
al., 1987; Kovacs, 1992). Our study used the total score, with higher scores indicating higher
levels of depressive symptoms. Our sample’s ranged from .80 to .84 (T1 to T3).
Maltreatment. Maltreatment was represented by a dichotomous variable (0 comparison/1
maltreatment). For this analysis, those who were referred to our study by child protective
services were included in the maltreatment group. And those who were recruited from the school
list, whose parents did not report any history of maltreatment, made up the comparison group
(for more details see: Negriff et al., 2020).
Covariates. Parenting attitudes, family conflict, and depressive symptoms have been
shown to vary by sociodemographic factors (e.g., Jambunathan et al., 2000; Lewis et al., 2015;
Salk et al., 2017; Wamser-Nanney & Campbell, 2020). As such, our study included as covariates
at T1, parental education which serves as a proxy for socioeconomic status (0 did not graduate
76
high school/1 high school graduate and beyond), parental age, adolescent’s age (centered around
grand mean), and sex (0 female/1 male).
Analytic plan
Preliminary analyses examined bivariate correlations as well as descriptive statistics (See
Tables 9 and 10). We conducted t and chi-square tests to examine differences between the
maltreatment and comparison group, as well as males and females on study variables and
covariates. Because our study included siblings, we randomized one sibling per parent report of
parenting attitudes. We used Mplus version 8.0 (Muthé n & Muthé n, 2018) for substantive
analysis. Two sets of structural equation models examined the relationship between parenting
attitudes and adolescent depressive symptoms, with family conflict and cohesion as mediators (in
the two separate models). This process involved inputting parenting attitudes, family
conflict/cohesion (separately), and adolescent depressive symptoms, along with covariates (i.e.,
maltreatment, age, parental age, sex, and parental education) to examine within-time associations
at T1, and then examining direct effects of parenting attitudes at T1 on subsequent adolescent
depressive symptoms – alongside covariates. This was followed by examining additional
pathways between parenting attitudes at T1 to adolescent depressive symptoms at T2-3 through
family conflict/cohesion at T2 (separate models), as well as autoregressive effects as controls for
conflict/cohesion and depressive symptoms at T1.
77
Table 10. Bivariate Correlations
1 2 3 4 5 6 7 8 9 10 11 12
1 Maltreatment
a
--
2 Age -0.15
*
--
3 Sex
b
0.12 -0.14
*
--
4 Education
c
-0.34
**
0.03 -0.02 --
5 Parent age 0.03 0.02 0.05 0.09 --
6 Parenting
Attitudes
-0.15
*
-0.03 0.10 0.32
**
0.16
*
--
7 Conflict T1
d
0.16
*
-0.01 0.02 0.02 -0.10 -0.12 --
8 Conflict T2
d
0.18
*
0.01 0.03 0.14 0.01 -0.05 0.54
**
--
9 Cohesion T1
e
0.05 -0.12 0.03 -0.05 0.05 0.06 -0.53
**
-0.39
**
--
10 Cohesion T2
e
-0.01 -0.14 -0.08 -0.13 -0.07 -0.01 -0.24
**
-0.43
**
0.58
**
--
11 Depression T1
f
0.13
*
0.04 0.01 -0.10 -0.04 -0.22
**
0.39
**
0.23
**
-0.32
**
-0.21
**
--
12 Depression T2
f
0.18
*
-0.05 -0.02 -0.06 -0.03 -0.12 0.31
**
0.48
**
-0.25
**
-0.32
**
0.54
**
--
13 Depression T3
f
0.09 -0.03 0.15 0.01 0.23
**
-0.01 0.15 0.30
**
-0.18
*
-0.35
**
0.42
**
0.59
**
Notes. *p < 0.05, ** p < 0.01
a
Maltreatment - Comparison group (referent)/Maltreatment
b
Sex – Male (referent)/female
c
Education – Parent did not graduate high school (referent)/Graduated high school or beyond
d
Family conflict – Family Environment Scale conflict subscale
e
Family cohesion– Family Environment Scale cohesion subscale
f
Adolescent depressive symptoms – Child Depression Inventory
78
PARENTING
ATTITUDES
T1
DEPRESSIVE
SYMPTOMS
T3
FAMILY
CONFLICT* /
COHESION*
T2
DEPRESSIVE
SYMPTOMS
T2
DEPRESSIVE
SYMPTOMS
T1
FAMILY
CONFLICT* /
COHESION*
T1
SEX
(1/2 FEMALE)
AGE
PARENTAL AGE
PARENTAL
EDUCATION
(0/1 HIGH SCHOOL GRAD OR BEYOND)
CHILD
MALTREATMENT
(0/1 MALTREATED)
Note. Multiple group models by maltreatment and sex excluded maltreatment and sex as covariates from their respective models.
Parenting attitudes were measured using the Adolescent Adult Parenting Inventory-2 (AAPI-2) and depressive symptoms were
measured using the Children’s Depression Inventory (CDI). *Family conflict and cohesion (measured by their respective Family
Environment subscales) were included in two separate models with all else remaining the same.
Figure 5.
79
Under the assumption of missing completely at random (MCAR), we utilized Full
Information Maximum Likelihood (FIML; Arbuckle et al., 1996) to handle missing data and the
maximum likelihood estimator (Schlomer et al., 2010). To adjust for non-normality, standard
errors were bootstrapped (500 iterations). To test mediation, we used the “model indirect”
command in Mplus (Muthé n, 2011). Finally, we conducted multiple group/multi-group analyses
using maltreatment and sex (grouping) for the two separate models (family conflict and
cohesion) and chi-square tests to examine differences in pathways between groups (Satorra &
Bentler, 2001). We assessed model fit by using the χ
2
test of model fit (where a null result at p <
.05 indicates good fit), Akaike Information Criteria (AIC) and Bayesian Information Criteria
(BIC). In addition, we evaluated fit using the following fit indices: the Comparative Fit Index
(CFI; > .90-.95), Root Mean Square Error of Approximation (RMSEA; < .06-.08), and the
Standardized Root Mean Square Residual (SRMR; < .08; Hu & Bentler, 1999).
Results
Preliminary analysis showed study variables differed by maltreatment. Broadly, the
maltreatment group had lower parenting attitude scores, higher adolescent-reported family
conflict scores, and higher adolescent depressive symptom scores at T1 and T2. Bivariate
correlations confirmed significant correlation between maltreatment and parenting attitudes,
depressive symptoms (T1 & T2), and family conflict (T1 & T2). More appropriate parenting
attitudes were correlated with older parental age, greater parental education, and lower
adolescent depressive symptoms at T1, while maltreatment was correlated with more
inappropriate parenting attitudes (Table 10).
Substantive analyses using structural equation modeling with covariates included fit our
data adequately (conflict model: χ
2
= 41.86, df = 23, p = .009, CFI = .94, RMSEA = .06, SRMR
80
= .05; cohesion model: χ
2
= 35.42, df = 23, p = .05, CFI = .96, RMSEA = .05, SRMR = .05; see
Figures 6 & 7). Parenting attitudes were significantly associated with adolescent depressive
symptoms at T1 (b = -.20, 95% CI [-.35, -.05]), but they were not significantly associated with
future depressive symptoms or family conflict or family cohesion. Adolescent depressive
symptoms were positively associated with family conflict, within-time (b = .38, [.25, .51], T1; b
= .41, [.29, .54], T2) and inversely associated with family cohesion (b = -.33, [-.52, -.15], T1; b =
-.22, [-.36, -.08], T2). Autoregressive effects controlling for conflict, cohesion, and depressive
symptoms at time t were all significantly associated at time t+1. Tests of mediation from
parenting attitudes at T1 to future depressive symptoms via conflict/cohesion at T2 yielded null
results, indicating no significant mediation in our models (indirect effect = -.02, 95% CI [-.11,
.07] via conflict; indirect effect = -.01, [-.10, .07] via cohesion); in our multi-group models
(separate models for maltreatment and sex), indirect pathways also yielded null results.
81
PARENTING
ATTITUDES
T1
DEPRESSIVE
SYMPTOMS
T3
FAMILY
CONFLICT
T2
SEX
(0/1 FEMALE)
AGE
PARENTAL AGE
PARENTAL
EDUCATION
(0/1 HIGH SCHOOL GRAD OR
BEYOND)
ETHNICITY/
RACE
(0/1 HISPANIC)
CHILD
MALTREATMENT
(0/1 MALTREATED)
FAMILY
CONFLICT
T1
DEPRESSIVE
SYMPTOMS
T1
DEPRESSIVE
SYMPTOMS
T2
PARENTING
ATTITUDES
T1
DEPRESSIVE
SYMPTOMS
T3
FAMILY
CONFLICT
T2
SEX
(0/1 FEMALE)
AGE
PARENTAL AGE
PARENTAL
EDUCATION
(0/1 HIGH SCHOOL GRAD OR
BEYOND)
ETHNICITY/
RACE
(0/1 HISPANIC)
FAMILY
CONFLICT
T1
DEPRESSIVE
SYMPTOMS
T1
DEPRESSIVE
SYMPTOMS
T2
Bold indicates significant differences between groups.
Figure 6a. Family conflict models
Figure 6b. Family conflict models by maltreatment
82
PARENTING
ATTITUDES
T1
DEPRESSIVE
SYMPTOMS
T3
FAMILY
CONFLICT
T2
AGE
PARENTAL AGE
PARENTAL
EDUCATION
(0/1 HIGH SCHOOL GRAD OR
BEYOND)
ETHNICITY/
RACE
(0/1 HISPANIC)
CHILD
MALTREATMENT
(0/1 MALTREATED)
FAMILY
CONFLICT
T1
DEPRESSIVE
SYMPTOMS
T1
DEPRESSIVE
SYMPTOMS
T2
Bold indicates significant differences between groups.
Notes. Solid lines indicate significant at the p < .05 level. Non-significant effects between study variables of interest are drawn out
in dotted lines and for ease of display, covariances between covariates as well as non-significant associations between covariates
and study variables at T1 are not shown. In multiple group models of maltreatment and sex, those covariates were excluded in
their respective models. Model fit: a) full sample (χ
2
= 41.86, df = 23, p = .009, CFI = .94, RMSEA = .06, SRMR = .05); b) multi-
group by maltreatment (χ
2
= 83.32, df = 63, p = .04, CFI = .93, RMSEA = .05, SRMR = .08); c) multi-group by sex (χ
2
= 113.89,
df = 63, p = .0001, CFI = .85, RMSEA = .09, SRMR = .09).
Figure 6c. Family conflict models by sex
83
PARENTING
ATTITUDES
T1
DEPRESSIVE
SYMPTOMS
T3
FAMILY
COHESION
T2
SEX
(0/1 FEMALE)
AGE
PARENTAL AGE
PARENTAL
EDUCATION
(0/1 HIGH SCHOOL GRAD OR
BEYOND)
ETHNICITY/
RACE
(0/1 HISPANIC)
CHILD
MALTREATMENT
(0/1 MALTREATED)
FAMILY
COHESION
T1
DEPRESSIVE
SYMPTOMS
T1
DEPRESSIVE
SYMPTOMS
T2
PARENTING
ATTITUDES
T1
DEPRESSIVE
SYMPTOMS
T3
FAMILY
COHESION
T2
SEX
(0/1 FEMALE)
AGE
PARENTAL AGE
PARENTAL
EDUCATION
(0/1 HIGH SCHOOL GRAD OR
BEYOND)
ETHNICITY/
RACE
(0/1 HISPANIC)
FAMILY
COHESION
T1
DEPRESSIVE
SYMPTOMS
T1
DEPRESSIVE
SYMPTOMS
T2
Bold indicates significant differences between groups.
Figure 7a. Family cohesion models
Figure 7b. Family cohesion models by maltreatment
84
PARENTING
ATTITUDES
T1
DEPRESSIVE
SYMPTOMS
T3
FAMILY
COHESION
T2
AGE
PARENTAL AGE
PARENTAL
EDUCATION
(0/1 HIGH SCHOOL GRAD OR
BEYOND)
ETHNICITY/
RACE
(0/1 HISPANIC)
CHILD
MALTREATMENT
(0/1 MALTREATED)
FAMILY
COHESION
T1
DEPRESSIVE
SYMPTOMS
T1
DEPRESSIVE
SYMPTOMS
T2
Bold indicates significant differences between groups.
Notes. Solid lines indicate significant at the p < .05 level. Non-significant effects between study variables of interest are drawn
out in dotted lines and for ease of display, covariances between covariates as well as non-significant associations between
covariates and study variables at T1 are not shown. In multiple group models of maltreatment and sex, those covariates were
excluded in their respective models. Model fit: a) full sample (χ
2
= 35.42, df = 23, p = .05, CFI = .96, RMSEA = .05, SRMR =
.05); b) multi-group by maltreatment (χ
2
= 82.54, df = 58, p = .02, CFI = .91, RMSEA = .06, SRMR = .08); c) multi-group by sex
(χ
2
= 111.83, df = 63, p = .0001, CFI = .84, RMSEA = .08, SRMR = .10).
Figure 7c. Family cohesion models by sex
85
For main effects in the multiple group models, we found significant differences in
maltreated adolescents, such that more appropriate parenting attitudes were associated with
lower depressive symptoms at T1 (b = -.31, [-.54, -.08]). This was not significant in comparison
adolescents. While maltreated and comparison adolescents differed in their relationship between
family conflict and parenting attitudes at T1, as well as between cohesion and parenting attitudes
at T1, these associations were not significant at the p < .05 level. As in the conflict model, the
cohesion model found maltreated and comparison adolescents differed with regard to parenting
attitudes and depressive symptoms, but maltreated adolescents had strong inverse associations
for depressive symptoms and cohesion (b = -.44, [-.68, -.20], T1; b = -.29, [-.51, -.07], T2),
which were significant at the p < .01 level, and significantly differed from comparison
adolescents. These pathways were not significant for the comparison adolescents. For maltreated
adolescents, an increase in cohesion at T2 was associated with a decrease in depressive
symptoms at T3 (b = -.26, [-.44, -.09]), but not comparison adolescents (b = .00, [-.16, .17]). The
total Δχ
2
was 30.33 (7) for the maltreatment group’s conflict model, while for the cohesion
model, the total Δχ
2
was 16.96 (5), which are both significant at p < .01.
Multi-group models by sex showed no significant differences between male and females
on most pathways of interest with the exception of maltreatment on conflict and depressive
symptoms at T1. For males, maltreatment was significantly associated with greater reports of
family conflict and depressive symptoms at T1 (b = .42, [.26, .48]; b = .20, [.04, .36], significant
at p < .001 and p < .05 level, respectively). However, these associations failed to reach
significance in the female group. The total Δχ
2
was 15.85 (3) for this group’s conflict model,
while for the cohesion model, the total Δχ
2
was 17.80 (3), which are both significant at p < .01.
Discussion
86
The goal of this study was to expand the SIP model through a family systems framework
to examine the mediating influence of family factors on the relationship between parenting
attitudes and adolescent depressive symptoms. Our findings suggest that parenting attitudes and
family cohesion are inversely associated with depressive symptoms when adolescents are
younger, while family conflict is positively associated with depressive symptoms, but these do
not seem to predict subsequent depressive symptoms. This conflicts with other studies that have
found parenting attitudes do influence adolescent depression (Park et al., 2016; Weed et al.,
2013). Despite this, we believe we are still able to glean insights from these findings and inform
future interventions to prevent adolescent depression. While the relationships between parenting
attitudes and family conflict and cohesion were null, this study found that cohesion may play an
important and possibly protective role in future depressive symptoms for maltreated adolescents.
This study’s first question examined the relationship between parenting attitudes and
adolescent depressive symptoms and found there was an association, but only within time (T1).
Lower/more inappropriate parenting attitude scores were related to higher depressive symptoms
for maltreated relative to comparison adolescents (multi-group model). This signals that either,
the more inappropriate the attitudes, the greater the depressive symptoms or alternatively, the
lower depressive symptoms, the more appropriate the attitudes. Cross-sectional association
allows for speculation of either interpretation, as does the SIP model, which posits parenting
attitudes precede behaviors, which are informed by the context (Milner, 2003). That is, though
parenting attitudes may exert influence on depressive symptoms, adolescent depressive
symptoms and other unobserved parenting stressors may disrupt the parenting attitudes-
depression connection such that parenting attitudes play a smaller role in outcomes. For instance,
Barnett et al. (2010) found youth who were depressed were seen as more needy by parents who
87
strongly endorsed parenting attitudes that believed in producing independent youth (i.e., not
spoiling them), which meant giving the youth less attention.
However, moving beyond the within-time association, we did not find parenting attitudes
predicted T2 or T3 adolescent depressive symptoms in contrast to others' findings (i.e., Park et
al., 2016; Weed et al., 2013). This null finding might be explained in different ways. First, our
rigorous longitudinal study design controlled for autoregressive effects of depressive symptoms,
which indicated a moderate level of stability across time (b = .53 to .59 from T1 to T3). Past
depressive symptoms are not surprisingly a strong predictor of subsequent depressive symptoms;
however, this autoregressive effect may also serve to attenuate otherwise larger effect sizes from
influences that are, in actuality, smaller than observed (without the autoregressive effects;
Adachi & Wiloughby, 2015) – such as with parenting attitudes. This possibility has support in
the literature (e.g., Weed et al., 2013), as well as in our findings, given the relatively small
association between parenting attitudes and depressive symptoms at T1 and the decreasing (and
non-significant) effect size from T1 to T2 and T3. This may suggest that though parenting
attitudes may have some influence on future adolescent depressive symptoms, the inclusion of
other more salient variables (e.g., past depressive symptoms) renders their signal statistically
undetectable. Another possibility is that depressive symptoms may be driving parenting attitudes,
which has been previously reported by Wamser-Nanney & Campbell (2020). This is an example
of the symptom driven model (Joiner et al., 1999; Rudolph et al., 2008), which in various studies
have shown adolescent depressive symptoms to be drivers of other family risks – such as family
conflict (Lee et al., under review). However, this interpretation is not explicitly supported by our
analysis, given we did not test whether depressive symptoms impacted future parenting attitudes.
88
Second, this study sought to examine mediating roles of family conflict and family
cohesion in the relationship between parenting attitudes and adolescent depressive symptoms.
While parenting attitudes did not predict depressive symptoms, family conflict and family
cohesion were associated with depressive symptoms within time (at T1 and T2). This partially
confirms our hypotheses regarding conflict and cohesion being associated with depressive
symptoms but fails in predicting association between parenting attitudes and conflict, and family
conflict on subsequent depressive symptoms. Interestingly, family cohesion at T2 does predict
lower depressive symptoms at T3 – but only for maltreated adolescents. There may be a few
explanations for this interesting finding, which may help expand the SIP model via the family
system framework, particularly during the period of adolescence. According to the SIP model,
parents with more inappropriate attitudes access biased and negative schemas to attend to cues
that require parental responses (Milner, 2003). While theoretically, inappropriate parenting
attitudes may trigger arguments or conflicts between parent and adolescent about autonomy
(Branje, 2018), high family conflict may trigger greater negative responses in parenting as well.
For instance, Estefan et al. (2013) found conflict between parents might activate harsher
parenting attitudes toward adolescents due to stress; in this case, family conflict is an antecedent
to greater inappropriate parenting attitudes, which may in turn increase youth depressive
symptoms (Franck & Buehler, 2007). However, our within-time association between T1
parenting attitudes and family conflict and cohesion does not support any significant association
(or such a small association that it does not reach significance), making the possibility regarding
reverse directionality less likely.
Estefan et al. (2013) also found that family conflict may be between different members
within the family, which may also explain why we do not see parenting attitudes and family
89
conflict associated. That is, certain conflicts may not necessarily trigger negative schemas that
parents attend to. In our sample, half our sample were maltreated, and almost all lived in lower
income households in urban environments (Negriff et al., 2020). As such, relationships between
members other than the parent-adolescent dyads (e.g., between parents/partners) may be strained,
leading to greater conflict but not resulting in inappropriate parenting attitudes that affect the rest
of the family. This may help explain the lack of association between parenting attitudes and
conflict/cohesion. Alternatively, our sample included parents with high rates of depressive
symptoms (Mennen & Trickett, 2011) – which is not uncommon for parents who have contact
with the child welfare system (Horwitz et al., 2011). Depression may have impaired parental
responsivity to cues in their environment (Lovejoy et al., 2000), which may disrupt the link
between parenting attitudes and family conflict or cohesion.
Other possibilities for the lack of association between parenting attitudes and family
conflict and cohesion may have to do with developmental tasks of adolescence. As they grow,
adolescents strive for more autonomy and independence from their parents, and as this happens,
they may rely on themselves more than parents (Branje, 2018; Pinquart, 2017). The associations
at T1 between parenting and family conflict/cohesion, during the younger years of adolescence,
seems to decrease in size at T2; this may be suggestive of a growing disconnect – presumably as
adolescents continue growing in independence. Moreover, this may be particularly true of
maltreated and neglected youth, who may be reluctant to rely on their parents or other caretakers
or have grown accustomed to relying on themselves and not others (Samuels & Pryce, 2008).
Indeed, there is some evidence that maltreated adolescents find less social support from their
families (Negriff et al., 2019); and if family interactions lead to less support, particularly for
more depressed adolescents, this may further discourage them from engaging their families as
90
they grow more independent/autonomous. This explanation finds empirical support in our
significant within-time T1 differences between the maltreatment and comparison adolescents.
Although, likely due to power, none of these parameters reach significance at the p < .05 level,
parenting attitudes and conflict and cohesion are associated in the expected direction for
comparison adolescents (i.e., more appropriate parenting attitudes, less conflict/greater
cohesion), relative to associations for maltreated adolescents (closer to zero association). This
may speak importantly to a difference in family relations between maltreated adolescents who
remain at home and comparison adolescents. That is, as maltreated adolescents age, many may
be less influenced by parenting attitudes than non-maltreated adolescents, which may be a reason
why parenting attitudes do not predict adolescent depressive symptoms.
The mediation hypotheses of the second question also resulted in null findings.
Specifically, this aim sought to build family-level understanding to the SIP model of parenting
attitudes explaining youth outcomes. But the null results for family conflict and family cohesion
mediating the relationship between parenting attitudes and adolescent depressive symptoms is
understandable, given the above explanations about attitudes not predicting family conflict or
family cohesion. Another possibility for why this family-level mediation was not seen may relate
to individual-level processes being more pertinent during adolescence than family-level
processes. For instance, Park et al. (2016) found, negative automatic thoughts and self-esteem
mediated the association between parenting attitudes and adolescent depressive symptoms,
further lending support that past depressive symptoms may account for far more variance than
the possibly diminishing effect of parenting attitudes over time during adolescence. As for
differences by sex, counter to our hypothesis, maltreated male adolescents seemed to fare worse
on T1 depressive symptoms and family conflict. Without consideration of maltreatment, we
91
found no differences between males and females in our sample related to study variables (see
Table 9). This was initially surprising, given the literature on female risk being greater, but made
sense in the context of maltreatment, which likely poses a greater risk for worse outcomes
(Jonson-Reid et al., 2012) over the risk conferred by sex alone. Overall, households with
maltreated adolescents and lower parental education seemed to be more vulnerable to
inappropriate parenting attitudes, greater conflict, and depressive symptoms.
Interestingly, this study found family cohesion significantly associated with lower
depressive symptoms in the future, serving as a potential protective factor for maltreated but not
comparison adolescents. This has salient implications for our population of interest, given family
cohesion can vary, even for families that have multiple risk factors; this is partially evidenced by
high mean levels of cohesion in both the maltreated and comparison groups, which notably did
not differ significantly between the two groups. Cohesion and depressive symptoms were
inversely associated for maltreated adolescents, which differed for comparison adolescents.
Further, no such difference between maltreated and comparison adolescents were found for
family conflict and depressive symptoms. This indicates, though family cohesion may not often
be seen as a target for intervention for this population, these findings suggest that perhaps even
more so than family conflict, cohesion may be a better target to improve adolescent depressive
symptoms. This idea is supported by others who found cohesion was protective against
depressive symptoms and global impairments, even in clinical samples (Xu et al., 2017), and is a
finding we believe is important to share.
Limitations and Conclusion
The present study has a number of strengths as well as limitations. The rigorous
longitudinal, comparison group design that controlled for covariates and autoregressive effects
92
was a strength. However, the model may have been underpowered to sufficiently detect small
effect sizes (e.g., the effect from parenting attitudes to depressive symptoms), since others have
shown when controlling for autoregressive effects, main effects may become smaller (Adachi &
Wiloughby, 2015). Further, our measure of family conflict covers global family conflict, but was
measured only using adolescent report, which may be sensitive to reporter bias. Relatedly, Lau et
al. (2006) have noted that self-report measures of parenting, particularly when related to
maltreatment, can also be prone to social bias; a recent study showed a small but statistically
significant mean-level difference of parent self-reported parenting as being more positive relative
to adolescent-reported parenting (Hou et al., 2020). This may also indicate parents may have
reported slightly more appropriate parenting than may be practiced, so when possible, future
studies may want to use multiple methods for gauging parenting attitudes and practices (e.g.,
Camilo et al., 2020). This study only included maltreated adolescents who remained at home,
which is a special subset of the population (albeit a large subset), limiting the generalizability of
this study’s findings to all maltreated children. Future research might test other mediators (e.g.,
specific adolescent-parent conflict), using longitudinal designs that examine different periods of
adolescence and or focus on moderators of family cohesion for depressive outcomes of
maltreated adolescents.
This study speaks to the importance of considering multiple contexts and adolescent
development to expand theories related to parenting and families – in particular, the SIP model in
the context of child maltreatment, family systems, and parenting attitudes for adolescents. These
findings suggest that despite parenting attitudes having an attenuated influence on adolescent
depressive symptoms over time, interventions targeting the increase of family cohesion, for
example via Interpersonal Therapy for adolescents (IPT-A; Toth et al., 2020) may be effective
93
for decreasing depressive symptoms – especially for maltreated adolescents. These findings may
speak to priorities when dealing with maltreated adolescents who remain at home indicating that
family-wide interventions may be more productive than focusing solely on parenting.
94
Chapter 5: Implications and Conclusion
Child maltreatment has negative consequences at the individual, family, and societal
level (Denholm et al., 2013; McLaughlin et al., 2010; Peterson et al., 2018). Past studies, using
nationally representative samples have shown that maltreated youth who remain at home
experience negative mental health and family-level risks, including greater odds relative to the
general population of externalizing and internalizing problems, high levels of family stress, poor
parenting attitudes and skills, as well as intrafamilial violence (Campbell et al., 2012; Horwitz,
Hurlburt, et al., 2011). However, extant studies have been limited in their understanding of the
longitudinal relationships between family factors and adolescent mental health. More
specifically, family conflict is a risk factor for maltreatment (Stith et al., 2009; Vial et al., 2020)
but it is unknown whether maltreated adolescents are at greater risk than non-maltreated
adolescents of facing elevated or atypical levels of family conflict. Moreover, while many
researchers have examined the relationship between family conflict and externalizing problems
during adolescence (e.g., Barr et al., 2012; Choe et al., 2014; Elam et al., 2018), the directionality
of this relationship remains unclear, with studies showing mixed results. And the majority of
studies that have been conducted, with some exceptions (Knopp et al., 2017; Mastrotheodoros et
al., 2020; Zemp et al., 2018), have not disaggregated the variance at the between- and within-
person level to clarify understanding of the true underlying relationships between family conflict,
family cohesion, and adolescent externalizing symptoms. Further, past studies have failed to
simultaneously consider how the important protective factor, family cohesion, may affect the
relationship between family conflict and externalizing problems. Relatedly, despite the important
proximal influence of family conflict and family cohesion on adolescent depression (White et al.,
2014; Yap et al., 2014), still unexamined is whether conflict and cohesion mediate the
95
relationship between parenting attitudes and adolescent depressive symptoms. This dissertation,
through three separate but related studies, addressed these limitations in past studies to inform
future research, practice, and policy.
The first study (Chapter 2) examined the trajectories of family conflict in earlier
adolescence, to test whether maltreated adolescents were at greater risk than their non-maltreated
peers for elevated and or atypical levels of family conflict. The results show across earlier
adolescence, within urban and ethnically/racially diverse families, adolescent-reported family
conflict might be categorized into four classes: low-increasing (8.0%), persistently-high (17.5%),
high-decreasing (11.2%), and persistently-low (63.3%). Further, maltreated adolescents were
more likely to face elevating and atypical trajectories of family conflict – specifically more likely
to belong to low-increasing and high-decreasing classes relative to non-maltreated adolescents.
These experiences across adolescence for these maltreated youth may present great risk for
negative mental health outcomes in the future (Choe et al., 2014). Thus, practitioners may want
to identify and intervene with families who may be more at-risk for these elevated and atypical
trajectories. Policy makers should also take note that maltreated adolescents are at greater risk
for increasing family conflict, and craft policies that take a longitudinal view of engagement and
support with these families – especially in light of increased efforts at family preservation
(Lindell et al., 2020. This may entail a requirement for more frequent periodic assessments of the
entire family, gathering viewpoints of family conflict from family members to help inform
practitioners.
The second study (Chapter 3) aimed to understand the relationship between family
conflict, family cohesion, and externalizing symptoms, disentangling between- and within-
person level effects and examining differences in models by maltreatment and comparison
96
adolescents. One of the aims, to test three competing conceptual models simultaneously (i.e.,
interpersonal risk, symptom driven, and transactional models), to examine within-person effects
revealed that though in the total sample, transactional effects emerged in the rule breaking
model, when parsed by groups, maltreated adolescent models reflected a symptom driven for
externalizing symptoms, whereas comparison adolescent models reflected an interpersonal risk
model, where family conflict drove future family conflict and future rule breaking. Given past
findings that show elevated risk for externalizing problems in maltreated adolescents (Allen et
al., 2021; Campbell et al., 2012), this finding can support practitioners by clarifying a specific
target for this particular population. To this end, practitioners could also note that recent
evidence shows youth with externalizing problems may be effectively treated by interventions
that target mechanisms such as emotion regulation through use of mindfulness interventions
(Miller et al., 2020).
In the third study (Chapter 4), family conflict and cohesion were further examined as
mediating factors in the relationship between parenting attitudes and adolescent depressive
symptoms. Studies reporting variations in the relationship between parenting attitudes and
adolescent depressive symptoms (Park et al., 2016), family conflict and depressive symptoms
(Lewis et al., 2015), as well as family cohesion and depressive symptoms (Fosco et al., 2016), by
sex (Salk et al., 2017; Weed et al., 2013) and maltreatment (Kaferly et al., 2020; Norman et al.,
2012), also prompted investigation into whether these pathways would differ by these respective
groups. The findings show that while parenting attitudes and adolescent depressive symptoms
were associated in earlier years, the effect fades over time confirming previous research (Weed
et al., 2013). This also supports recent reports that during adolescence, “child effects” may
become stronger as “parent effects” wane on adolescent outcomes (De Los Reyes et al., 2019).
97
Further, family cohesion was shown to be protective against future adolescent depressive
symptoms for maltreated adolescents, suggesting a target for intervention for maltreated
adolescents who may be at risk for depression.
These findings, taken together, suggest a number of important considerations for
practitioners and policy makers working to improve the lives of maltreated adolescents. In light
of the recent passage of the Families First Prevention and Services Act of 2018, which provides
federal funding to states to prevent foster care placement via increases for family preservation
services (Lindell et al., 2020), policy makers and child welfare practitioners may want to
consider the longitudinal impacts of family conflict, and shape policy accordingly. Family
conflict can have a negative impact of the long-term wellbeing of adolescents (Benson &
Buehler, 2012; Choe et al., 2014; Karriker-Jaffe et al., 2013), and based on the findings of Study
1, for families that are screened for further services, a consideration for longer follow-up periods
to assess for increasing family conflict levels during adolescence seems appropriate. Practitioners
may want to collect multiple sources of reports for family conflict, which may differ between
parent and adolescent reports, as others have also found (De Los Reyes et al., 2019). Based on
our findings, we believe divergent reports may be a flag for further assessment and intervention,
as seen as appropriate by practitioners. In conjunction, based on findings from Study 2,
practitioners are encouraged to assess for and treat externalizing symptoms in maltreated
adolescents, as this may drive future family conflict and decrease family cohesion. Incorporating
mindfulness skills, which has an expanding base of empirical support, into pre-existing programs
may be appropriate. For instance, the Mindful Power Coping program (Miller et al., 2020), an
empirically supported intervention, may be useful for maltreated adolescents with externalizing
problems. Future research should aim to disaggregate between- and within-family effects to
98
better clarify the drivers of within-person and -family effects in order to improve intervention
efforts (e.g., Berry & Willoughby, 2017).
Finally, the findings of Study 3 that show parenting attitudes had a waning influence on
adolescent depressive symptoms in later years. Further, since family cohesion was protective
against depressive symptoms, interventions to address maltreated adolescents’ depressive
symptoms may want to target family cohesion as a mechanism for change through evidence-
based interventions such as Interpersonal Psychotherapy for Depressed Adolescents (IPT-A)
focusing on strengthening family relationships (Toth et al., 2020). These studies have aimed to
provide a longitudinal view of the family-level influences and mental health symptoms and their
relationship with one another in maltreated adolescents, with the goal of promoting effective
interventions and policies for this at-risk population.
99
References
Adachi, P., & Willoughby, T. (2015). Interpreting effect sizes when controlling for stability
effects in longitudinal autoregressive models: Implications for psychological
science. European Journal of Developmental Psychology, 12(1), 116-128.
https://doi.org/10.1080/17405629.2014.963549
Allen, E. K., Desir, M. P., & Shenk, C. E. (2021). Child maltreatment and adolescent
externalizing behavior: Examining the indirect and cross-lagged pathways of prosocial
peer activities. Child Abuse & Neglect, 111, 104796.
Arbuckle, J. L. (1996). Full information estimation in the presence of incomplete data. In
Marcoulides, G. A., Schumacker, R. E. (Eds.), Advanced structural equation modeling
(pp. 243–277). Mahwah, NJ: Lawrence Erlbaum.
Azar, S. T., Okado, Y., Stevenson, M. T., & Robinson, L. R. (2013). A preliminary test of a
social information processing model of parenting risk in adolescent males at risk for later
physical child abuse in adulthood. Child Abuse Review, 22(4), 268-286.
https://doi.org/10.1002/car.2244
Azar, S. T., Reitz, E. B., & Goslin, M. C. (2008). Mothering: Thinking is part of the job
description: Application of cognitive views to understanding maladaptive parenting and
doing intervention and prevention work. Journal of Applied Developmental Psychology,
29(4), 295-304. https://doi.org/10.1016/j. appdev.2008.04.009
Bandura, A., & McClelland, D. C. (1977). Social learning theory (Vol. 1). Prentice Hall:
Englewood cliffs.
100
Barker, M. M., Beresford, B., Bland, M., & Fraser, L. K. (2019). Prevalence and incidence of
anxiety and depression among children, adolescents, and young adults with life-limiting
conditions: a systematic review and meta-analysis. JAMA Pediatrics, 173(9), 835-844.
doi:10.1001/jamapediatrics.2019.1712
Barnett, M. A., Shanahan, L., Deng, M., Haskett, M. E., & Cox, M. J. (2010). Independent and
interactive contributions of parenting behaviors and beliefs in the prediction of early
childhood behavior problems. Parenting: Science and Practice, 10(1), 43-59.
https://doi.org/10.1080/15295190903014604
Barr, S. C., Hanson, R., Begle, A. M., Kilpatrick, D. G., Saunders, B., Resnick, H., & Amstadter,
A. (2012). Examining the moderating role of family cohesion on the relationship between
witnessed community violence and delinquency in a national sample of
adolescents. Journal of Interpersonal Violence, 27(2), 239-262.
Bavolek, S. J., & Keene, R. G. (1999). Adult-Adolescent Parenting Inventory-AAPI-2:
Administration and development handbook. Park City, UT: Family Development
Resources, Inc.
Bavolek, S. J., Kline, D. F., McLaughlin, J. A., & Publicover, P. R. (1979). Primary prevention
of child abuse and neglect: Identification of high-risk adolescents. Child Abuse &
Neglect, 3(3-4), 1071-1080. https://doi.org/10.1016/0145-2134(79)90152-2
Benson, M. J., & Buehler, C. (2012). Family process and peer deviance influences on adolescent
aggression: Longitudinal effects across early and middle adolescence. Child
Development, 83(4), 1213-1228. DOI: 10.1111/j.1467-8624.2012.01763.x
101
Berlin, K. S., Parra, G. R., & Williams, N. A. (2014). An introduction to latent variable mixture
modeling (part 2): Longitudinal latent class growth analysis and growth mixture models.
Journal of Pediatric Psychology, 39(2), 188-203.
Berry, D., & Willoughby, M. T. (2017). On the practical interpretability of cross‐lagged panel
models: Rethinking a developmental workhorse. Child Development, 88(4), 1186-1206.
https://doi.org/10.1111/cdev.12660
Bi, X., Moos, R. H., Timko, C., & Cronkite, R. C. (2015). Family conflict and somatic symptoms
over 10 years: a growth mixture model analysis. Journal of Psychosomatic Research,
78(5), 459-465. https://doi.org/10.1016/j.jpsychores.2015.01.013
Bornstein, M. H., Putnick, D. L., & Lansford, J. E. (2011). Parenting attributions and attitudes in
cross-cultural perspective. Parenting, 11(2-3), 214-237.
https://doi.org/10.1080/15295192.2011.585568
Branje, S. (2018). Development of parent–adolescent relationships: Conflict interactions as a
mechanism of change. Child Development Perspectives, 12(3), 171-176.
https://doi.org/10.1111/cdep.12278
Brumley, L. D., & Jaffee, S. R. (2016). Defining and distinguishing promotive and protective
effects for childhood externalizing psychopathology: a systematic review. Social
Psychiatry and Psychiatric Epidemiology, 51(6), 803-815.
https://doi.org/10.1007/s00127-016-1228-1
Camilo, C., Garrido, M. V., & Calheiros, M. M. (2020a). Parental Attitudes in Child
Maltreatment. Journal of Interpersonal Violence. doi: 10.1177/0886260520943724
102
Camilo, C., Garrido, M. V., & Calheiros, M. M. (2020b). The social information processing
model in child physical abuse and neglect: a meta-analytic review. Child Abuse &
Neglect, 108, 104666. https://doi.org/10.1016/j.chiabu.2020.104666
Campbell, K. A., Thomas, A. M., Cook, L. J., & Keenan, H. T. (2012). Longitudinal experiences
of children remaining at home after a first-time investigation for suspected maltreatment.
The Journal of Pediatrics, 161(2), 340-347. https://doi.org/10.1016/j.jpeds.2012.02.039
Carter, A. (2019). The consequences of adolescent delinquent behavior for adult employment
outcomes. Journal of Youth and Adolescence, 48(1), 17-29.
https://doi.org/10.1007/s10964-018-0934-2
Castellani, V., Pastorelli, C., Eisenberg, N., Gerbino, M., Di Giunta, L., Ceravolo, R., & Milioni,
M. (2014). Hostile, aggressive family conflict trajectories during the transition to
adulthood: Associations with adolescent Big Five and emerging adulthood adjustment
problems. Journal of Adolescence, 37(5), 647-658.
https://doi.org/10.1016/j.adolescence.2013.12.002
Chan, G. C., Kelly, A. B., & Toumbourou, J. W. (2013). Accounting for the association of
family conflict and heavy alcohol use among adolescent girls: The role of depressed
mood. Journal of Studies on Alcohol and Drugs, 74(3), 396-405.
https://doi.org/10.15288/jsad.2013.74.396
Choe, D. E., & Zimmerman, M. A. (2014). Transactional process of African American
adolescents’ family conflict and violent behavior. Journal of Research on
Adolescence, 24(4), 591-597. https://doi.org/10.1111/jora.12056
103
Choe, D. E., Stoddard, S. A., & Zimmerman, M. A. (2014). Developmental trajectories of
African American adolescents’ family conflict: Differences in mental health problems in
young adulthood. Developmental Psychology, 50(4), 1226-1232.
https://doi.org/10.1037/a0035199
Clark, C., Young, M. S., & Dow, M. G. (2013). Can strengthening parenting couples’
relationships reduce at-risk parenting attitudes?. The Family Journal, 21(3), 306-312.
https://doi.org/10.1177/1066480713476841
Clayborne, Z. M., Varin, M., & Colman, I. (2019). Systematic review and meta-analysis:
adolescent depression and long-term psychosocial outcomes. Journal of the American
Academy of Child & Adolescent Psychiatry, 58(1), 72-79.
https://doi.org/10.1016/j.jaac.2018.07.896
Cox, M., & Paley, B. (1997). Families as systems. Annual Review of Psychology, 48(1), 243–
267. https://doi.org/10.1146/annurev.psych.48.1.243
Cui, M., Donnellan, M. B., & Conger, R. D. (2007). Reciprocal influences between parents'
marital problems and adolescent internalizing and externalizing behavior. Developmental
Psychology, 43(6), 1544–1552. https://doi.org/10.1037/0012-1649.43.6.1544
Cummings, E. M., & Schatz, J. N. (2012). Family conflict, emotional security, and child
development: translating research findings into a prevention program for community
families. Clinical Child and Family Psychology Review, 15(1), 14–27.
https://doi.org/10.1007/s10567-012-0112-0
Curran, P. J., & Bauer, D. J. (2011). The disaggregation of within-person and between-person
effects in longitudinal models of change. Annual Review of Psychology, 62, 583-619.
doi: 10.1146/annurev.psych.093008.100356
104
Curran, P. J., Howard, A. L., Bainter, S. A., Lane, S. T., & McGinley, J. S. (2014). The
separation of between-person and within-person components of individual change over
time: A latent curve model with structured residuals. Journal of Consulting and Clinical
Psychology, 82(5), 879. doi: 10.1037/a0035297
Curry, J., Silva, S., Rohde, P., Ginsburg, G., Kratochvil, C., Simons, A., ... & March, J. (2011).
Recovery and recurrence following treatment for adolescent major depression. Archives
of General Psychiatry, 68(3), 263-269. doi:10.1001/archgenpsychiatry.2010.150
Davis, J. P., Pedersen, E. R., Tucker, J. S., Dunbar, M. S., Seelam, R., Shih, R., & D'Amico, E. J.
(2019). Long-term associations between substance use-related media exposure,
descriptive norms, and alcohol use from adolescence to young adulthood. Journal of
Youth and Adolescence, 48(7), 1311–1326. https://doi.org/10.1007/s10964-019-01024-z
De Los Reyes, A., Ohannessian, C.M., & Racz, S.J. (2019). Discrepancies between adolescent
and parent reports about family relationships. Child Development Perspectives, 13(1), 53-
58. doi: 10.1111/cdep.12306
Deane, K., Richards, M., Mozley, M., Scott, D., Rice, C., & Garbarino, J. (2018). Posttraumatic
stress, family functioning, and externalizing in adolescents exposed to violence: A
moderated mediation model. Journal of Clinical Child & Adolescent
Psychology, 47(sup1), S176-S189. DOI: 10.1080/15374416.2016.1197836
Denholm, R., Power, C., Thomas, C., & Li, L. (2013). Child maltreatment and household
dysfunction in a British birth cohort. Child Abuse Review, 22(5), 340-353.
https://doi.org/10.1002/car.2235
105
Elam, K. K., Chassin, L., & Pandika, D. (2018). Polygenic risk, family cohesion, and adolescent
aggression in Mexican American and European American families: Developmental
pathways to alcohol use. Development and Psychopathology, 30(5), 1715–1728.
https://doi.org/10.1017/S0954579418000901
Erel, O., & Burman, B. (1995). Interrelatedness of marital relations and parent-child relations: A
meta-analytic review. Psychological Bulletin, 118(1), 108–132.
https://doi.org/10.1037/0033-2909.118.1.108
Estefan, L. F., Coulter, M. L., VandeWeerd, C. L., Armstrong, M., & Gorski, P. (2013).
Relationships between stressors and parenting attitudes in a child welfare parenting
program. Journal of Child and Family Studies, 22(2), 199-208. doi:10.1007/s10826-012-
9569-1
Finkelhor, D. (1993). Epidemiological factors in the clinical identification of child sexual abuse.
Child Abuse & Neglect, 17(1), 67-70. https://doi.org/10.1016/0145-2134(93)90009-T
Fosco, G. M., & Lydon‐Staley, D. M. (2020). Implications of family cohesion and conflict for
adolescent mood and well‐being: Examining within‐and between‐family processes on a
daily timescale. Family Process, 59(4), 1672-1689. https://doi.org/10.1111/famp.12515
Fosco, G. M., Van Ryzin, M. J., Connell, A. M., & Stormshak, E. A. (2016). Preventing
adolescent depression with the family check-up: Examining family conflict as a
mechanism of change. Journal of Family Psychology, 30(1), 82–92.
https://doi.org/10.1037/fam0000147
106
Fosco, G. M., Van Ryzin, M. J., Xia, M., & Feinberg, M. E. (2016). Trajectories of adolescent
hostile-aggressive behavior and family climate: Longitudinal implications for young
adult romantic relationship competence. Developmental Psychology, 52(7), 1139-1150.
DOI:10.1037/dev0000135
Franck, K. L., & Buehler, C. (2007). A family process model of marital hostility, parental
depressive affect, and early adolescent problem behavior: The roles of triangulation and
parental warmth. Journal of Family Psychology, 21(4), 614. doi: 10.1037/0893-
3200.21.4.614
Hadiwijaya, H., Klimstra, T. A., Vermunt, J. K., Branje, S. J., & Meeus, W. H. (2017). On the
development of harmony, turbulence, and independence in parent–adolescent
relationships: A five-wave longitudinal study. Journal of Youth and Adolescence, 46(8),
1772-1788. https://doi.org/10.1007/s10964-016-0627-7
Hamby, S., Finkelhor, D., Turner, H., & Ormrod, R. (2010). The overlap of witnessing partner
violence with child maltreatment and other victimizations in a nationally representative
survey of youth. Child Abuse & Neglect, 34(10), 734-741.
https://doi.org/10.1016/j.chiabu.2010.03.001
Hanson, J. L., Nacewicz, B. M., Sutterer, M. J., Cayo, A. A., Schaefer, S. M., Rudolph, K. D., ...
& Davidson, R. J. (2015). Behavioral problems after early life stress: contributions of the
hippocampus and amygdala. Biological Psychiatry, 77(4), 314-323.
https://doi.org/10.1016/j.biopsych.2014.04.020
107
Heleniak, C., Jenness, J. L., Vander Stoep, A., McCauley, E., & McLaughlin, K. A. (2016).
Childhood maltreatment exposure and disruptions in emotion regulation: A
transdiagnostic pathway to adolescent internalizing and externalizing
psychopathology. Cognitive Therapy and Research, 40(3), 394-415. DOI
10.1007/s10608-015-9735-z
Herrenkohl, T. I., Lee, J. O., Kosterman, R., & Hawkins, J. D. (2012). Family influences related
to adult substance use and mental health problems: a developmental analysis of child and
adolescent predictors. The Journal of Adolescent Health, 51(2), 129–135.
https://doi.org/10.1016/j.jadohealth.2011.11.003
Herrenkohl, T. I., Sousa, C., Tajima, E. A., Herrenkohl, R. C., & Moylan, C. A. (2008).
Intersection of child abuse and children’s exposure to domestic violence. Trauma,
Violence, & Abuse, 9(2), 84-99. https://doi.org/10.1177/1524838008314797
Hinshaw, S. P. (1992). Externalizing behavior problems and academic underachievement in
childhood and adolescence: causal relationships and underlying
mechanisms. Psychological Bulletin, 111(1), 127. https://doi.org/10.1037/0033-
2909.111.1.127
Hirschi, T. (1969). A control theory of delinquency. Criminology theory: Selected classic
readings, 1969, 289-305.
Horwitz, B. N., Ganiban, J. M., Spotts, E. L., Lichtenstein, P., Reiss, D., & Neiderhiser, J. M.
(2011). The role of aggressive personality and family relationships in explaining family
conflict. Journal of Family Psychology, 25(2), 174–183.
https://doi.org/10.1037/a0023049
108
Horwitz, S. M., Hurlburt, M. S., Cohen, S. D., Zhang, J., & Landsverk, J. (2011). Predictors of
placement for children who initially remained in their homes after an investigation for
abuse or neglect. Child Abuse & Neglect, 35(3), 188-198.
https://doi.org/10.1016/j.chiabu.2010.12.002
Hou, Y., Benner, A. D., Kim, S. Y., Chen, S., Spitz, S., Shi, Y., & Beretvas, T. (2020).
Discordance in parents’ and adolescents’ reports of parenting: A meta-analysis and
qualitative review. American Psychologist, 75(3), 329–348.
https://doi.org/10.1037/amp0000463
Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis:
Conventional criteria versus new alternatives. Structural Equation Modeling: A
Multidisciplinary Journal, 6(1), 1-55. doi.org/10.1080/10705519909540118
Jambunathan, S., Burts, D. C., & Pierce, S. (2000). Comparisons of parenting attitudes among
five ethnic groups in the United States. Journal of Comparative Family Studies, 31(4),
395-406. https://doi.org/10.3138/jcfs.31.4.395
Jennings, W. G., & Reingle, J. M. (2012). On the number and shape of developmental/life-course
violence, aggression, and delinquency trajectories: A state-of-the-art review. Journal of
Criminal Justice, 40(6), 472-489. https://doi.org/10.1016/j.jcrimjus.2012.07.001
Joiner, T., Coyne, J. C., & Blalock, J. (1999). On the interpersonal nature of depression:
Overview and synthesis. In T. Joiner & J. C. Coyne (Eds.), The interactional nature of
depression: Advances in interpersonal approaches (pp. 3-19). Washington, DC, US:
American Psychological Association. https://doi.org/10.1037/10311-013
Jonson-Reid, M., Kohl, P. L., & Drake, B. (2012). Child and adult outcomes of chronic child
maltreatment. Pediatrics, 129(5), 839-845. DOI: https://doi.org/10.1542/peds.2011-2529
109
Juang, L. P., & Alvarez, A. A. (2010). Discrimination and adjustment among Chinese American
adolescents: Family conflict and family cohesion as vulnerability and protective factors.
American Journal of Public Health, 100(12), 2403-2409.
doi: 10.2105/AJPH.2009.185959
Kaferly, J., Furniss, A., & Allison, M. A. (2020). Transmission of intergenerational parenting
attitudes and natural mentorship: associations within the LONGSCAN population. Child
Abuse & Neglect, 108, 104662. https://doi.org/10.1016/j.chiabu.2020.104662
Karriker-Jaffe, K. J., Foshee, V. A., Ennett, S. T., & Suchindran, C. (2013). Associations of
neighborhood and family factors with trajectories of physical and social aggression
during adolescence. Journal of Youth and Adolescence, 42(6), 861–877.
https://doi.org/10.1007/s10964-012-9832-1
Kelly, A. B., Mason, W. A., Chmelka, M. B., Herrenkohl, T. I., Kim, M. J., Patton, G. C.,
Hemphill, S. A., Toumbourou, J. W., & Catalano, R. F. (2016). Depressed mood during
early to middle adolescence: A bi-national longitudinal study of the unique impact of
family conflict. Journal of Youth and Adolescence, 45(8), 1604-1613.
https://doi.org/10.1007/s10964-016-0433-2
Keyes, K. M., Gary, D., O’Malley, P. M., Hamilton, A., & Schulenberg, J. (2019). Recent
increases in depressive symptoms among US adolescents: trends from 1991 to 2018.
Social Psychiatry and Psychiatric Epidemiology, 54(8), 987-996.
https://doi.org/10.1007/s00127-019-01697-8
110
Knopp, K., Rhoades, G. K., Allen, E. S., Parsons, A., Ritchie, L. L., Markman, H. J., & Stanley,
S. M. (2017). Within- and Between-Family Associations of Marital Functioning and
Child Well-being. Journal of marriage and the family, 79(2), 451–461.
https://doi.org/10.1111/jomf.12373
Koepke, S., & Denissen, J. J. (2012). Dynamics of identity development and separation–
individuation in parent–child relationships during adolescence and emerging adulthood–
A conceptual integration. Developmental Review, 32(1), 67-88.
https://doi.org/10.1016/j.dr.2012.01.001
Lau, A. S., Valeri, S. M., McCarty, C. A., & Weisz, J. R. (2006). Abusive parents’ reports of
child behavior problems: Relationship to observed parent-child interactions. Child Abuse
& Neglect, 30(6), 639-655. https://doi.org/10.1016/j.chiabu.2005.11.009
Lee, F. S., Heimer, H., Giedd, J. N., Lein, E. S., Šestan, N., Weinberger, D. R., & Casey, B. J.
(2014). Adolescent mental health—opportunity and obligation. Science, 346(6209), 547-
549. DOI: 10.1126/science.1260497
Lereya, S. T., Samara, M., & Wolke, D. (2013). Parenting behavior and the risk of becoming a
victim and a bully/victim: A meta-analysis study. Child Abuse & Neglect, 37(12), 1091-
1108. https://doi.org/10.1016/j.chiabu.2013.03.001
Lewis, A. J., Kremer, P., Douglas, K., Toumborou, J. W., Hameed, M. A., Patton, G. C., &
Williams, J. (2015). Gender differences in adolescent depression: Differential female
susceptibility to stressors affecting family functioning. Australian Journal of Psychology,
67(3), 131–139. https://doi.org/10.1111/ajpy.12086
111
Li, M., Johnson, S. B., Musci, R. J., & Riley, A. W. (2017). Perceived neighborhood quality,
family processes, and trajectories of child and adolescent externalizing behaviors in the
United States. Social Science & Medicine, 192, 152-161.
https://doi.org/10.1016/j.socscimed.2017.07.027
Lindell, K. U., Sorenson, C. K., & Mangold, S. V. (2020). The Family First Prevention Services
Act: A new era of child welfare reform. Public Health Reports, 135(2), 282-286.
https://doi.org/10.1177/0033354919900892
Liu, M., & Hancock, G. R. (2014). Unrestricted mixture models for class identification in growth
mixture modeling. Educational and Psychological Measurement, 74(4), 557-584.
https://doi.org/10.1177/0013164413519798
Lovejoy, M. C., Graczyk, P. A., O'Hare, E., & Neuman, G. (2000). Maternal depression and
parenting behavior: A meta-analytic review. Clinical Psychology Review, 20(5), 561-592.
https://doi.org/10.1016/S0272-7358(98)00100-7
Marsiglia, F. F., Parsai, M., & Kulis, S. (2009). Effects of familism and family cohesion on
problem behaviors among adolescents in Mexican immigrant families in the southwest
United States. Journal of Ethnic & Cultural Diversity in Social Work, 18(3), 203-220.
https://doi.org/10.1080/15313200903070965
Mastrotheodoros, S., Van der Graaff, J., Deković, M., Meeus, W. H., & Branje, S. (2020).
Parent–adolescent conflict across adolescence: Trajectories of informant discrepancies
and associations with personality types. Journal of Youth and Adolescence, 49(1), 119-
135. https://doi.org/10.1007/s10964-019-01054-7
112
McLaughlin, K. A., Green, J. G., Gruber, M. J., Sampson, N. A., Zaslavsky, A. M., & Kessler,
R. C. (2010). Childhood adversities and adult psychiatric disorders in the national
comorbidity survey replication II: associations with persistence of DSM-IV
disorders. Archives of General Psychiatry, 67(2), 124–132.
https://doi.org/10.1001/archgenpsychiatry.2009.187
Mennen, F. E., & Trickett, P. K. (2011). Parenting attitudes, family environments, depression,
and anxiety in caregivers of maltreated children. Family Relations. 60(3), 259-271.
https://doi.org/10.1111/j.1741-3729.2011.00646.x
Mennen, F. E., Negriff, S., Schneiderman, J. U., & Trickett, P. K. (2018). Longitudinal
associations of maternal depression and adolescents’ depression and behaviors:
Moderation by maltreatment and sex. Journal of Family Psychology, 32(2), 240–250.
https://doi.org/10.1037/fam0000394
Miller, S., Boxmeyer, C., Romero, D., Powell, N., Jones, S., & Lochman, J. (2020). Theoretical
model of Mindful Coping Power: Optimizing a cognitive behavioral program for high-
risk children and their parents by integrating mindfulness. Clinical Child and Family
Psychology Review, 1-14. https://doi.org/10.1007/s10567-020-00312-6
Mills, R., Scott, J., Alati, R., O'Callaghan, M., Najman, J. M., & Strathearn, L. (2013). Child
maltreatment and adolescent mental health problems in a large birth cohort. Child Abuse
& Neglect, 37(5), 292–302. https://doi.org/10.1016/j.chiabu.2012.11.008
Milner, J. S. (1993). Social information processing and physical child abuse. Clinical Psychology
Review, 13(3), 275-294. https://doi.org/10.1016/0272-7358(93)90024-G
Milner, J. S. (2003). Social information processing in high-risk and physically abusive parents.
Child Abuse & Neglect, 27(1), 7–20. https://doi.org/10.1016/S0145-2134(02)00506-9
113
Molleda, L., Estrada, Y., Lee, T. K., Poma, S., Terán, A. M. Q., Tamayo, C. C., & Prado, G.
(2017). Short-term effects on family communication and adolescent conduct problems:
familias Unidas in Ecuador. Prevention Science, 18(7), 783–792. https://doi.
org/10.1007/s11121-016-0744-2.
Mollerstrom, W. W., Patchner, M. A., & Milner, J. S. (1992). Family functioning and child
abuse potential. Journal of Clinical Psychology, 48(4), 445-454. doi: 10.1002/1097-
4679(199207)48:4<445::AID-JCLP2270480404>3.0.CO;2-2
Moos, R. H., & Moos, B. S. (1994). Family Environment Scale manual. Consulting
Psychologists Press.
Moreira, J. F. G., & Telzer, E. H. (2015). Changes in family cohesion and links to depression
during the college transition. Journal of Adolescence, 43, 72-82.
https://doi.org/10.1016/j.adolescence.2015.05.012
Muthén, B. (2011). Applications of causally defined direct and indirect effects in mediation
analysis using SEM in Mplus.
Muthé n, L. K., & Muthé n, B. O. (1998-2017). Mplus user’s guide (Version 8.0). Los Angeles,
CA: Muthé n & Muthé n.
Negriff, S., Cederbaum, J. A., & Lee, D. S. (2019). Does Social Support Mediate the Association
Between Maltreatment Experiences and Depressive Symptoms in Adolescence. Child
Maltreatment, 24(2), 203–212. https://doi.org/10.1177/1077559518814680
Negriff, S., Gordis, E., Susman, E., Kim, K., Peckins, M., Schneiderman, J., & Mennen, F.
(2020). The Young Adolescent Project: A longitudinal study of the effects of
maltreatment on adolescent development. Development and Psychopathology, 32(4),
1440-1459. Doi:10.1017/S0954579419001391
114
Norman, R. E., Byambaa, M., De, R., Butchart, A., Scott, J., & Vos, T. (2012). The long-term
health consequences of child physical abuse, emotional abuse, and neglect: A systematic
review and meta-analysis. PLoS Medicine, 9(11), e1001349.
https://doi.org/10.1371/journal.pmed.1001349
O’Keefe, M., Mennen, F., & Lane, C. J. (2006). An examination of the factor structure for the
Youth Self Report on a multiethnic population. Research on Social Work Practice, 16(3),
315-325. https://doi.org/10.1177/1049731505285340
Olson, A. E., Shenk, C. E., Noll, J. G., & Allen, B. (2021). Child maltreatment and substance use
in emerging adulthood: Internalizing and externalizing behaviors at the transition to
adolescence as indirect pathways. Child Maltreatment, 10775595211010965.
https://doi.org/10.1177/10775595211010965
Olson, D. H., Waldvogel, L., & Schlieff, M. (2019). Circumplex model of marital and family
systems: An update. Journal of Family Theory & Review, 11(2), 199-211.
https://doi.org/10.1111/jftr.12331
Park, S., Kim, B. N., & Park, M. H. (2016). The relationship between parenting attitudes,
negative cognition, and the depressive symptoms according to gender in Korean
adolescents. International Journal of Mental Health Systems, 10(1), 1-8.
https://doi.org/10.1186/s13033-016-0069-3
Patterson, G. R., DeBaryshe, B. D., & Ramsey, E. (1989). A developmental perspective on
antisocial behavior. American Psychologist, 44(2), 329–
335. https://doi.org/10.1037/0003-066X.44.2.329
115
Peckins, M. K., Negriff, S., Reader, J. M., & Susman, E. J. (2018). Perception of Maltreatment:
Gender-Specific Mental Health Outcomes Associated with Maltreatment as Most
Upsetting Experience. In Child Maltreatment Research, Policy, and Practice (pp. 27-47).
Springer, Cham.
Peterson, C., Florence, C., & Klevens, J. (2018). The economic burden of child maltreatment in
the United States, 2015. Child Abuse & Neglect, 86, 178-183.
https://doi.org/10.1016/j.chiabu.2018.09.018
Pinquart, M. (2017). Associations of Parenting Dimensions and Styles with Internalizing
Symptoms in Children and Adolescents: A Meta-Analysis. Marriage and Family Review.
53(7), 613-640. https://doi.org/10.1080/01494929.2016.1247761
Ram, N., & Grimm, K. J. (2009). Methods and measures: Growth mixture modeling: A method
for identifying differences in longitudinal change among unobserved groups.
International Journal of Behavioral Development, 33(6), 565-576.
https://doi.org/10.1177/0165025409343765
Rodriguez, C. M. (2010). Parent–child aggression: Association with child abuse potential and
parenting styles. Violence and Victims, 25(6), 728-741. doi: 10.1891/0886-6708.25.6.728
Rodriguez, C. M. (2013). Analog of parental empathy: Association with physical child abuse risk
and punishment intentions. Child Abuse & Neglect, 37(8), 493-499.
https://doi.org/10.1016/j.chiabu.2012.10.004
Rodriguez, C. M., Silvia, P. J., & Gaskin, R. E. (2019). Predicting Maternal and Paternal Parent-
Child Aggression Risk: Longitudinal Multimethod Investigation using Social Information
Processing Theory. Psychology of Violence, 9(3), 370–382.
https://doi.org/10.1037/vio0000115
116
Rothenberg, W. A., Hussong, A. M., & Chassin, L. (2016). Intergenerational continuity in high-
conflict family environments. Development and Psychopathology, 28(1), 293–308.
https://doi.org/10.1017/S0954579415000450
Rothenberg, W. A., Hussong, A. M., & Chassin, L. (2017). Modeling trajectories of adolescent‐
perceived family conflict: Effects of marital dissatisfaction and parental alcoholism.
Journal of Research on Adolescence, 27(1), 105-121. https://doi.org/10.1111/jora.12259
Rothenberg, W. A., Solis, J. M., Hussong, A. M., & Chassin, L. (2017). Profiling families in
conflict: Multigenerational continuity in conflict predicts deleterious adolescent and adult
outcomes. Journal of Family Psychology, 31(5), 616–628.
https://doi.org/10.1037/fam0000319
Roubinov, D. S., & Luecken, L. J. (2013). Family conflict in childhood and adolescence and
depressive symptoms in emerging adulthood: Mediation by disengagement coping.
Journal of Divorce & Remarriage, 54(7), 576-595.
https://doi.org/10.1080/10502556.2013.828988
Rudolph, K. D., Flynn, M., & Abaied, J. L. (2008). A developmental perspective on
interpersonal theories of youth depression. In J. R. Z. Abela & B. L. Hankin (Eds.),
Handbook of Depression in Children and Adolescents (pp. 79–102). The Guilford Press.
Salk, R. H., Hyde, J. S., & Abramson, L. Y. (2017). Gender differences in depression in
representative national samples: Meta-analyses of diagnoses and symptoms.
Psychological Bulletin, 143(8), 783–822. https://doi.org/10.1037/bul0000102
Sameroff, A. J. (1983). Developmental systems: Contexts and evolution. In W. Kessen (Ed.),
Handbook of Child Psychology: Vol. 1. History, theory, and methods (pp. 237–294). New
York: Wiley.
117
Samuels, G. M., & Pryce, J. M. (2008). “What doesn't kill you makes you stronger”: Survivalist
self-reliance as resilience and risk among young adults aging out of foster care. Children
and Youth Services Review, 30(10), 1198-1210.
https://doi.org/10.1016/j.childyouth.2008.03.005
Satorra, A., & Bentler, P. M. (2001). A scaled difference chi-square test statistic for moment
structure analysis. Psychometrika, 66(4), 507–514. https://doi.org/10.1007/BF02296192
Schlomer, G. L., Bauman, S., & Card, N. A. (2010). Best practices for missing data management
in counseling psychology. Journal of Counseling Psychology, 57(1), 1–10.
https://doi.org/10.1037/a0018082
Sears, M. S., Repetti, R. L., Reynolds, B. M., Robles, T. F., & Krull, J. L. (2016). Spillover in
the home: The effects of family conflict on parents' behavior. Journal of Marriage and
Family, 78(1), 127-141. doi:10.1111/jomf.12265
Selig, J. P., & Little, T. D. (2012). Autoregressive and cross-lagged panel analysis for
longitudinal data. In B. Laursen, T. D. Little, & N. A. Card (Eds.), Handbook of
Developmental Research Methods (pp. 265-278). New York, NY, US: The Guilford
Press.
Sherrill, R. B., Lochman, J. E., DeCoster, J., & Stromeyer, S. L. (2017). Spillover between
interparental conflict and parent–child conflict within and across days. Journal of Family
Psychology, 31(7), 900–909. https://doi.org/10.1037/fam0000332
Sousa, C., Mason, W. A., Herrenkohl, T. I., Prince, D., Herrenkohl, R. C., & Russo, M. J.
(2018). Direct and indirect effects of child abuse and environmental stress: A lifecourse
perspective on adversity and depressive symptoms. The American Journal of
Orthopsychiatry, 88(2), 180–188. https://doi.org/10.1037/ort0000283
118
Stark, K. D., Banneyer, K. N., Wang, L. A., & Arora, P. (2012). Child and adolescent depression
in the family. Couple and Family Psychology: Research and Practice, 1(3), 161–184.
https://doi.org/10.1037/a0029916
Steeger, C. M., & Gondoli, D. M. (2013). Mother-adolescent conflict as a mediator between
adolescent problem behaviors and maternal psychological control. Developmental
Psychology, 49(4), 804–814. https://doi.org/10.1037/a0028599
Stevens, K. I., Schneiderman, J. U., Negriff, S., Brinkmann, A., & Trickett, P. K. (2015). The
whole picture: Child maltreatment experiences of youths who were physically abused.
Child Abuse & Neglect, 43, 30–41. https://doi.org/10.1016/j.chiabu.2014.12.006
Stith, S. M., Liu, T., Davies, L. C., Boykin, E. L., Alder, M. C., Harris, J. M., Som, A.,
McPherson, M. & Dees, J. E. M. E. G. (2009). Risk factors in child maltreatment: A
meta-analytic review of the literature. Aggression and Violent Behavior, 14(1), 13-29.
https://doi.org/10.1016/j.avb.2006.03.006
Taylor, L. K., Merrilees, C. E., Goeke-Morey, M. C., Shirlow, P., & Cummings, E. M. (2016).
Trajectories of adolescent aggression and family cohesion: The potential to perpetuate or
ameliorate political conflict. Journal of Clinical Child & Adolescent Psychology, 45(2),
114-128. https://doi.org/10.1080/15374416.2014.945213
Thornberry, T. P., Matsuda, M., Greenman, S. J., Augustyn, M. B., Henry, K. L., Smith, C. A., &
Ireland, T. O. (2014). Adolescent risk factors for child maltreatment. Child Abuse &
Neglect, 38(4), 706-722. https://doi.org/10.1016/j.chiabu.2013.08.009
Timmons, A. C., & Margolin, G. (2015). Family conflict, mood, and adolescents’ daily school
problems: Moderating roles of internalizing and externalizing symptoms. Child
Development, 86(1), 241-258. https://doi.org/10.1111/cdev.12300
119
Toth, S. L., Handley, E. D., Manly, J. T., Sturm, R., Adams, T. R., Demeusy, E. M., & Cicchetti,
D. (2020). The moderating role of child maltreatment in treatment efficacy for adolescent
depression. Journal of Abnormal Child Psychology, 48(10), 1351-1365.
https://doi.org/10.1007/s10802-020-00682-z
Trentacosta, C. J., Criss, M. M., Shaw, D. S., Lacourse, E., Hyde, L. W., & Dishion, T. J. (2011).
Antecedents and outcomes of joint trajectories of mother–son conflict and warmth during
middle childhood and adolescence. Child Development, 82(5), 1676-1690.
https://doi.org/10.1111/j.1467-8624.2011.01626.x
Tsai, K. M., Telzer, E. H., & Fuligni, A. J. (2013). Continuity and discontinuity in perceptions of
family relationships from adolescence to young adulthood. Child Development, 84(2),
471-484. DOI: 10.1111/j.1467-8624.2012.01858.x
U.S. Department of Health & Human Services, Administration for Children and Families,
Administration on Children, Youth and Families, Children’s Bureau. (2021). Child
Maltreatment 2019. https://www.acf.hhs.gov/cb/research-data-technology/statistics-
research/child-maltreatment.
Vial, A., van der Put, C., Stams, G. J. J., Kossakowski, J., & Assink, M. (2020). Exploring the
interrelatedness of risk factors for child maltreatment: a network approach. Child Abuse
& Neglect, 107, 104622. https://doi.org/10.1016/j.chiabu.2020.104622
Villodas, M. T., Litrownik, A. J., Thompson, R., Jones, D., Roesch, S. C., Hussey, J. M., Block,
S., English, D. J., & Dubowitz, H. (2015). Developmental transitions in presentations of
externalizing problems among boys and girls at risk for child maltreatment. Development
and Psychopathology, 27(1), 205–219. https://doi.org/10.1017/S0954579414000728
120
Wagner, K. D., Ritt-Olson, A., Chou, C. P., Pokhrel, P., Duan, L., Baezconde-Garbanati, L.,
Soto, D. W., & Unger, J. B. (2010). Associations between family structure, family
functioning, and substance use among Hispanic/Latino adolescents. Psychology of
Addictive Behaviors, 24(1), 98–108. https://doi.org/10.1037/a0018497
Wamser-Nanney, R., & Campbell, C. L. (2020). Predictors of parenting attitudes in an at-risk
sample: results from the LONGSCAN study. Child Abuse & Neglect, 99, 104282.
https://doi.org/10.1016/j.chiabu.2019.104282
Weed, K., Morales, D. A., & Harjes, R. (2013). Symptoms of depression depend on rigid
parenting attitudes, gender, and race in an at-risk sample of early adolescents. Journal of
Emotional and Behavioral Disorders, 21(4), 250-263.
https://doi.org/10.1177/1063426611424748
Weymouth, B. B., Buehler, C., Zhou, N., & Henson, R. A. (2016). A meta‐analysis of parent–
adolescent conflict: Disagreement, hostility, and youth maladjustment. Journal of Family
Theory & Review, 8(1), 95-112. https://doi.org/10.1111/jftr.12126
White, J., Shelton, K. H., & Elgar, F. J. (2014). Prospective associations between the family
environment, family cohesion, and psychiatric symptoms among adolescent girls. Child
Psychiatry & Human Development, 45(5), 544-554. doi: 10.1007/s10578-013-0423-5
Xu, Y., Boyd, R. C., Butler, L., Moore, T. M., & Benton, T. D. (2017). Associations of parent-
adolescent discrepancies in family cohesion and conflict with adolescent impairment.
Journal of Child and Family Studies, 26(12), 3360-3369. https://doi.org/10.1007/s10826-
017-0825-2
121
Yap, M. B. H., Pilkington, P. D., Ryan, S. M., & Jorm, A. F. (2014). Parental factors associated
with depression and anxiety in young people: A systematic review and meta‐analysis.
Journal of Affective Disorders, 156, 8–23. doi:10.1016/j.jad.2013.11.007
Yu, J. J. (2019). Longitudinal typologies of perceived parent-child conflict and their correlates in
adolescence. Children and Youth Services Review, 98, 132-142.
https://doi.org/10.1016/j.childyouth.2018.12.026
Zemp, M., Johnson, M. D., & Bodenmann, G. (2018). Within-family processes: Interparental and
coparenting conflict and child adjustment. Journal of Family Psychology, 32(3), 299–
309. https://doi.org/10.1037/fam0000368
Abstract (if available)
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Examining the longitudinal relationships between community violence exposure and aggressive behavior among a sample of maltreated and non-maltreated adolescents
PDF
Untangling the developmental relations between depression and externalizing behavior among maltreated adolescents
PDF
Childhood adversity across generations and its impact on externalizing behavior
PDF
Neighborhood context and adolescent mental health: development and mechanisms
PDF
The impact of child maltreatment on the mental health of adolescents: a longitudinal study of social anxiety and self-perception
PDF
Understanding allegations of childhood neglect using structured and unstructured administrative data
PDF
The link between maternal depression and adolescent daughters' risk behavior: the mediating and moderating role of family
PDF
Trajectories of internalizing problems among maltreated girls and boys: differences by maltreatment type and developmental timing
PDF
The relationship between pubertal timing and delinquent behavior in maltreated male and female adolescents
PDF
Examining the longitudinal influence of the physical and social environments on social isolation and cognitive health: contextualizing the role of technology
PDF
Childhood cancer survivorship: parental factors associated with survivor's follow-up care behavior and mental health
PDF
Strength-Based Reporting: a trauma-informed practice for mandated reporters, to address behavioral health concerns in children at risk of child welfare involvement
PDF
Low-socioeconomic status families: the role of parental involvement and its association with early childhood academic achievement trajectories
PDF
U.S. Latinx youth development and substance use risk: adversity and strengths
PDF
Post-investigation experiences of families with complex needs at risk of maltreatment: an examination of need, matched services, and changes in need
PDF
Complicated relationships: the prevalence and correlates of sexual assault, dating violence victimization, and minority stress among sexual minority adolescents throughout the United States
PDF
Impact of change in sexual identity on mental health risks among sexual minority adolescents
PDF
County of San Diego Child Welfare Services Hotline Redesign
PDF
Effectiveness and implementation of a maternal depression treatment in a Head Start setting: a mixed method study
PDF
Emotion regulation as a mechanism linking parents’ marital aggression to adolescent behavioral problems: a longitudinal analysis
Asset Metadata
Creator
Lee, Daniel SeungChul
(author)
Core Title
Maltreated adolescents and their families: a longitudinal examination of family functioning, parenting attitudes, & youth mental health
School
Suzanne Dworak-Peck School of Social Work
Degree
Doctor of Philosophy
Degree Program
Social Work
Degree Conferral Date
2021-08
Publication Date
07/22/2021
Defense Date
05/26/2021
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
adolescence,child maltreatment,Depression,externalizing,family cohesion,family conflict,longitudinal analysis,Mental Health,OAI-PMH Harvest
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Mennen, Ferol E. (
committee chair
), Cederbaum, Julie A. (
committee member
), Davis, Jordan P. (
committee member
), Immordino-Yang, Mary Helen (
committee member
)
Creator Email
dslee@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC15616317
Unique identifier
UC15616317
Legacy Identifier
etd-LeeDanielS-9822
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Lee, Daniel SeungChul
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Repository Email
cisadmin@lib.usc.edu
Tags
child maltreatment
externalizing
family cohesion
family conflict
longitudinal analysis