Professional Documents
Culture Documents
Four distinct patterns of adolescents’ behavioral, emotional, and physiological responses to family conflict were identi-
fied during mother–father–adolescent (M = 13.08 years) interactions. Most youth displayed adaptively regulated pat-
terns comprised of low overt and subjective distress. Under-controlled adolescents exhibited elevated observable and
subjective anger. Over-controlled adolescents were withdrawn and reported heightened subjective distress. Physiologi-
cally reactive adolescents had elevated cortisol coupled with low overt and subjective distress. Regulation patterns
were associated with unique mental health trajectories. Under-controlled adolescents had elevated conduct and peer
problems whereas over-controlled adolescents had higher anxiety and depressive symptoms. Physiologically reactive
adolescents had low concurrent, but increasing levels of depressive, anxiety, and peer problem symptoms. Findings
underscore the importance of examining organizations of regulatory strategies in contributing to adolescent mental
health.
Children’s strategies for managing stress are critical responses in service of one’s goals (Thompson,
to their mental health. Building self- and emotion- 1994) and has been conceptualized as a multi-
regulation skills is a key developmental task faceted system comprised of several component
(Sameroff, 2009; Thompson, 1994); the ability to processes including emotional, behavioral, and bio-
successfully regulate stressful and emotionally logical components (Cicchetti, Ganiban, & Barnett,
arousing events has positive implications for child 1991; Thompson, Lewis, & Calkins, 2008). This
and adolescent development (e.g., Zeman, Cassano, study sought to explore how unique patterns of
Perry-Parrish, & Stegall, 2006). By comparison, the adolescent regulation, manifested across emotional,
inability to effectively regulate emotion and behav- behavioral, and biological responses to a family
ior during stressful events is linked to later social, disagreement task, contribute to the emerging risk
emotional, and behavioral difficulties (e.g., Eisen- for psychopathology.
berg et al., 2001; Shields, Cicchetti, & Ryan, 1994). Regulation strategies may be best understood in
Emotion regulation has been defined as the ability the context in which they occur. An especially sali-
to monitor, evaluate, and modify emotional ent context in which youth are faced with stressors
is within the family. The family provides everyday
opportunities for youth to encounter stressful situa-
This research was supported by grant R01 MH57318 from the
National Institute of Mental Health awarded to Patrick T. Davies tions and develop strategies for handling and cop-
and E. Mark Cummings. Support was provided to Kalsea J. Koss ing with stress (Zeman et al., 2006). In risky
by a dissertation fellowship from the American Psychological families, youth’s responses develop as an adaptive
Foundation and was supported by National Institute of Mental means to cope with multiple threats, including the
Health training grants (T32 MH015755 and T32 MH018921) dur-
immediate threat of family discord (Repetti, Taylor,
ing the preparation of this article. The authors are grateful to the
families and teachers who participated in this project. Their grat- & Seeman, 2002); over time, these short-term
itude is also expressed to the staff and students who assisted on responses may serve as precursor outcomes to later
various stages of the project at the University of Notre Dame health and well-being deficits (Repetti, Robles, &
and the University of Rochester and in particular to Jana Lam Reynolds, 2011). Emotional security theory (Davies
for her assistance with the observational coding.
Requests for reprints should be sent to Kalsea Koss, Depart-
ment of Psychology and Human Development, Vanderbilt © 2016 The Authors
University, Nashville, TN 37203. E-mail: kalsea.j.koss@ Journal of Research on Adolescence © 2016 Society for Research on Adolescence
vanderbilt.edu DOI: 10.1111/jora.12269
230 KOSS, CUMMINGS, DAVIES, AND CICCHETTI
& Cummings, 1994) posits that youth’s regulatory problems whereas under-controlled regulation is
processes together reflect goal-directed strategies related to higher rates of externalizing problems
for restoring and maintaining felt security. As a (e.g., Cole, Zahn-Waxler, Fox, Usher, & Welsh,
threat to security, family discord activates multiple 1996; Eisenberg et al., 2001).
emotional, behavioral, and cognitive regulatory Dysregulated patterns of responses to family
responses that serve to regain a sense of security in conflict may reflect goal-directed strategies to alle-
the family. For example, youth may exhibit viate the threat and occurrence of discord, helping
increased emotional distress, including more youth to preserve a sense of safety despite the
intense and prolonged instances of anger, fear, and threats posed by their family environment. In
sadness in response to family stressors. Youth also response to family adversity, both under- and
utilize a variety of behavioral responses including over-controlled regulation may reflect hyper-
avoidance, vigilance, involvement to mediate con- responsivity to threats; however, these may mani-
flict, and dysregulated anger as strategies to regu- fest in different strategies with under-controlled
late their exposure to discord (Cummings & youth exhibiting overtly aggressive strategies
Davies, 2010). While such response strategies may whereas over-controlled youth appear inhibited
be adaptive in the short term, serving to alleviate and vigilant. Histories of family functioning influ-
the immediate threat of conflict and help restore a ence the meaning discord may have for youth’s
sense of security, over time these same regulatory security and, as such, discord may not pose a
processes (e.g., withdrawal, aggression) may be threat to the safety and stability of the family unit
maladaptive in broader social contexts and have for all youth. Adaptive regulation strategies for
cumulative developmental consequences that lead youth who preserve a sense of security during fam-
to the emergence of psychopathology. ily discord may be characterized by low to moder-
ate levels of arousal. Several studies have explored
patterns of emotion regulation in response to inter-
Emotional and Behavioral Regulation Patterns
adult anger in samples of both maltreated and non-
The stress and coping literature has long highlighted maltreated youth (e.g., Cummings, 1987; Davies &
the importance of understanding the higher order Forman, 2002; El-Sheikh, Cummings, & Goetsch,
organization of regulatory processes which reflect 1989; Maughan & Cicchetti, 2002). For example,
an organized system of responses (Fleming, Baum, Maughan and Cicchetti (2002) classified children
& Singer, 1984). Emotional security theory adopts an into three patterns reflecting adaptive regulation,
organizational perspective on regulation (Cum- over-controlled regulation, and under-controlled
mings & Davies, 1996); inherent in this perspective regulation. Adaptively concerned children dis-
is the notion that across multiple levels of responses, played moderate levels of negative affect arousal
strategies together reflect goal-directed behavior to during a background anger interaction that sub-
preserve or restore felt security. However, many sided during a post-task resolution period. Under-
investigations of youth’s insecurity have not focused controlled/ambivalent children exhibited under-
on the ways in which these different responses inter- regulated, disorganized, and ambivalent behaviors
act and collectively represent organized systems of and higher levels of both positive and negative
responses. Presently, little is known about how these affect. These children also lacked goal-directed
regulatory processes relate to one another and how behavior and exhibited higher levels of dysregu-
differences in the organization of responses relate to lated arousal. Over-controlled/unresponsive chil-
adjustment across adolescence. dren exhibited a desire to withdraw or avoid the
In the emotion regulation literature, dysregula- situation and evidenced an absence or low levels of
tion commonly takes two forms: patterns of over- observable affective responses in combination with
control and under-control (e.g., Smith & Eisenberg, heightened subjective experiences of negative emo-
2005). Youth exhibiting over-controlled regulation tion and distress. Maltreated youth were more
may seem well-regulated to an observer but these likely to display both over- and under-controlled
youth are characterized by heightened levels of patterns compared with nonmaltreated youth.
inhibition, avoidance, and emotional distress. Alter- Additionally, under-controlled youth had higher
natively, youth exhibiting under-controlled regula- rates of social and internalizing problems, support-
tion lack the ability to constrain their emotions and ing the notion that for a subset of children under-
behaviors. Both forms of dysregulation have been controlled regulation serves as an explanatory
linked to adjustment problems; over-controlled reg- mechanism between family risk and maladjust-
ulation is associated with risk for internalizing ment.
PATTERNS OF ADOLESCENT REGULATION 231
Little research has explored youth’s regulation acute stressor consists of elevated levels as well as
patterns in the context of conflict beyond child- an intensified reactivity peak and a prolonged
hood and less is known whether these identifiable return to prestressor levels; over time, this height-
and distinct patterns continue into adolescence. ened HPA activity may cause wear and tear on the
Family conflict may be particularly salient during body. On the other hand, hyporeactivity of the
adolescence; conflicts within families peak in late HPA axis is characterized as decreased sensitivity
childhood and early adolescence (Shanahan, of the HPA response to acute stressors resulting in
McHale, Osgood, & Crouter, 2007) as children lower levels and blunted, flattened reactivity.
transition to adolescence and seek a balance Downregulation of the HPA axis in response to
between autonomy and relatedness within family chronic stress is thought to serve to protect against
relationships. Furthermore, adolescence is charac- repeated elevations in cortisol production (e.g.,
terized by vast changes in cognition, emotion, Fries, Hesse, Hellhammer, & Hellhammer, 2005)
social relationships, and physical and brain devel- and may represent an adaptive mechanism in
opment, which may contribute to the developmen- response to heightened family adversity (e.g.,
tal stage as a sensitive period for reorganization in Saxbe, Margolin, Shapiro, & Baucom, 2012). In sup-
how responses relate to one another. Examining port of examining the ways in which the HPA axis
regulation patterns during early adolescence prior contributes to organizations of insecurity, cortisol
to the development of mental health problems response patterns have been associated with emo-
may allow for understanding the etiology of psy- tional and behavioral reactivity and regulation
chopathology. (e.g., Davies, Sturge-Apple, Cicchetti, & Cummings,
2008; Koss et al., 2013; Lisonbee, Pendry, Mize, &
Gwynn, 2010) and both forms of altered responses
Physiological Reactivity and Regulation
are associated with maladjustment (e.g., Davies
Thompson et al. (2008) called attention to the need et al., 2011; Gunnar & Vazquez, 2006).
to incorporate the interactive nature of multiple
regulatory processes involved in emotion regula-
Adopting a Person-Oriented Approach
tion, including the integration of physiological pro-
cesses. For example, identifying the psychological Studies of pattern-based or person-centered
and biological processes emerging during child- approaches have typically focused on identifying a
hood and adolescence in risky families may be predetermined set of patterns (e.g., Cummings,
important for understanding how family adversity 1987; Maughan & Cicchetti, 2002). To date, data-
leads to cascading mental health problems (Repetti driven approaches are an emerging direction for
et al., 2011). At a biological level, the hypothala- uncovering profiles of responses in the broader
mic–pituitary–adrenal axis (HPA) serves to provide emotion regulation literature (e.g., Zalewski, Len-
protection against environmental stressors and is gua, Wilson, Trancik, & Bazinet, 2011); however,
responsible for assembling and activating the investigations utilizing this approach have not been
resources necessary to cope with stressors. The conducted in response to family conflict during
HPA axis response consists of a cascade of events childhood or adolescence. Pattern-based or person-
resulting in the release of the hormone cortisol. oriented analytic approaches provide unique
Cortisol production is found in response to threat advantages to understanding individual differences
in social contexts (Chen, Cohen, & Miller, 2010), inherent in development. These analytic
including when goals of self-preservation are approaches allow for identifying and examining
threatened (Dickerson & Kemeny, 2004). The HPA qualitatively different sets of responses (Muth!en &
axis is also activated in response to family conflict Muth!en, 2000). Distinct patterns reflecting differ-
(e.g., Davies, Sturge-Apple, & Cicchetti, 2011; Koss ences in how responses relate to one another may
et al., 2013). advance our understanding of the divergence in
Efficient, adaptive HPA responses include effec- functioning that emerges throughout development
tive regulation, activation, and termination of this and allow for uncovering relations that may be
system. According to allostatic load theory masked in variable-centered approaches.
(McEwen & Seeman, 1999), repeated stressors can
lead to alterations in the HPA axis. Two forms of
Present Study and Hypotheses
altered HPA reactivity have been commonly
observed in relation to environmental adversity. This study extends prior research in several key
Hyperactivity of the HPA axis in response to an directions. First, this investigation expands
232 KOSS, CUMMINGS, DAVIES, AND CICCHETTI
assessments: T1–T2 M = 1.18 years, SD = .20; T2- topic in a similar manner as they would in their
T3 M = 1.14 years, SD = .22). The majority of the home. As a goal for the task, families were also
240 families included in analyses at T1 were instructed to work toward a resolution or solution
retained throughout the longitudinal study (T1 to to their problem during the 7-minute period. Dis-
T2 retention rate: 93.3%; T2 to T3 retention rate: cussions were videotaped. Upon completion of the
94.2%). There were no differences at T1 for families discussion, each family member completed a ques-
lost to attrition between T1 and T3 and those tionnaire about the task. The FPST was conducted
retained throughout the study period among T1 consistent with other established procedures for tri-
study variables or most family demographics adic parent–child interaction tasks (e.g., Gordis,
including family income, child gender, marital sta- Margolin, & John, 2001; Lindahl & Malik, 1999).
tus, and race/ethnicity. However, adolescents lost Families discussed a wide variety of topics
to attrition were older (M = 13.31 years, SD = .55) (more than 20 distinct topics). Frequently chosen
compared to adolescents retained in the study topics included the following: responsibilities and
(M = 13.05 years, SD = .52; F(1,234) = 6.12, p < .05). chores (20.3%); sibling relationships (18.6%); clean-
At each time point, adolescents’ teachers were ing (8.1%); rules (9.7%); video game, computer,
recruited to complete survey questions about the and/or television use (6.4%); and school (5.9%).
child through the mail. Teachers were contacted The discussed topic was most often identified by
based on maternal and adolescent nominations of the child (58.2%; 57.8% mother; 43.9% father) and
the teacher that knew the child the best. Teachers was identified by two or more family members in
were chosen if they taught core subjects (e.g., 48.1% of families suggesting agreement among
math, English, science, history) to maximize consis- family members regarding the sources of family
tency among the type of classroom settings in discord. To assess whether the identified regulation
which they interacted with the adolescent. The patterns were due to differences in the laboratory
majority of adolescents had teacher assessments FSPT, adolescents provided evaluations of the task.
(T1 N = 218; T2 N = 183, T3 N = 184). Across all Adolescents rated how similar the discussion
time points, 96.6% of teachers reported knowing resembled disagreements occurring in the home on
the child moderately or very well (length of time a 7-point Likert scale (1 = a lot more negative,
knowing the child in months: T1 M = 11.62, 4 = about the same, 7 = a lot more positive). Adoles-
SD = 9.37; T2 M = 14.12, SD = 10.58; T3 M = 13.03, cents also rated the seriousness of the discussed
SD = 9.85). topic in their relationship with their parents on a 6-
point Likert scale (1 = not at all, 6 = a whole lot).
Procedure
Saliva collection. Adolescents provided one
At each of the annual assessments, mothers, pretask and three post-task saliva samples to capture
fathers, and adolescents visited the laboratory HPA reactivity to the FPST. Adolescents rinsed their
designed to resemble a home living space. Parents mouths with water 10 min prior to providing the
and adolescents provided consent and assent prior series of samples to reduce the number of contami-
to each family session. Parents were provided mon- nants in the saliva. Samples were collected through
etary compensation for their time and adolescents passive drool with the aid of a straw. A pretask sal-
received a gift card for their time. Teachers pro- iva sample was collected approximately 40 min after
vided consent and received monetary compensa- arrival to the laboratory setting to allow sufficient
tion for each completed survey packet. time for the HPA axis to return to baseline levels. To
capture adolescents’ reactivity to the family dis-
Triadic family problem-solving task (FPST). At agreement, post-task samples were collected 10, 20,
T1, mothers, fathers, and adolescents engaged in a and 30 min after the peak of the FPST. The peak of
7-minute problem-solving discussion task designed the FPST was predetermined to be the midpoint of
to elicit adolescents’ stress and regulatory the discussion to capture responses during the dis-
responses during a family disagreement. Prior to cussion should the family begin to come to a resolu-
the start of the task, each family member was tion during the latter portion of the task. This
asked to individually identify a problematic topic resulted in the collection of saliva samples at 14, 24,
for their family. Families were then given a 2-min- and 34 min after the start of the 7-min FPST. Fami-
ute period to collectively decide on one topic to lies visited the laboratory in the late afternoon and
discuss. Families were instructed to discuss the early evening hours to minimize the effects of the
234 KOSS, CUMMINGS, DAVIES, AND CICCHETTI
diurnal cortisol pattern (M pretask sampling time Interactions and Family Functioning (SCIFF; Lin-
5:23 pm; SD = 1 hr 52 min). dahl & Malik, 2000). The individual-level child
codes provided global ratings of adolescents’ affec-
tive-behavioral responses during the task. Video-
Measures
tapes were coded using the Anger and Frustration,
Adolescent self-reports of emotional dis- Withdrawal, Opposition/Defiance, and Positive
tress. Following the completion of the FPST, ado- Affect scales. Each behavioral and emotional
lescents provided self-reports of the intensity of response was assessed on a 5-point Likert scale
specific emotions felt during the task. Adolescents ranging from 1 (very low) to 5 (high) reflecting the
reported how much they felt each emotion on a degree to which each response was exhibited by
6-point Likert scale ranging from 0 (not at all) to 5 the adolescent during the FPST. The Anger and
(a whole lot). Emotions included subjective reports Frustration scale assessed the overall level of anger,
of feeling angry, sad, scared, worried, upset, and irritation, and frustration exhibited by the child
happy throughout the FPST. Data reduction strate- coding for verbalizations, overt behavior, and emo-
gies were employed to reduce the overlap among tional tone indicative of anger or frustration. The
highly correlated specific emotions. Scared and Withdrawal scale assessed the degree to which the
worried feelings (r = .57, p < .001) were averaged child retreated from or avoided the discussion.
to create a composite report of fear, and angry and This code captured the degree to which the child
upset feelings were averaged to create a composite emotionally or physically shut down or backed off
report of anger (r = .71, p < .001). This resulted in from the discussion. Body language, attitude, and
self-reports of emotional responses in this study tone of voice were considered in the withdrawal
utilizing one-item reports of feeling sad and happy code. The Opposition/Defiance scale captured the
and composite reports of feeling mad (angry and extent to which the child exhibits deliberate disre-
upset items; a = .83) and afraid (scared and wor- spectful, noncompliant, argumentative, or distract-
ried items; a = .69). ing behaviors. The Positive Affect scale assessed
the degree to which the child was happy, affection-
Cortisol. Saliva samples were assayed for sali- ate, and relaxed during the discussion by assessing
vary cortisol using a highly sensitive immunoassay the child’s body language, facial expressions, and
at Salimetrics Inc. (State College, PA). The assay tone of voice. The SCIFF has good concurrent and
test process utilized 25 ll of saliva, and samples constructive validity (Lindahl & Malik, 2001).
were tested in duplicate form. The test had a lower Twenty percent of videotaped interactions were
test sensitivity of .007 lg/dl and an upper test sen- coded by a separate trained coder to calculate relia-
sitivity of 3.00 lg/dl. The average intra-assay coef- bility in the current sample. Reliability coefficients
ficient was 5.75% for the current sample. Samples reported are single-item intraclass correlation coef-
were examined for outliers of four standard devia- ficients (ICC) based on a one-way random effects
tions above or below the mean. Cortisol variables analysis of variance model and absolute agreement
for adolescents with outliers were dropped from of raters. Correlations and ICCs based on a subset
analyses. Adolescents with missing cortisol assess- of 50 tapes were Anger/Frustration: r = .77,
ments or outliers were included in analyses with ICC = .77; Withdrawal: r = .76, ICC = .75; Opposi-
missing values for their cortisol variables (n = 7). tion/Defiance: r = .82, ICC = .81; and Positive
Cortisol reactivity was calculated using the area Affect: r = .70, ICC = .65.
under the curve (AUC) with respect to ground
(AUCg) utilizing the pretask and three post-task Depressive symptoms. Adolescents completed
cortisol assessments (Pruessner, Kirschbaum, the Center for Epidemiologic Studies—Depression
Meinlschmid, & Hellhammer, 2003). Time of day scale (CES-D; Radloff, 1977) at each of the time
was controlled for in pretask and AUC variables points. The 20-item measure assesses depressive
using residualized scores. Residualized scores were symptomology over the past week. Adolescents
created by subtracting predicted scores controlling reported how frequently they experienced each
for time of day from observed scores. item on a 4-point Likert scale; scores were summed
with higher scores indicating more depressive
Observations of adolescent affective-behavioral symptoms. The CES-D had adequate internal relia-
responses. Adolescents’ affective-behavioral re- bility in the current sample (a range: .86–.87).
sponses were observationally coded from the Scores of 16 or higher indicate clinical levels of
videotaped FPST using the System for Coding depression in adult samples. Adolescents in this
PATTERNS OF ADOLESCENT REGULATION 235
study reported scores of 16 or greater at 18.4% Adolescent and teacher reports were significantly
(T1), 19.3% (T2), and 23.4% (T3). The CES-D has correlated (T1–T3 r range: .33–.46, all ps < .001). The
been found to be suitable for assessing depression composite SDQ conduct problems scale had ade-
during adolescence (Radloff, 1991; Roberts, quate internal reliability in the current sample (T1–
Andrews, Lewinsohn, & Hops, 1990). T3 a range: .71–.75).
236
Correlations and Descriptive Statistics
Note. Pretask and AUCg cortisol are residualized variables. *p < .05; **p < .01.
PATTERNS OF ADOLESCENT REGULATION 237
a member (LPA model results provide the proba- subjective distress, displayed lower levels of
bility of being assigned to each distinct profile). In observable distress, and had relatively low cortisol
simulation studies, most likely class membership responses to the FPST. Three additional patterns
has been found to be a suitable method for examin- emerged among the remaining adolescents. A pat-
ing auxiliary variables when entropy is .80 or tern consistent with under-controlled (UC) regulation
greater (Clark & Muth!en, 2009). Entropy is a statis- emerged (15%, n = 37). These adolescents dis-
tic reflecting the degree of classification, with val- played higher levels of overt opposition/defiance
ues of 1.0 reflecting perfect classification. Entropy and overt anger/frustration coupled with elevated
for the best-fitting model in the current study was subjective reports of feeling angry. A pattern con-
.95; thus, participants were classified into the cate- sistent with over-controlled (OC) regulation also
gorical regulation pattern with which they had the emerged (9%, n = 22). These adolescents displayed
highest probability of being a member. increased levels of overt withdrawal along with
elevated subjective reports of feeling scared,
angry, and sad. Lastly, a pattern emerged consist-
Four-Class Solution of Adolescent Regulation
ing of adolescents who displayed a larger increase
Patterns
in cortisol reactivity in response to the FPST.
Adolescents’ emotional, behavioral, and cortisol These physiologically reactive (PR) adolescents (9%,
responses to family conflict were included in the n = 21) were similar to the adaptively regulated
latent profile analysis. Means for all indicators adolescents in terms of low subjective distress and
and variances of the cortisol variables were free to overt behavioral responses; however, these youth
vary across profiles. One-, two-, three-, four-, and displayed higher pretask and increased cortisol
five-profile models were examined. The five-pro- reactivity.
file model was not identified. Results of the LPA
support a four-profile solution (see Table 2). Ado- Demographic group differences. A chi-squared
lescents were assigned to the regulation profile in test was conducted to examine differences in child
which they had the highest probability of being a gender among the four patterns; results revealed
member. Average posterior probabilities for class no differences in the proportion of boys and girls
membership of the four-class model ranged from among the four patterns of regulation (v2(3) = 3.44,
.97 to .99; values of 1.00 indicate perfect classifica- ns). A one-way analysis of variance (ANOVA) was
tion. conducted to determine whether family income, T1
Means and standard deviations of model indi- adolescent age, time of day of the FPST, and wake
cators for each regulation profile are displayed in time differed across the four regulation patterns.
Table 3. Figure 1 displays the deviation of each Results revealed no significant differences among
profile mean from the sample mean for the model the four patterns (F(3,234) = .32, ns; F(3,232) = .58,
indicators. Each profile was given a descriptive ns; F(3,235) = 1.57, ns; F(3,235) = .21, ns, respec-
label based on the pattern of regulation indicators. tively).
The majority of adolescents (67%, n = 160) were ANOVAs and chi-square difference tests were
classified into an adaptively regulated (AR) pattern. conducted to examine whether regulation patterns
These adolescents reported lower levels of were due to different characteristics of the FPST.
There were no differences among the four pat-
terns in the degree to which adolescents reported
TABLE 2 that the FPST resembled disagreements in the
Model Selection Criteria for Comparison of the 1-, 2-, 3-, 4-, and home (F(3,234) = 1.26, ns sample M = 4.76,
5-Class Models SD = 1.24; Likert scale point 4 = about the same),
No. Classes BIC Entropy BLRT
the degree to which adolescents reported the seri-
ousness of the topic discussed (F(3,237) = 1.27, ns;
1 6595.38 – – sample M = 3.89, SD = 1.26), the topic discussed
2 6083.90 .92 <.001 (v2(63) = 63.39, ns), or which combination of fam-
3 5842.99 .94 <.001 ily members generated the selected topic
4 5691.14 .95 <.001
(v2(18) = 8.05, ns). Collectively, these results sug-
5 Model not identified – – –
gest that the adolescent regulation patterns were
Note. BIC = Bayesian information criteria. BLRT = parametric not attributed to differences in the specific topics
bootstrapped likelihood ratio test. Bolded values indicated best- discussed or the degree to which the topic was
fitting model. problematic for the family.
238 KOSS, CUMMINGS, DAVIES, AND CICCHETTI
TABLE 3
Profile Means and Standard Deviations for Self-Report of Emotions, Observational Affective Behaviors, and Physiological Reponses
During the Family Problem-Solving Task
Profile
Self-Report Emotions
Mad 1.26ade (1.47) .46ac (.73) .50df (.76) 2.57cef (1.38) .78 (1.14)
Scared .26e (.48) .24c (.54) .24f (.58) 1.18cef (1.03) .33 (.65)
Sad .30e (.57) .18c (.46) .24f (.44) 3.00cef (.87) .47 (.97)
Happy 2.17 (1.81) 2.36c (1.60) 3.00f (1.64) 1.45cf (1.71) 2.30 (1.67)
Observational Affective Behaviors
Anger/Frustration 3.57ade (1.28) 1.48ac (.70) 1.10df (.30) 2.09cef (1.23) 1.83 (1.15)
Withdrawal 2.05ae (.97) 2.60a (1.32) 2.52 (1.44) 3.14e (1.32) 2.56 (1.31)
Opposition/Defiance 4.54ade (.69) 1.43ac (.69) 1.29df (.78) 2.14cef (1.28) 1.96 (1.36)
Positive Affect 1.97 (1.07) 2.05 (1.01) 1.95 (1.28) 1.77 (.81) 2.00 (1.03)
Physiological Responses
Adjusted Pretask Cortisol Level .00ad (.04) !.02ab (.03) .16bcd (.13) !.02c (.03) .00 (.07)
Adjusted Area Under the Curve !.58d (1.26) !.82b (1.25) 7.70bcd (7.02) !.66c (1.29) .00 (3.40)
Notes. Superscript denotes mean differences between groups at p < .05. aAR vs. UC comparison; bAR vs. PR comparison; cAR vs.
OC comparison; dUC vs. PR comparison; eUC vs. OC comparison; fPR vs. OC comparison.
…
4
Under-Controlled
1 Adaptively Regulated
Physiologically Reactive
Over-Controlled
-1
Multigroup Latent Growth Curve Modeling problems and conduct problems. Full-information
maximum-likelihood estimation was utilized; this
The structural equation modeling (SEM) framework
approach allows for the inclusion of participants
was used to fit multigroup linear latent growth
with partial data. A chi-square difference test was
models (LGM) to examine differences in the growth
used to compare nested models with parameters
of adolescent mental health symptoms among regu-
free to vary or constrained to be equal among the
lation patterns identified in the LPA. Separate LGMs
groups. Residual error variances were constrained
were fit using manifest variables of adolescent
to be equal across time points. Adolescent age was
reports of symptoms of depression and anxiety and
examined as a time-varying covariate of mental
composite teacher and adolescent reports of peer
PATTERNS OF ADOLESCENT REGULATION 239
health symptoms; age was not a significant covariate adolescents reporting the lowest levels of anxiety.
of symptoms in any of the four subsequent models PR adolescents had increasing anxiety symptoms
and was thus excluded as a covariate in the subse- across early adolescence. Wald tests of parameter
quent model results presented. constraints were conducted to examine differences
in intercept and slope means among the regulation
Depressive symptoms. A chi-square difference patterns. OC adolescents had significantly higher
test was used to compare nested models to examine levels of concurrent anxiety symptoms compared to
whether a model with group-specific parameters fit UC, PR, and AR adolescents (v2(1) = 10.02, p < .01;
the data better than a constrained model. The con- v2(1) = 4.85, p < .05; v2(1) = 4.74, p < .05, respec-
strained model fit the data significantly worse tively). Additionally, UC adolescents had signifi-
(v2(30) = 82.21) than the model with the means of cantly lower initial levels of anxiety symptoms
the intercept and slope parameters and the residual compared to AR adolescents (v2(1) = 4.08, p < .05).
variances free to vary among the four regulation The PR adolescents had significantly greater growth
patterns (v2(21) = 54.31, Ddf = 9, v2 differ- compared to AR adolescents (v2(1) = 5.47, p < .05)
ence = 27.90, p < .001), providing support for group and a marginal trend compared to the UC adoles-
differences in growth of depressive symptoms. The cents (v2(1) = 3.62, p = .06).
linear growth curves of depressive symptoms for
each regulation pattern are depicted in Figure 2a. Peer problems. A multigroup model with
Group-specific parameters are displayed in Table 4. means of the intercept and slope variables and the
OC adolescents had elevated depressive symptoms residual variance free to vary across groups fit the
at T1. OC, UC, and AR adolescents had stable levels data significantly better (v2(21) = 54.71) than a
of depressive symptoms across early adolescence. model with parameters constrained to be equal
PR adolescents had the lowest levels of depressive among groups (v2(30) = 73.31, Ddf = 9, v2 differ-
symptoms at T1; however, this group also had sig- ence = 18.60, p < .05). Linear growth curves of peer
nificant growth and the steepest rise in symptoms problem symptoms for each regulation pattern are
across early adolescence. Wald tests of parameter depicted in Figure 2c, and group-specific parame-
constraints were conducted to examine differences ters are displayed in Table 4. UC adolescents had
in intercept and slope means among the regulation elevated levels of peer problems at T1. The slope fac-
patterns. OC, UC, and AR adolescents had higher tor for changes in peer problem symptoms was non-
initial depressive symptoms compared to the PR significant for UC, OC, and AR adolescents. PR
group (v2(1) = 10.82, p < .001; v2(1) = 5.81, p < .05; adolescents displayed the lowest concurrent
v2(1) = 6.84, p < .01, respectively). Additionally, levels of peer problems; however, these adolescents
there was a trend for OC adolescents reporting showed significant increases in peer problems across
higher initial depressive symptoms compared to early adolescence. Wald tests of parameter con-
AR adolescents (v2(1) = 3.28, p = .07). PR adoles- straints were conducted to examine differences in
cents had significantly greater growth in depressive intercept and slope means among the regulation pat-
symptoms compared to AR and UC adolescents terns. UC adolescents had significantly higher initial
(v2(1) = 9.28, p < .01; v2(1) = 6.31, p < .05, levels of peer problems compared to the PR adoles-
respectively). cents (v2(1) = 5.85, p < .05). PR adolescents had sig-
nificantly greater growth in peer problems
Anxiety symptoms. The constrained model fit compared to AR adolescents (v2(1) = 9.75, p < .01).
the data significantly worse (v2(30) = 54.09) than the PR adolescents also had significantly different
model with the means of the intercept and slope growth curves compared to UC adolescents
parameters and the residual variances free to vary (v2(1) = 11.58, p < .001).
among the four regulation patterns (v2(21) = 27.02;
Ddf = 9, v2 difference = 27.07, p < .01), providing Conduct problems. A multigroup model with
support for group differences in the growth of anxi- group-specific parameters did not fit the data better
ety symptoms. Linear growth curves of anxiety (v2(21) = 53.00) than a constrained model
symptoms for each regulation pattern are depicted (v2(30) = 69.45, Ddf = 9, v2 difference = 16.45, ns)
in Figure 2b, and group-specific parameters are dis- suggesting no group differences in growth curves
played in Table 4. OC adolescents had stable, ele- of conduct problems. Subsequently, an intercept-
vated anxiety symptoms across the study period. only (no-growth) model was fit to examine group
UC and AR adolescents had stable levels of anxiety differences in T1 conduct problems. The multi-
symptoms across early adolescence with UC group model with group-specific intercept means
240 KOSS, CUMMINGS, DAVIES, AND CICCHETTI
(a) (b)
16 13
12
14
Depressive Symptoms
11
Anxiety Symptoms
12
10
10 9
8
8
7
6
6
4 5
T1 T2 T3 T1 T2 T3
(c) (d)
2.0
2.2
1.8
1.8 1.4
1.6 1.2
1.4 1.0
0.8
1.2 Under-Controlled
0.6
1.0 Adaptively Regulated
0.4 Physiologically Reactive
0.8 0.2 Over-Controlled
0.6 0.0
T1 T2 T3 T1 T2 T3
FIGURE 2 Trajectories of adolescent depressive, anxiety, peer problem, and conduct problem symptoms for each regulation pattern.
Note. Conduct problem trajectories reflect intercept-only, no-growth model results.
TABLE 4
Multigroup Growth Parameters for Change in Adolescent Mental Health Symptoms by Regulation Pattern
Regulation Patterns
Adaptively Physiologically
Under-Controlled Regulated Reactive Over-Controlled
Depressive Symptoms
Intercept Mean 10.76*** 10.03*** 5.59*** 13.62***
Variancea 42.15***
Slope Mean !.18 .08 3.44*** .79
Variancea 13.28***
Anxiety Symptoms
Intercept Mean 6.06*** 8.18*** 7.30*** 11.28***
Variancea 22.82***
Slope Mean .03 .02 1.55* .19
Variancea 3.18**
Peer Problem Symptoms
Intercept Mean 1.76*** 1.35*** .89** 1.46***
Variancea 1.02***
Slope Mean !.17 .02 .54*** .16
Variancea .07
Conduct Problem Symptomsb
Intercept Mean 1.53*** 1.04*** .92*** 1.26***
Variancea .75***
Note. aVariances constrained to be equal across regulation patterns. bAn intercept-only, no-growth model was fit for Conduct Problems.
*p ≤ .05, **p ≤ .01, ***p ≤ .001.
and residual variances (v2(27) = 60.33) fit the data p < .05) suggesting differences in T1 intercepts. UC
significantly better than the constrained model adolescents had the highest rates of T1 conduct
(v2(33) = 75.79, Ddf = 6, v2 difference = 15.46, problems (see Figure 2d). Wald tests of parameter
PATTERNS OF ADOLESCENT REGULATION 241
constraints were conducted to examine differences broader developmental contexts and contribute to
in intercept means among the regulation patterns. maladjustment. Consistent with this notion, the
UC adolescents had significantly higher rates of T1 multigroup latent growth curve analyses indicated
conduct problems compared to AR and PR adoles- differences in adolescents’ concurrent and subse-
cents (v2(1) = 5.50, p < .05; v2(1) = 3.94, p < .05, quent mental health symptoms.
respectively). Adolescents characterized as under-controlled
exhibited elevated levels of observable anger,
opposition, and defiance during family conflict in
DISCUSSION
combination with increased feelings of anger dur-
The present study contributes to the growing litera- ing the task. Adolescents who utilize disruptive
ture examining differences in youth’s regulation pat- regulation strategies during conflict may also apply
terns reflecting unique constellations of regulatory these strategies of delinquent and aggressive
strategies utilized during times of stress. Consistent behaviors to other broader developmental contexts.
with hypotheses, distinct patterns of adolescent As expected, under-controlled adolescents had
responses during a family conflict task were identi- higher concurrent levels of conduct problems (see
fied across multiple systems, including adolescents’ also Cummings, 1987; Maughan & Cicchetti, 2002).
subjective distress, observable affective behaviors, Under-controlled adolescents also reported ele-
and physiological stress responses. The under-con- vated levels of peer problems. The behaviors asso-
trolled, over-controlled, and adaptively regulated ciated with an under-controlled regulation pattern,
patterns were consistent with previous research on namely increased anger, defiance, and opposition,
younger children’s patterns of responses to conflict likely produce social situations in which peers find
(e.g., Cummings, 1987; El-Sheikh et al., 1989; interaction with these adolescents difficult. On the
Maughan & Cicchetti, 2002) suggesting that higher other hand, over-controlled adolescents were char-
order regulation patterns remain prevalent in early acterized by withdrawn and avoidant behaviors
adolescence. Unique to this investigation, the emer- during the family discussion. In addition, these
gence of the physiologically reactive pattern high- adolescents reported elevated levels of subjective
lights the importance of including additional distress, including increased rates of feeling anger,
indicators of stress, beyond behavioral and subjec- fear, and sadness during the family conflict task.
tive distress, to capture adolescents at risk for mental Although withdrawn behaviors may serve to mini-
health problems. Adopting a narrow focus on any mize feelings of emotional insecurity, they may
one level of analysis would have missed identifying also place adolescents at risk for internalizing prob-
youth at heightened risk for concurrent or emerging lems. This pattern of responses to family conflict
mental health problems. was also related to feelings of emotional distress
Consistent with research in community sam- and withdrawal in broader contexts of functioning.
ples, a large majority of adolescents displayed Consistent with hypotheses, elevated anxiety and
adaptive regulation in the face of family conflict, depressive symptoms remained stable across the
that is, low levels of observable behavioral disrup- study for these adolescents. Both under-controlled
tions and subjective distress during the family and over-controlled patterns were associated with
conflict task were characteristic of these adoles- concurrent heightened mental health symptoms,
cents. This group evidenced stability in low men- suggesting that regulation strategies serving as
tal health symptoms across adolescence. For many mechanisms for disruptions in broader functioning
youth, conflict may represent a manageable stres- may be well established prior to adolescence for
sor that does not pose a threat to their sense of some youth. Research on emotional insecurity dur-
security and the family represents a context in ing childhood does find support for the longitudi-
which they build effective regulation skills. While nal prediction of heightened internalizing and
the majority of the sample was characterized as externalizing problems prior to adolescence (Cum-
adaptively regulated, three distinct patterns of mings & Davies, 2010). Because heightened symp-
dysregulation emerged. The family context pro- toms were already present at the FPST assessment,
vides adolescents with everyday opportunities to we cannot assess directionality between regulation
encounter and manage stress. In accordance with strategies in the family context and the emergence
emotional security theory, dysregulated strategies of mental health symptoms for the under-con-
for handling family stressors may serve to allevi- trolled and over-controlled adolescents. Use of lon-
ate stress for adolescents in the short term; how- gitudinal designs earlier in development will be
ever, these strategies may lead to problems in necessary to disentangle whether the specific
242 KOSS, CUMMINGS, DAVIES, AND CICCHETTI
patterns uncovered in the present study serve as necessary to assess whether this heightened HPA
precursors to specific constellations of mental activity is associated with newly emerging disrup-
health symptoms. tions in the family system.
Lastly, a physiologically reactive pattern There were no differences in youth’s cortisol reac-
emerged; these adolescents displayed marked acti- tivity among the under-controlled, over-controlled,
vation of the HPA axis in response to the family and adaptively regulated adolescents. Elevated
conflict task. Despite this increased cortisol levels of cortisol in response to the FPST were found
response, these adolescents displayed effective reg- among a minority of participants; this is consistent
ulation of their affective and behavioral responses. with investigations of cortisol during conflict para-
This group may represent a sophisticated form of digms and is hypothesized to reflect individual dif-
regulation resulting in masking of observable dis- ferences in experiences with family adversity
tress. However, masking strategies are hypothe- (Gunnar, Talge, & Herrera, 2009). Previous research
sized to be associated with increased subjective on basal cortisol levels and mental health supports
distress (Davies & Forman, 2002); the present group the notion that concurrent mental health symptoms
of adolescents did not report elevated levels of sub- are associated with lower HPA activity (Shirtcliff &
jective distress. While the physiologically reactive Essex, 2008). Both the under-controlled and over-
adolescents were observably and subjectively simi- controlled groups had elevated concurrent mental
lar to the adaptively regulated adolescents, there health problems at the time of the family conflict
were differences in their trajectories of mental task. which may contribute to the lack of an elevated
health. The physiologically reactive adolescents had cortisol response for these groups. However, lack of
lower concurrent depressive symptoms and peer distinction from the adaptively regulated group pre-
problems; however, these adolescents displayed cludes interpretations of low cortisol as blunted
pronounced growth in depressive, anxiety, and responses in the present investigation that have been
peer problem symptoms across the study period. found in previous investigations of family adversity
This pattern of regulation may represent a group of and the HPA axis (e.g., Saxbe et al., 2012).
adolescents who, although effectively regulated and There were no differences in the proportion of
well adjusted during childhood, experience emerg- girls and boys in each regulation pattern. The lack of
ing regulation problems during adolescence. This gender differences is consistent with previous
activation of the HPA axis in response to conflict research on child gender and strategies used to reg-
may serve as precursor to later problems. The ulated exposure to conflict experienced in the family
results in the present investigation are consistent (Cummings & Davies, 2010). Research supports the
with research on adolescents’ mental health and notion that conflict may be an equally threatening
HPA functioning; Shirtcliff and Essex (2008) found context for both boys and girls. Given the small sam-
elevated basal cortisol was related to increases in ple sizes within the dysregulated regulation pat-
mental health problems across the transition to ado- terns, gender differences were not examined in
lescence. Repeated activation of the HPA axis in mental health growth curves; however, gender dif-
response to family stressors can take a toll on the ferences in internalizing and externalizing problems
body and may result in subsequent downregulation become more pronounced throughout adolescence.
of the HPA axis. Longitudinal research is needed to The findings in the present study support the
examine whether the heightened HPA activity in notion that differences in regulation patterns serve
response to conflict that preceded the emergence of as a mechanism to different forms of maladjustment
mental health problems in the present study is fol- and provide further evidence that regulation strate-
lowed by a blunting of the system over time. The gies utilized in the family may relate to general
identification of this physiologically reactive pattern strategies and approaches for dealing with stress in
highlights the need to incorporate multiple levels of multiple developmental contexts. Additional
analysis in the study of stress regulation. The lack research is necessary to understand whether the pat-
of correspondence between physiological and terns identified in the present study develop as an
behavioral regulation has been supported in pat- adaptation to the family context or reflect inherent
tern-based approaches in infants exposed to marital differences in the way in which the child interacts
conflict (Towe-Goodman, Stifter, Mills-Koonce, & with the world. Youth’s regulation strategies in
Granger, 2011), suggesting activation of the HPA response to family stressors are expected to not only
response may reflect an at-risk group that remains reflect the immediate context of discord (e.g., the
unidentified in studies solely using behavioral FPST paradigm), but also reflect their histories of
assessments of regulation. Future research will be experience with discord and family adversity
PATTERNS OF ADOLESCENT REGULATION 243
(Cummings & Davies, 1996). Thus, additional assessments that would allow for the creation of
research is necessary to examine how youth’s expo- latent variables to better model the underlying con-
sure to family adversity impacts the development of struct of adolescent mental health. Given the nature
specific profiles of regulation. Drawing on emotional of the task and measures, distinctions between reac-
security theory, it is expected that all three dysregu- tivity and regulation in adolescents’ emotional,
lated patterns would be associated with heightened behavioral, and physiological responses were not
family adversity. Histories of family adversity and possible and these responses likely capture both
the timing of risk likely contribute to the develop- reactivity and regulation. Nevertheless, the method-
ment of specific patterns of responses. Future ology in the present study allowed for assessing the
research is needed to disentangle the contributions unique family context in which adolescents experi-
of strategies developed as a result of characteristics ence and respond to discord on a regular basis.
of family adversity and intrinsic characteristics of Lastly, it should be noted that the groups identified
youth, such as temperament and genetic factors that through the latent profile analysis may not reflect
account for individual differences in the manifesta- true subgroups within the population (Bauer &
tion of specific forms of insecurity despite exposure Shanahan, 2007). However, the regulation profiles
to similar risky environments. identified are consistent with previous investiga-
The present study is not without limitations. The tions within the literature during childhood (Cum-
sample utilized in the present study is a community mings, 1987; Maughan & Cicchetti, 2002) and the
sample, largely represented by well-functioning profiles identified in the present study relate to
families. Consistent with this notion, the majority of unique constellations of adolescent outcomes.
adolescents displayed an adaptively regulated pat- Future research replicating the identified regulation
tern of responses. This resulted in unequal and patterns is necessary.
smaller sample sizes among the dysregulated pat- The present study provides initial support for
terns and may contribute to a reduction in power to the continuance of regulation patterns during ado-
detect effects. Future studies utilizing a balance of lescence. Adolescents utilize a variety of strategies
normative and high-risk families may allow for a to handle and cope with conflict; however, not all
better understanding of differences in the dysregu- adolescents utilize the same strategies for manag-
lated patterns. While the use of the family conflict ing stress and differences in these strategies may
task increased the ecological validity of how adoles- relate to differences in mental health trajectories.
cents respond to family conflict, this task dimin- Employing person-oriented approaches for under-
ished generalizability and control such that not all standing higher order regulation patterns is a
families experienced the same levels of family con- promising direction for regulation research. Pat-
flict throughout the task. Differences in the occur- terns of regulation represent coherent goal-directed
rence and manner in which families handle conflict organizations of individual responses, providing a
may differentially result in the necessity for some new understanding of the role of these strategies in
adolescents to employ strategies to manage a stress- the larger network of regulatory process that may
ful situation. However, adolescents reported no dif- lead to unique trajectories of adolescent mental
ferences in the degree of similarity between the health.
FPST and conflict in the home or the seriousness of
the issue discussed during the FPST among the dif-
REFERENCES
ferent regulation patterns. Although the present
study sought to examine families at yearly intervals, Achenbach, T. M., McConaughy, S. H., & Howell, C. T.
there is variation in the length of time between mea- (1987). Child/adolescent behavioral and emotional
surement occasions due to accommodating families’ problems: Implications of cross-informant correlations
busy lives that may contribute additional error in for situational specificity. Psychological Bulletin, 101,
the longitudinal growth models. While the magni- 213–232. doi:10.1037/0033-2909.101.2.213
Bauer, D. J., & Shanahan, M. J. (2007). Modeling complex
tude of correlations across reports in the present
interactions: Person-centered and variable-centered
study is consistent with the larger literature (e.g.,
approaches. In T. D. Little, J. A. Bovaird, & N. A. Card
Achenbach et al., 1987; Muris et al., 2004), it should (Eds.), Modeling contextual effects in longitudinal studies
also be noted that correlations between adolescent (pp. 255–283). Mahwah, NJ: Lawrence Erlbaum.
and teacher report of behavior problems were low Chen, E., Cohen, S., & Miller, G. E. (2010). How low
in some cases, indicating little shared variance socioeconomic status affects 2-year hormonal trajecto-
among these different reporters. Future research ries in children. Psychological Science, 21, 31–37.
would benefit from additional reporters and doi:10.1177/0956797609355566
244 KOSS, CUMMINGS, DAVIES, AND CICCHETTI
Cicchetti, D., Ganiban, J., & Barnett, D. (1991). Contribu- Personality and Social Psychology, 46, 939–949.
tions from the study of high-risk populations to under- doi:10.1037/0022-3514.46.4.939
standing the development of emotion regulation. In J. Fries, E., Hesse, J., Hellhammer, J., & Hellhammer, D. H.
Garber, & K. A. Dodge (Eds.), The development of emo- (2005). A new view on hypocortisolism. Psychoneuroen-
tion regulation and dysregulation. (pp. 15–48). New York, docrinology, 30, 1010–1016. doi:10.1016/j.psyneuen.20
NY: Cambridge University Press. 05.04.006
Clark, S. L., & Muth!en, B. (2009). Relating latent class Goodman, R. (1997). The strengths and difficulties ques-
analysis results to variables not included in the analysis. tionnaire: A research note. Journal of Child Psychology
https://www.statmodel.com/download/relatinglca.pdf and Psychiatry, 38, 581–586. doi:10.1111/j.1469-
Cole, P. M., Zahn-Waxler, C., Fox, N. A., Usher, B. A., & 7610.1997.tb01545.x
Welsh, J. D. (1996). Individual differences in emotion Goodman, R., Ford, T., Simmons, H., Gatward, R., &
regulation and behavior problems in preschool chil- Meltzer, H. (2000). Using the strengths and difficulties
dren. Journal of Abnormal Psychology, 105, 518–529. questionnaire (SDQ) to screen for child psychiatric dis-
doi:0.1037/0021-843X.105.4.518 orders. British Journal of Psychiatry, 177, 534–539.
Cummings, E. M. (1987). Coping with background anger doi:10.1192/bjp.177.6.534
in early childhood. Child Development, 58, 976–984. Gordis, E. B., Margolin, G., & John, R. S. (2001). Parents’
doi:10.2307/1130538 hostility in dyadic marital and triadic family settings
Cummings, E. M., & Davies, P. T. (1996). Emotional security and children’s behavioral problems. Journal of Consult-
as a regulatory process in normal development and the ing and Clinical Psychology, 69, 727–734. doi:10.1037/
development of psychopathology. Development and Psy- 0022-006X.69.4.727
chopathology, 8, 123–139. doi:10.1017/S0954579400007008 Gunnar, M. R., Talge, N. M., & Herrera, A. (2009). Stres-
Cummings, E. M., & Davies, P. T. (2010). Marital conflict sor paradigms in developmental studies: What does
and children: An emotional security perspective. New and does not work to produce mean increases in sali-
York, NY: Guilford Press. vary cortisol. Psychoneuroendocrinology, 34, 953–976.
Davies, P. T., & Cummings, E. M. (1994). Marital conflict doi:10.1016/j.psyneuen.2009.02.010
and child adjustment: An emotional security hypothe- Gunnar, M. R., & Vazquez, D. V. (2006). Stress neurobiol-
sis. Psychological Bulletin, 116, 387–411. doi:10.1037/ ogy and developmental psychopathology. In D. Cic-
0033-2909.116.3.387 chetti, & D. J. Cohen (Eds.), Developmental
Davies, P. T., & Forman, E. M. (2002). Children’s patterns psychopathology (Vol. 2): Developmental neuroscience (pp.
of preserving emotional security in the interparental 533–577). Hoboken, NJ: John Wiley.
subsystem. Child Development, 73, 1880–1903. Koss, K. J., George, M. R. W., Davies, P. T., Cicchetti, D.,
doi:10.1111/1467-8624.t01-1-00512 Cummings, E. M., & Sturge-Apple, M. L. (2013). Pat-
Davies, P. T., Sturge-Apple, M. L., & Cicchetti, D. (2011). terns of children’s adrenocortical reactivity to inter-
Interparental aggression and children’s adrenocortical parental conflict and associations with child adjustment:
reactivity: Testing an evolutionary model of allostatic A growth mixture modeling approach. Developmental
load. Development and Psychopathology, 23, 801–814. Psychology, 49, 317–326. doi:10.1037/a0028246
doi:10.1017/S0954579411000319 Lindahl, K. M., & Malik, N. M. (1999). Observations of
Davies, P. T., Sturge-Apple, M. L., Cicchetti, D., & Cum- marital conflict and power: Relations with parenting in
mings, E. M. (2008). Adrenocortical underpinnings of the triad. Journal of Marriage and the Family, 61, 320–
children’s psychological reactivity to interparental con- 330. doi:10.2307/353751
flict. Child Development, 79, 1693–1706. doi:10.1111/ Lindahl, K. M., & Malik, N. M. (2000). System for Coding
j.1467-8624.2008.01219.x Interactions and Family Functioning (SCIFF). Unpub-
Dickerson, S. S., & Kemeny, M. E. (2004). Acute stressors lished manual, University of Miami, Miami, FL.
and cortisol responses: A theoretical integration and Lindahl, K. M., & Malik, N. M. (2001). The System for
synthesis of laboratory research. Psychological Bulletin, Coding Interactions and Family Functioning. In P. K.
130, 355–391. doi:10.1037/0033-2909.130.3.355 Kerig, & K. M. Lindahl (Eds.), Family observational cod-
Eisenberg, N., Cumberland, A., Spinrad, T. L., Fabes, R. ing systems: Resources for systemic research (pp. 75–90).
A., Shepard, S. A., Reiser, M., . . . Guthrie, I. K. (2001). Philadelphia, PA: Brunner/Mazel.
The relations of regulation and emotionality to chil- Lisonbee, J. A., Pendry, P., Mize, J., & Gwynn, E. P.
dren’s externalizing and internalizing problem behav- (2010). Hypothalamic-pituitary-adrenal and sympa-
ior. Child Development, 72, 1112–1134. doi:10.1111/1467- thetic nervous system activity and children’s behav-
8624.00337 ioral regulation. Mind, Brain, and Education, 4, 171–181.
El-Sheikh, M., Cummings, E. M., & Goetsch, V. L. (1989). doi:10.1111/j.1751-228X.2010.01096.x
Coping with adults’ angry behavior: Behavioral, physio- Maughan, A., & Cicchetti, D. (2002). Impact of child mal-
logical, and verbal responses in preschoolers. Developmen- treatment and interadult violence on children’s emo-
tal Psychology, 25, 490–498. doi:10.1037/0012-1649.25.4.490 tion regulation abilities and socioemotional
Fleming, R., Baum, A., & Singer, J. E. (1984). Toward an adjustment. Child Development, 73, 1525–1542.
integrative approach to the study of stress. Journal of doi:10.1111/1467-8624.00488
PATTERNS OF ADOLESCENT REGULATION 245
McEwen, B. S., & Seeman, T. (1999). Protective and dam- using the Center for Epidemiologic Studies of Depres-
aging effects of mediators of stress: Elaborating and sion scale. Psychological Assessment: A Journal of Consult-
testing the concepts of allostasis and allostatic load. ing and Clinical Psychology, 2, 122–128. doi:10.1037/1040-
Annals of the New York Academy of Sciences, 896, 30–47. 3590.2.2.122
doi:10.1111/j.1749-6632.1999.tb08103.x Sameroff, A. J. (2009). Conceptual issues in studying the
Muris, P., Meesters, C., Eijkelenboom, A., & Vincken, M. development of self-regulation. In S. L. Olson & A. J.
(2004). The self-report version of the strengths and dif- Sameroff (Eds.), Biopsychosocial regulatory processes in
ficulties questionnaire: Its psychometric properties in the development of childhood behavioral problems (pp. 238–
8- to 13-year old non-clinical children. British Journal of 257). New York, NY: Cambridge University Press.
Clinical Psychology, 43, 437–448. doi:10.1348/ Saxbe, D. E., Margolin, G., Shapiro, L. A. S., & Baucom,
0144665042388982 B. R. (2012). Does dampened phsyiolgoical reactivity
Muth!en, B. O., & Muth!en, L. K. (2000). Integrating per- protect youth in aggressive family environments? Child
son-centered and variable-centered analyses: Growth Development, 83, 821–830. doi:10.1111/j.1467-
mixture modeling with latent trajectory classes. Alco- 8624.2012.01752.x
holism: Clinical and Experimental Research, 24, 882–891. Shanahan, L., McHale, S. M., Osgood, D. W., & Crouter,
doi:10.1111/j.1530-0277.2000.tb02070.x A. C. (2007). Conflict frequency with mothers and
Muth!en, L. K., & Muth!en, B. O. (1998–2010). Mplus user’s fathers from middle childhood to late adolescence:
guide (6th ed.). Los Angeles, CA: Muth!en & Muth!en. Within- and between-families comparisons. Develop-
Nylund, K. L., Asparouhov, T., & Muth!en, B. O. (2007). mental Psychology, 43, 539–550. doi:10.1037/0012-
Deciding on the number of classes in latent class anal- 1649.43.3.539
ysis and growth mixture modeling: A Monte Carlo Shields, A. M., Cicchetti, D., & Ryan, R. M. (1994). The
simulation study. Structural Equation Modeling, 14, 535– development of emotional and behavioral self-regula-
569. doi:10.1080/10705510701575396 tion and social competence among maltreated school-
Pruessner, J. C., Kirschbaum, C., Meinlschmid, G., & age children. Development and Psychopathology, 6, 57–75.
Hellhammer, D. H. (2003). Two formulas for computa- doi:10.1017/S0954579400005885
tion of area under the curve represent measures of Shirtcliff, E. A., & Essex, M. J. (2008). Concurrent and
total hormone concentration versus time-dependent longitudinal associations of basal and diurnal cortisol
change. Psychoneuroendocrinology, 28, 916–931. with mental health symptoms in early adolescence.
doi:10.1016/S0306-4530(02)00108-7 Developmental Psychobiology, 50, 690–703. doi:10.1002/
Radloff, L. S. (1977). The CES-D scale: A self-report dev.20336
depression scale for research in the general population. Smith, C. L., & Eisenberg, N. (2005). Self-regulation. In C.
Applied Psychological Measurement, 1, 385–401. B. Fisher, & R. M. Lerner (Eds.), Encyclopedia of applied
doi:10.1177/014662167700100306 developmental science (pp. 977–980). Thousand Oaks,
Radloff, L. S. (1991). The use of the Center for Epidemio- CA: Sage.
logic Studies of Depression scale in adolescents and Thompson, R. A. (1994). Emotion regulation: A theme in
young adults. Journal of Youth and Adolescence, 20, 149– search of definition. Monographs of the Society for
166. doi:10.1007/BF01537606 Research in Child Development, 59, 25–52. doi:10.2307/
Repetti, R. L., Robles, T. F., & Reynolds, B. (2011). Allostatic 1166137
processes in the family. Development and Psychopathology, Thompson, R. A., Lewis, M. D., & Calkins, S. D. (2008).
23, 921–938. doi:10.1017/S095457941100040X Reassessing emotion regulation. Child Development Per-
Repetti, R. L., Taylor, S. E., & Seeman, T. E. (2002). Risky spectives, 2, 124–131. doi:10.1111/j.1750-8606.2008.
families: Family social environments and the mental 00054.x
and physical health of offspring. Psychological Bulletin, Towe-Goodman, N. R., Stifter, C. A., Mills-Koonce, W.
128, 330–366. doi:10.1037/0033-2909.128.2.330 R., Granger, D. A., & The Family Life Project Key
Reynolds, C. R. (1980). Concurrent validity of what I Investigators. (2011). Interparental aggression and
think and feel: The Revised Children’s Manifest Anxi- infant patterns of adrenocortical and behavioral stress
ety scale. Journal of Consulting and Clinical Psychology, responses. Developmental Psychobiology, 54, 685–699.
48, 774–775. doi:10.1037/0022-006X.48.6.774 doi:10.1002/dev.20618
Reynolds, C. R., & Paget, K. D. (1983). National normative Zalewski, M., Lengua, L. J., Wilson, A. C., Trancik, A., &
and reliability data for the Revised Children’s Manifest Bazinet, A. (2011). Emotion regulation profiles, temper-
Anxiety scale. School Psychology Review, 12, 324–336. ament, and adjustment problems in preadolescents.
Reynolds, C. R., & Richmond, B. O. (1978). What I think Child Development, 82, 951–966. doi:10.1111/j.1467-
and feel: A revised measure of children’s manifest anx- 8624.2011.01575.x
iety. Journal of Abnormal Child Psychology, 6, 271–280. Zeman, J., Cassano, M., Perry-Parrish, C., & Stegall, C.
doi:10.1007/BF00919131 (2006). Emotion regulation in children and adolescents.
Roberts, R. E., Andrews, J. A., Lewinsohn, P. M., & Hops, Developmental and Behavioral Pediatrics, 27, 155–168.
H. (1990). Assessment of depression in adolescents doi:10.1097/00004703-200604000-00014