Está en la página 1de 13

Child Development, November/December 2003, Volume 74, Number 6, Pages 1869 – 1880

Adolescents’ Emotion Regulation in Daily Life: Links to Depressive


Symptoms and Problem Behavior
Jennifer S. Silk, Laurence Steinberg, and Amanda Sheffield Morris

This study examined links between emotion regulation and adjustment in a sample of 152 adolescents in
Grades 7 (M age 5 12) and 10 (M age 5 15). Emotion regulation was assessed using the experience sampling
method, in which adolescents provided multiple reports about the intensity, lability, and strategies used to
regulate negative emotions across 1 week. Adolescents also completed self-report measures of adjustment.
Adolescents who reported more intense and labile emotions and less effective regulation of these emotions also
reported more depressive symptoms and problem behavior. Responding to negative emotions with
disengagement (e.g., denial) or involuntary engagement (e.g., rumination) was less effective in regulating
negative affect, and greater use of these strategies was related to higher levels of depressive symptoms and
problem behavior.

The concept of emotion regulation is popular among adolescence, studies indicate that adolescents ex-
researchers and clinicians seeking to elucidate indi- perience more frequent and intense emotions than
vidual differences in the development of psycho- younger or older individuals (Larson, Csikszentmi-
pathology. According to Thompson’s (1994, pp. 27 – 28) halyi, & Graef, 1980; Larson & Lampman-Petraitis,
widely cited definition, ‘‘emotion regulation consists 1989). Second, many of the hormonal, neural, and
of the extrinsic and intrinsic processes responsible cognitive systems thought to underlie the regulation
for monitoring, evaluating, and modifying emotional of emotion appear to mature throughout the
reactions, especially their intensive and temporal adolescent period (Spear, 2000). Third, the preva-
features, to accomplish one’s goals.’’ Emerging the- lence of various forms of psychopathology, includ-
ories of both child and adult psychopathology frame ing affective and behavioral disorders, increases
maladjustment in terms of difficulties regulating affect dramatically during the adolescent period. A better
(Bradley, 2000; Steinberg & Avenevoli, 2000). understanding of emotion regulation during adoles-
Adolescence is an opportune time to explore cence may help us understand individual differ-
processes and correlates of emotion regulation for ences in mental health and adjustment during this
several reasons. First, the transition through adoles- period of increased risk.
cence is accompanied by physical, psychological, Emotion regulation may be implicated in diverse
and social transformations that elicit novel experi- forms of adolescent psychopathology, including
ences of emotional arousal. Consistent with the notion both internalizing disorders and externalizing dis-
that emotional experience is especially intense in orders. Internalizing disorders, such as depression,
likely involve deficits in the ability to down-regulate
negative emotions such as sadness or difficulty up-
Jennifer S. Silk, Western Psychiatric Institute and Clinic; regulating and maintaining positive emotions (Cole,
Laurence Steinberg, Department of Psychology, Temple Univer- Michel, & Teti, 1994). Depressed children and
sity; Amanda S. Morris, Department of Psychology, University of adolescents may also lack facility with strategies
New Orleans. used by other children and adolescents to ameliorate
This research was supported by the John D. and Catherine
negative affect, such as problem solving or cognitive
T. MacArthur Foundation Research Network on Psychopathology
and Development. An earlier version of this manuscript served as restructuring (Dodge & Garber, 1991). Externalizing
a part of the first author’s doctoral dissertation. Appreciation is disorders are characterized by behavioral dysregula-
expressed to the students, teachers, administrative assistants, and tion; however, dysregulated affect, particularly
administrators of the two participating schools. We also wish to anger, is also theorized to be a feature of these
acknowledge the thoughtful and instructive review of an earlier disorders (Bradley, 2000; Dearing et al., 2002).
version of this manuscript by Philip C. Kendall, Jerome Resnick,
Brian Marx, Denise Sloan, and Julie Hubbard, and the generous In exploring the link between emotion regulation
assistance of Ronald Dahl and his staff in data coding and data and adolescent psychosocial adjustment, we focused
management.
Correspondence concerning this article should be addressed to
Jennifer S. Silk, 4423 Sennott Square, 210 South Bouquet Street, r 2003 by the Society for Research in Child Development, Inc.
Pittsburgh, PA 15260. Electronic mail may be sent to jss4@pitt.edu. All rights reserved. 0009-3920/2003/7406-0019
1870 Silk, Steinberg, and Morris

on two interrelated components of emotion regula- Goldsmith, 1998; Gilliom, Shaw, Beck, Schonberg, &
tion: (a) emotional dynamics, including the intensity, Lukon, 2002; Grolnick, Bridges, & Connell, 1996).
lability, and down-regulation of emotional states; Related findings come as well from the literature on
and (b) emotion-regulation strategies, the behavioral coping among adolescents. Studies of coping in
and cognitive processes adolescents use to modulate adolescence suggest that approaching or engaging
or change affective states. With respect to emotional with the stressor is associated with better adjustment
dynamics, Larson and colleagues presented relevant (Compas, Connor, Saltzman, Thomsen, & Wads-
findings from an experience sampling study in worth, 1999). In contrast, depressive symptomatol-
which adolescents recorded their mood states seven ogy is associated with the use of avoidance and
times a day in response to signals from an electronic rumination, as well as limited use of active strategies
pager. Findings suggested that adolescents who are such as support seeking, problem solving, and
more emotionally labile (e.g., who have a greater cognitive restructuring (Herman-Stahl, Stemmler, &
degree of fluctuation in mood state), or who report Petersen, 1995; Nolen-Hoeksema & Morrow, 1993;
more intense negative affect, tend to score higher on Sandler, Tein, & West, 1994).
measures of depressive symptomatology (Larson, Compas and colleagues (Connor-Smith, Compas,
Raffaelli, Richards, Ham, & Jewell, 1990). Little is Wadsworth, Thomsen, & Saltzman, 2000) recently
known about the relation between emotional dy- presented an empirically based classification of
namics and externalizing problems among adoles- coping dimensions that has been replicated among
cents; however, research conducted with younger diverse samples of adolescents (Langrock, Compas,
children demonstrates that children high in negative Keller, Merchant, & Copeland, 2002; Wadsworth &
affective intensity exhibit higher rates of externaliz- Compas, 2002). In this model, responses to stress are
ing behavior over time (Eisenberg et al., 1995; first distinguished on the basis of whether they are
Eisenberg et al., 1997). Interestingly, Eisenberg and voluntary or involuntary. Voluntary responses, also
colleagues found that children with internalizing called controlled responses, are volitional strategies
and externalizing problems exhibit elevated levels of enacted by the individual. Involuntary responses,
both anger and sadness (Eisenberg et al., 2001). also called automatic responses, are responses that
Few studies have directly examined the relation may or may not be under conscious awareness and
between emotion-regulation strategies and adjust- are not enacted volitionally. Involuntary or auto-
ment among adolescents. In a notable exception, matic responses, such as rumination, are typically
Garber and colleagues conducted a series of studies experienced as intrusive and often seem to ‘‘pop
in which young adolescents reported on their typical into’’ one’s mind and take on a life of their own
emotion-regulation strategies. They found that ado- beyond the control of the individual.
lescents diagnosed with a depressive disorder Voluntary or controlled responses are next differ-
nominated fewer problem-focused and active dis- entiated with respect to whether they involve active
traction strategies, but more avoidant, passive, and engagement with, or passive disengagement from,
aggressive strategies, than adolescents in a control the stressor. Within voluntary engagement, a final
group (Garber, Braafladt, & Zeman, 1991; Garber, distinction is made between primary and secondary
Braafladt, & Weiss, 1995). Depressed adolescents control (Thurber & Weisz, 1997). Primary control
also reported lower expectations that strategies involves engaging with a stressor in an attempt to
would be effective in altering their negative emo- modify negative aspects of the situation, whereas
tions than adolescents in the control group (Garber secondary control is aimed at maximizing one’s fit to
et al., 1995). In these studies, however, adolescents the situation through such mechanisms as distrac-
reported on the expected efficacy and the estimated tion or cognitive restructuring. Compas and collea-
frequency of strategies, and it remains unclear how gues reported that primary and secondary control
well these reports reflect actual emotion-regulation responses are associated with fewer depressive
behaviors in affectively charged, real-life contexts. symptoms and less aggression, whereas disengaged
These findings from studies of adolescents are and involuntary responses are associated with
consistent with those from research on infants and higher depressive symptomatology and aggression
young children, which suggest that active and (Langrock et al., 2002; Wadsworth & Compas, 2002).
attention-based strategies such as self-distraction Although this hierarchy is intended to apply to
and attention shifting are linked with decreases in responses to stress, it can also be used to classify
distress, whereas passive focus on the distressing strategies for regulating emotion. It should be noted
stimulus is associated with increases in distress and that the use of the term strategies does not imply
the development of externalizing behaviors (Buss & that all emotion-regulation strategies are conscious,
Adolescent Emotion Regulation 1871

volitional, or adaptive. In fact, involuntary engage- whereas disengagement and involuntary engage-
ment strategies are by nature likely to occur outside ment strategies would be associated with less
of volitional control but are still referred to as effective regulation of negative affect. We also
strategies in this study. expected that greater usage of primary and second-
A fundamental limitation of most of the studies ary control strategies would be associated with
reviewed here on both coping and emotion regula- lower levels of depressive symptomatology and
tion is their acontextual nature. In most studies, problem behavior, whereas greater usage of disen-
adolescents are either told to imagine that a gagement and involuntary engagement strategies
particular stressful or emotional experience has just would be associated with higher levels of depressive
occurred or are asked to nominate a stressful or symptomatology and problem behavior.
emotional experience that has occurred over the past
week or month. Adolescents then answer a series of Method
questions about how they would or did respond to
Participants
such a situation. It remains unclear whether the
strategies and regulatory processes endorsed by Participants were recruited from two public
adolescents on checklists and in response to vign- schools in a suburban community outside of a major
ettes adequately reflect their actual behaviors in Northeastern city. During assemblies in each school,
affectively charged situations. The ‘‘hot cognition’’ research assistants described the project to all 7th-
that occurs during affective experience has now and 10th-grade students and provided each student
cooled, and responses may reflect what adolescents with a packet containing a description of the study; a
know they should do in such situations, rather than parental consent form; and a stamped, self-ad-
what they actually do when they are influenced by dressed envelope. Students were instructed to share
the heat of the emotion. In fact, research comparing the materials with their parents and to return a
retrospective reports of coping with real-time reports signed parental consent form if they wished to
of coping during the same period has revealed participate in the study. Packets were distributed to
several biases in retrospective reporting of coping approximately 350 students across both schools, and
(Stone et al., 1998). 185 students returned parental consent forms. This
In the current study, we used the experience 53% participation rate is within the range of 45% to
sampling method (ESM) to circumvent many of 75% reported in other ESM studies using adolescent
these limitations. The ESM was modified to provide samples (Larson, 1989; Larson et al., 1980; Larson &
detailed data not only on the intensity and lability of Richards, 1994). Based on the order in which consent
adolescents’ emotions but also on the regulation of forms were received, 170 students were selected for
emotional states in response to specific emotional participation. School personnel requested that 3 be
experiences. This method allowed us to gather removed from the study because of pending dis-
information on emotion regulation as it actually ciplinary action, resulting in 167 students enrolled
unfolds in adolescents’ daily lives. The goals of the in the study. Twelve participants were dropped from
study were twofold: (a) to address the effectiveness the analyses because of incomplete or inaccurate
of specific emotion-regulation strategies in regulat- data, resulting in a final sample of 152 adolescents.
ing negative affect in adolescent daily life, and (b) to Data were not available for all participants on all
explore associations between patterns of emotional measures.
dynamics and strategy use and concurrent adoles- The final sample was composed of 152 adoles-
cent adjustment. Because changes in the neural and cents (73 males, 79 females) ages 12 to 17 (M 5 13.89,
hormonal substrate as well as the social context of SD 5 1.65). The sample consisted of 80 7th-grade
emotion regulation are thought to occur across the students (M age 5 12.43) and 72 10th-grade students
adolescent period, we also explored whether emo- (M age 5 15.51). Approximately 89% of the partici-
tion-regulation processes operate in a similar fashion pants were Caucasian, 5% were African American,
for both early and middle adolescents. 3% were Latino, and 3% were from other ethnic
We hypothesized that adolescents who reported groups. Approximately 64% of the adolescents
more intense and labile emotions and more difficulty reported that their parents were married and 36%
regulating negative affect would report greater reported that their parents were divorced, separated,
symptoms of depression and more problem behav- never married, or widowed. Participants came
ior. With respect to strategies, we expected that primarily from working- and middle-class families,
primary and secondary control strategies would be as indexed by maternal education. Five percent of
associated with greater regulation of negative affect, the mothers had not completed high school, 35%
1872 Silk, Steinberg, and Morris

held a high school diploma or GED, 15% had com- Participants first provided information about
pleted some college, and 22% percent of the mothers their current affect and activities. To assess emotion
held a college degree. Thirty adolescents did not regulation, we solicited information about one
know their mother’s highest level of education. experience over the past hour perceived by the
Students were compensated with $20, a gift adolescent as emotionally salient. Adolescents were
certificate to a local movie theatre, and entry into a prompted to think over the previous hour and
weekly raffle drawing for a stereo or video game record what they had been doing. They were then
system. Teachers with at least one participating provided with the following prompt: ‘‘Try to
student in their class were compensated with $10 at remember your feelings and thoughts over the past
the conclusion of the study for their help in hour. Think about the time when you felt the worst,
distributing materials to students. or most negative (e.g., mad, upset, disappointed,
sad, worried).’’ Participants were asked to estimate
the time at which this occurred in 15-min intervals
Procedure
ranging from ‘‘about an hour ago’’ to ‘‘right before I
Data were collected in groups of 15 to 20 was beeped.’’ After identifying this experience,
participants in weeklong sampling periods. At the participants responded to a series of questions
beginning of the designated week, students attended designed to obtain information on emotional dy-
a group orientation during which the research team namics and the use of strategies in relation to the
described the procedures of the study and adminis- identified emotion-eliciting experience. Experiences
tered a battery of questionnaires. After obtaining included both discrete events (e.g., ‘‘I failed my math
voluntary assent from each participant (parental test’’) and emotion-eliciting thoughts (e.g., ‘‘I was
consent had already been received), the research thinking about how much I miss my ex-boyfriend’’).
team reviewed a sample Experience Sampling Form Although we did not expect that adolescents would
(ESF) with participants. At the conclusion of the experience a significant negatively valenced emo-
meeting, students were given binders containing tional experience in every sampling interval, we
ESFs and preprogrammed wristwatches. requested that they identify their most negative
Beginning immediately after the orientation, parti- experience on every form, even if it was something
cipants were instructed to complete an ESF each time very minor. Participants also were asked to report on
their wristwatch beeped. A systematic sampling a positive emotional experience at the end of each
approach was used to determine signaling schedules. sampling form to avoid inadvertently encouraging
Within the target hours of each day, one signal was the adolescent to dwell on negative experiences.
programmed for every 90- to 150-min block, using a The compliance rate in the current study was 88%,
table of random numbers, with the provision that no with participants completing ESFs for an average of
signals occur within 60 min of each other. During 42 of 48 signals. Participants who responded to fewer
school days, signaling began at 8:00 a.m. and ended by than 50% of signals were considered noncompleters.
9:30 p.m. On Friday and Saturday nights, the signaling In addition, participants whose responses appeared
period was extended to 11:00 p.m., and on Saturday invalid because they had filled out all of the responses
and Sunday mornings, the signaling period began at at once (indicated by pen color and mismatch
10:00 a.m. Both schools agreed to allow students to between times and activities) or because responses
complete ESFs while in school. At the end of the week, were obviously fictitious were also considered non-
participants met with research staff to receive pay- completers. Data from 12 participants could not be
ment and participate in a debriefing interview. analyzed for these reasons.
Participants recorded data on current thoughts
and feelings in response to a brief electronic signal
Measures
from their wristwatch. Thoughts and experiences
were recorded on an ESF, adapted from Larson Emotional dynamics. Emotional dynamics were
(1989) and influenced by the work of Stone et al. assessed through participants’ ratings of anger,
(1998). The ESF was a brief 2- to 5-min survey used sadness, and anxiety on each ESF. The words mad,
to assess emotional dynamics and strategies in sad, and nervous were chosen to reflect each of these
response to experiences over the previous hour. emotions in terms that have been shown to be
The window of 1 hour was chosen to maximize understandable by children as young as fourth
chances of accessing a naturally occurring emotional grade (Laurent et al., 1999). Each ESF elicited ratings
experience while minimizing biases of retrospective for current affect and peak affect. Current affect was
recall over a longer period. defined as the participants’ affect at the time at
Adolescent Emotion Regulation 1873

which they completed the form (e.g., ‘‘How are you acceptance, distraction, positive thinking), (c) disen-
feeling right now?’’). Participants rated the degree to gagement (e.g., denial, avoidance, wishful thinking,
which they felt mad, sad, or nervous at the current escape, inaction), and (d) involuntary engagement
time. Participants also rated their peak affect, (e.g., rumination, impulsive or involuntary action).
defined as their affect during the negative emotion- Involuntary disengagement and voluntary disen-
eliciting experience previously identified. As de- gagement strategies were combined into one cate-
scribed earlier, participants were prompted to think gory to simplify the model for the present analyses.
over the past hour and record their most negative ex- Dichotomous variables were constructed to indicate
perience within the past hour. They were then asked the presence or absence of each of the four categories
to rate the degree to which they felt mad, sad, or of emotion-regulation strategies for each sampling
nervous at the worst point of this experience. Current moment. For some analyses, strategy use was repre-
and peak ratings were made on a 5-point scale sented as a proportion of sampling moments in which
ranging from 1 (not at all) to 5 (very much) for each the strategy was used out of the entire number of
emotion. relevant sampling moments for that participant (e.g.,
Indexes of intensity, lability, and regulation were containing a peak affect rating of 3 or greater).
constructed from ratings of current and peak affect. Demographics questionnaire. Adolescents com-
Intensity was calculated separately for anger, sadness, pleted a brief questionnaire designed to obtain basic
and anxiety by averaging current affect ratings for demographic information, including the partici-
each emotion across ESF reports. Thus, intensity pant’s age, gender, race, maternal education (used
reflected the mean level of anger, sadness, or anxiety to index socioeconomic status), and family structure.
that the adolescent typically experienced in his or her Youth Self-Report Form. Problem behavior was
daily life. Lability was computed for each emotion assessed using the Externalizing scale of the Youth
by calculating the standard deviation of current Self-Report Form (YSR; Achenbach, 1991). The YSR
affect ratings for that emotion across ESF reports is an adolescent self-report checklist of behavior
(Larson et al., 1990). Lability therefore reflected the problems during the past 6 months, each rated on a
adolescent’s typical degree of fluctuation within each 3-point scale. The Externalizing scale reflects out-
emotion across a week. Regulation of negative affect wardly directed negative behaviors, including lying,
was assessed by comparing current levels of anger, swearing, arguing, stealing, truancy, and destroying
sadness, and anxiety with peak levels of anger, property. The YSR has been extensively normed on a
sadness, and anxiety on each ESF report. In other nationally representative sample of adolescents.
words, a report of high peak anger and low current Achenbach (1991) reported acceptable internal con-
anger would suggest greater regulation, whereas a sistency, test – retest reliability, and discriminant and
report of high peak anger and moderate to high convergent validity for the YSR.
current anger would suggest lower regulation. Children’s Depression Inventory. Symptoms of de-
Emotion-regulation strategies. Emotion-regulation pression were assessed using the Children’s Depres-
strategies were examined for sampling moments sion Inventory (CDI; Kovacs, 1985). The CDI is a
containing an affect rating of 3 or greater (N 5 3,756), widely used measure of depressive symptomatology
indicating that the participant experienced a nega- in both clinically referred and nonclinical samples
tive emotion of at least a moderate degree of of children and adolescents. The CDI assesses
intensity (mean sampling moments with affect depressed mood and a wide range of associated
ratings of 3 or greater 5 25). On each ESF, adoles- symptoms, including somatic and interpersonal
cents were asked to describe how they reacted to the symptoms. Respondents choose one of three re-
identified emotion-eliciting experience by checking sponses that best describes them for each of 27 items.
one of the strategies on a checklist provided or by The CDI possesses adequate internal consistency
writing in their own response. The checklist in- (a 5 .87) and test – retest reliability (Kovacs, 1985).
cluded 13 strategies adapted from the Responses to
Stress Questionnaire (RSQ; Connor-Smith et al.,
2000). Responses written in the blank space pro-
Results
vided were also coded into these 13 strategies, with
kappa coefficients ranging from .92 to 1.00.
Data Analytic Procedures
The 13 emotion-regulation strategies were
grouped into four categories: (a) primary control Special analytic techniques were needed to
(e.g., problem solving, emotional expression), (b) accommodate the nested structure of the data, which
secondary control (e.g., cognitive restructuring, included repeated reports for each individual. Data
1874 Silk, Steinberg, and Morris

were collected for 6,560 moments in time (Level 1 emotion regulation; however, noncompleters were
data); however, these moments were nested within more likely to be male (w2 5 8.16, po.01) and
152 individuals (Level 2 data), therefore violating reported more problem behavior (M 5 64.9) than
assumptions of independence. Hierarchical linear completers (M 5 54.7; t 5  2.58, po.05). No differ-
modeling (HLM) is a regression procedure devel- ences between completers and noncompleters ex-
oped for modeling nested or hierarchical data (Bryk isted with respect to age, socioeconomic status, race,
& Raudenbush, 1992). HLM employs an iterative or depressive symptomatology.
process to estimate regression equations for Level 2 Gender and age differences. There were several
data (persons) as a step toward arriving at the best gender differences in emotional dynamics and
regression equation for the entire set of sampling emotion-regulation strategies. As shown in Table 1,
moments. HLM has the important advantage of females reported higher intensity of anger, sadness,
remaining unbiased in the presence of unequal and anxiety than males. Female adolescents also
numbers of observations across individuals and reported greater lability of sadness and anxiety but
unequally spaced sampling moments. Analyses not anger. Females used more primary control and
were conducted using the HLM 5.02 software involuntary engagement than males, although there
package (Bryk & Raudenbush, 1992). Full maximum were no gender differences in secondary control or
likelihood estimation was used for all models. In disengagement. Contrary to expectations, there were
some cases, mean scores or proportions aggregated few age differences in emotional dynamics, except
across sampling moments were used in more that 7th graders reported a higher intensity of anx-
traditional analytic approaches (e.g., correlations, iety than did 10th graders. No differences in strategy
analysis of covariance [ANCOVA]). use were found between 7th and 10th graders.

Preliminary Analyses Emotional Dynamics


Completers versus noncompleters. Means and stan- Correlations between emotional dynamics and adjust-
dard deviations for all major variables are presented ment. It was hypothesized that emotional intensity
in Table 1, and intercorrelations among major and lability would be associated with adjustment. As
variables are presented in Table 2. Noncompleters shown in Table 2, adolescents who reported greater
did not differ from completers on any measures of intensity of anger and sadness reported significantly

Table 1
Means and Standard Deviations of Major Variables for Entire Sample and by Gender

Range M (SD) Male Female t d

Emotional dynamics
Mean anger intensity 1.00 – 2.95 1.41 (0.36) 1.34 1.47  2.34 .36
Mean sadness intensity 1.00 – 3.21 1.34 (0.41) 1.23 1.43  3.12 .49
Mean anxiety intensity 1.00 – 3.31 1.33 (0.41) 1.25 1.40  2.32 .37
Anger lability 0.00 – 1.72 0.77 (0.39) 0.73 0.82  1.56 .23
Sadness lability 0.00 – 1.70 0.64 (0.44) 0.53 0.73  2.80 .45
Anxiety lability 0.00 – 1.95 0.64 (0.43) 0.55 0.73  2.57 .42
Emotion regulation strategies
Primary control 0.00 – 1.00 0.49 (0.23) 0.40 0.59  5.60 .83
Secondary control 0.00 – 1.00 0.47 (0.22) 0.48 0.46 0.43 .09
Disengagement 0.00 – 1.00 0.20 (0.16) 0.19 0.20  0.43 .07
Involuntary engagement 0.00 – .35 0.04 (0.07) 0.02 0.05  2.79 .45
Symptomatology
Depressive symptoms 0.00 – 40.00 8.48 (6.88) 7.41 9.47  1.86 .30
Problem behavior 30.00 – 87.00 54.78 (11.22) 52.77 56.51  1.98 .33
Note. N 5 138 for problem behavior; N 5 152 for all other variables. For males N 5 73 and for females N 5 79. Emotional intensity
represents the mean level of intensity for each emotion across all of the participants’ sampling moments. Strategy use is in proportions of
sampling moments in which the strategy was used out of the entire number of sampling moments for that participant.
po.05. po.01. po.001.
Adolescent Emotion Regulation 1875

Table 2
Correlation Matrix of Variables

1 2 3 4 5 6 7 8 9 10 11

Emotional dynamics
1. Anger intensity
2. Sadness intensity .78
3. Anxiety intensity .61 .59
4. Anger lability .85 .58 .47
5. Sadness lability .67 .83 .54 .68
6. Anxiety lability .53 .49 .85 .54 .57
Emotion regulation strategies
7. Primary control .17 .19 .15 .25 .30 .20
8. Secondary control  .07  .04  .04  .11  .11  .03  .15
9. Disengagement .05 .08 .09 .08 .09 .09 .00 .03
10. Involuntary engagement .20 .22 .28 .22 .24 .39 .19 .11 .31
Symptomatology
11. Depressive symptoms .20 .40 .15 .20 .41 .23 .08  .08 .35 .29
12. Problem behavior .27 .33 .22 .31 .35 .27 .14  .17 .37 .34 .57
Note. N 5 138 for problem behavior; N 5 152 for all other variables. Strategy use is represented in proportions.
po.05. po.01. po.001.

more depressive symptomatology and problem Cluster 2 represents sampling moments in which
behavior. Adolescents who reported greater inten- participants reported high peak levels of anger,
sity of anxiety reported significantly more problem sadness, or anxiety but lower levels of these
behavior and tended to report more depressive emotions in their current ratings. Cluster 1 repre-
symptomatology. A similar pattern of relations sents the larger and less informative set of sampling
emerged with respect to emotional lability. As shown moments in which both peak and current affect
in Table 2, adolescents who reported greater lability ratings were relatively low.
of anger, sadness, and anxiety reported significantly The second step in this analysis was to classify
more depressive symptomatology and problem participants into high- and low-regulation groups
behavior. based on the percentage of their sampling moments
Cluster analysis of regulation. We hypothesized falling into each cluster. First, participants who fell
that participants who showed lower regulation of 1 SD above the mean in percentage of Cluster 3
negative affect would report more depressive symp- sampling moments were classified into a low-
tomatology and problem behavior than participants regulation group (n 5 21). Second, participants who
who showed greater regulation of negative affect. were 1 SD above the mean in percentage of Cluster 2
To test this hypothesis, a two-step cluster analysis sampling moments were classified into a high-
procedure was used. In the first step, cluster analysis regulation group (n 5 23). Table 4 shows the results
was performed on the entire set of sampling of an ANCOVA comparing the high- and low-
moments using current and peak ratings of anger, regulation groups on self-reports of depressive
sadness, and anxiety. The solution was restricted to symptomatology and problem behavior. Gender
three clusters, with the goal of identifying a cluster was included as a covariate given gender differences
of sampling moments representing high regulation revealed in preliminary analyses. As hypothesized,
of negative affect, a cluster of sampling moments adolescents in the low-regulation group scored
representing low regulation of negative affect, and a significantly higher on both depressive symptoma-
cluster of sampling moments in which participants tology and problem behavior than adolescents in the
did not reach high levels of negative affect. Note that high-regulation group.
the unit of analysis in this stage was the sampling
moment, not the individual.
Emotion-Regulation Strategies
As shown in Table 3, Cluster 3 represents a subset
of sampling moments in which participants main- Emotion-regulation strategies were coded for all
tained relatively high ratings of anger, sadness, or sampling moments in which a negative emotion was
anxiety from peak to current ratings, whereas given a peak rating of 3 or greater. This included
1876 Silk, Steinberg, and Morris

Table 3 Time elapsed between peak and current ratings and


Final Mean Cluster Centers gender were included in all models as covariates.
Following the approach described earlier, a sig-
Peak Current
nificant and positive regression coefficient would
Cluster N Mad Sad Nervous Mad Sad Nervous indicate that use of the strategy was associated with
higher levels of current negative affect (e.g., lower
1 4,060 1.91 1.30 1.29 1.17 1.10 1.20
regulation of affect) compared with the use of other
2 1,630 4.09 2.79 1.89 1.26 1.22 1.35
strategies. A significant negative regression coeffi-
3 561 4.16 3.62 2.31 3.57 3.33 2.12
cient would indicate that use of the strategy was
associated with lower levels of current negative
affect (e.g., greater regulation of affect) compared
3,756 of 6,560 sampling moments, indicating that with the use of other strategies. Adolescents’ current
participants experienced a subjectively strong nega- affect was lower than their peak affect for virtually
tive emotion within the hour before sampling in all sampling moments; therefore, the regression
more than half of the instances during which they analyses provide information on the relative amount
were sampled (57%). As Table 1 indicates, partici- of regulation rather than on whether regulation
pants reported using primary and secondary control occurred at all.
strategies relatively frequently (49% and 47%, Table 5 summarizes the regression coefficients for
respectively), whereas participants less frequently each of the hierarchical linear models. Contrary to
reported using strategies involving disengagement our hypotheses, the use of primary control strategies
(20%) and involuntary engagement (4%). was not significantly associated with higher or lower
Hierarchical linear models predicting current intensity levels of regulation. The use of secondary control
of affect from strategy use. A central hypothesis of this strategies, such as cognitive restructuring or positive
study was that primary and secondary control thinking, was marginally associated with greater
strategies would be associated with more effective regulation of anger (coefficient 5  .06, t 5  1.76,
regulation of negative affect, whereas disengage- p 5 .08), although not with the regulation of sadness
ment and involuntary engagement strategies would or anxiety. As hypothesized, strategies involving
be associated with less effective regulation of nega- disengagement, such as avoidance, denial, or escape,
tive affect. This hypothesis was evaluated in a series were significantly associated with lower regulation
of hierarchical linear models with strategy used to of anger (coefficient 5 .14, t 5 3.04, po.01) and sad-
predict current intensity of affect, controlling for ness (coefficient 5 .12, t 5 2.92, po.01), although not
peak intensity of affect. Separate models were anxiety. The same pattern emerged for involuntary
estimated predicting current intensity of anger, engagement strategies, such as rumination, with
sadness, and anxiety from each of the four strategy these strategies significantly associated with lower
categories. Categories were dummy coded, with a 1 regulation of anger (coefficient 5 .40, t 5 3.93, po.001)
indicating that a strategy within the category was and sadness (coefficient 5 .24, t 5 2.45, po.05) but not
used during that sampling moment and a 0 indicat- anxiety.
ing that a strategy from that category was not used. Correlations between strategy use and adjustment. In
addition to examining the utility of particular
strategies in regulating negative affect, we also
Table 4 examined whether there were relations between
Analysis of Covariance (ANCOVA) Comparing Depressive Symptoms strategy use and adolescent depressive symptoma-
and Problem Behavior for High- Versus Low-Regulation Groups, tology and problem behavior. This question was
Controlling for Gender addressed by examining bivariate correlations be-
tween rate of use of each strategy and adolescent
Cluster 2a Cluster 3b reports of adjustment. As shown in Table 2, use of
High Low Cluster
primary control strategies was not correlated with
regulation regulation ANCOVA 2 vs. 3
depressive symptomatology or problem behavior,
M M F Cohen’s d
and secondary control was weakly negatively
Depressive 6.17 14.48  12.35 1.21 correlated with problem behavior only. Use of
symptoms disengagement and involuntary engagement strate-
Problem 52.53 63.20  5.22 .95 gies, was, however, associated with adjustment. As
behavior
shown in Table 2, adolescents who used involuntary
N 5 23. bN 5 21. po.05. po.01.
a
engagement or disengagement strategies at a higher
Adolescent Emotion Regulation 1877

Table 5
Summary of Fixed Effects for Hierarchical Linear Models Predicting Current Intensity from Strategy Use (Controlling for Peak Intensity, Time Since
Event, and Gender)

Mad Sad Nervous

Coefficient (SE) t d Coefficient (SE) t d Coefficient (SE) t d

Strategies
Primary Control .05 (.04) 1.32 .22 .04 (.03) 1.41 .23 .01 (.03) .29 .05
Secondary Control  .06w (.04)  1.76w .29  .05 (.03)  1.60 .26  .00 (.03)  .13 .02
Disengagement .14 (.05) 3.04 .50 .12 (.04) 2.92 .48 .02 (.04) .48 .08
Involuntary engagement .40 (.10) 3.93 .64 .24 (.10) 2.45 .40  .02 (.09)  .20 .03
p
Note. Level 1 N 5 3,752; Level 2 N 5 150. Cohen’s d is based on (2t/ df)
wpo.10. po.05. po.01. po.001.

rate reported more depressive symptomatology and In this study, adolescents who were less likely or
problem behavior. less able to regulate negative affect during real-life
emotional experiences reported more symptomatol-
ogy compared with those who recovered from
Discussion
negative emotional experiences more easily. Adoles-
This study provides initial support for the proposi- cents who were able to recover from feeling sad,
tion that emotion regulation is a central correlate of angry, or anxious were much less likely to report
behavioral and emotional problems among adoles- depressive symptoms and problem behavior than
cents. The similarity of findings across different adolescents who were not able to modulate these
aspects of emotion regulation (e.g., intensity, lability, negative emotions.
regulation patterns, strategy use) and across symp- As hypothesized, greater emotional intensity and
tom measures supports the notion that adolescents lability were associated with elevated depressive
who have problems regulating their emotions are symptomatology and behavior problems, suggesting
more vulnerable to both internalizing and externa- that intensity of emotion and variability in levels of
lizing problems. emotion from moment to moment are also related to
An important strength of the current study is its adolescent adjustment. In line with previous reports
ecological validity. Emotion regulation was assessed and theory (Larson et al., 1990), intensity and lability
in a naturalistic context, providing a lens into real- of sadness were associated with depressive symp-
life, real-time emotions and emotion regulation. This toms, and intensity and lability of anger were
methodology not only minimizes the problems associated with behavior problems. Consistent with
associated with retrospective recall; it also facilitates Eisenberg et al.’s (2001) findings, however, associa-
the recording of hot cognitions and affect-laden tions between emotion regulation and adjustment
responses and reactions. In addition to the ecological also followed along less predictable pathways.
validity of the data, the weeklong sampling period Intense and labile levels of anger and anxiety were
provided a depth and quantity of data on emotion associated with depressive symptoms, and intense
regulation that is more comprehensive and general- and labile levels of sadness and anxiety were related
izable than single-point assessments. The repeated to behavior problems. These findings support the
assessment allowed for the discernment of patterns contention that a nonspecific emotion dysregulation
of emotion regulation for each adolescent. A factorFincluding dysregulation of sadness, anger,
qualitative examination of the adolescents’ re- and anxietyFis a common correlate of both inter-
sponses provides confidence that participants de- nalizing and externalizing among adolescents (see
scribed their emotions and reactions with honesty. Steinberg & Avenevoli, 2000).
For example, the adolescents in this study wrote In addition to exploring the dynamics of emotion
about drug use, sexual activity, secret crushes, regulation, this study also built on previous research
embarrassments, and dreams. They did not describe by examining the effectiveness of the strategies
themselves as cool-headed problem solvers but adolescents use to regulate their emotions. Although
reported that they slammed the door on their sisters, we hypothesized that using primary control strate-
ate junk food until they felt better, or cried them- gies would be especially helpful for adolescents in
selves to sleep. regulating their negative emotions, this was not the
1878 Silk, Steinberg, and Morris

case. It was surprising that responding to feelings of (Garber et al., 1995; Nolen-Hoeksema & Morrow,
anger, sadness, or anxiety by using a primary control 1993; Wadsworth & Compas, 2002). Contrary to
strategy, such as problem solving or emotional hypotheses and previous research (e.g., Compas et
expression, had no systematic effect on the regula- al., 1999; Garber et al., 1995), however, neither
tion of these emotions. There was also little support primary nor secondary control strategies were
for the hypothesized efficacy of secondary control highly associated with adolescent symptomatology.
strategies, such as cognitive restructuring or distrac- The difference in findings may be attributable to
tion, in regulating affect. differences in methodologies, with most previous
We suspect that these strategies were relatively studies employing retrospective or hypothetical
effective in regulating negative affect in some reports of coping and emotion regulation as opposed
instances and relatively ineffective in other in- to the contextual approach employed here.
stances, thus averaging out to a negligible overall In the present study links between strategy use
effect. It also may be that the effectiveness of such and psychopathology were strongest with respect to
strategies depends on the specific nature of the strategies that involved involuntary engagement or
emotional experience. For example, coping research disengagement. To the extent that this finding can be
suggests that primary control strategies are more further replicated it may have implications for
helpful in stressful situations that are control- mental health prevention and intervention programs,
lable and less effective in uncontrollable situations which often focus more on enhancing the use of
(Thurber & Weisz, 1997). Collecting and analyzing ‘‘effective’’ coping skills than diminishing the use of
information about the controllability of the situation ‘‘ineffective’’ strategies. Our findings suggest that
in future research may prove important in further a key prevention and intervention aim with adoles-
evaluating the effectiveness of primary and second- cents might be discouraging the use of counter-
ary control strategies. Relatedly, collecting informa- effective emotion-regulation strategies, such as
tion on other external events and individuals that rumination, escape, avoidance, denial, and inaction.
may have influenced the success or failure of specific An important limitation of the present study
attempts at regulating emotions beyond the influ- is the cross-sectional and correlational nature of
ence of strategy would provide a more stringent test the data. Problems in emotion regulation may be
of the effectiveness of particular strategies and antecedents, concomitants, or consequences of psy-
would help round out our understanding of the chopathology, and the present study cannot shed
social context of emotion regulation. light on the directionality of this link. Nevertheless, a
The emotion-regulation strategies that proved first step in building theories about the relation
most influential were disengagement and invo- between emotion regulation and psychopathology is
luntary engagement, although their impact on re- to establish that difficulties modulating emotions are
gulation was countereffective. Specifically, when indeed correlated with various forms of psycho-
adolescents dealt with their emotions by disengag- pathology. This study provides initial evidence for
ing (e.g., by using denial, avoidance, escape, or wish- this link.
ful thinking) or reacted to their anger or sadness A second limitation is the exclusive use of self-
involuntarily (e.g., by ruminating or acting impul- report measures, which creates the potential for
sively), they maintained higher levels of anger and inflated shared-source variance. It should be noted,
sadness. One possible explanation for the detrimen- however, that although the adolescent was the
tal impact of disengagement and involuntary en- informant in all measures, the method of obtaining
gagement strategies is that they share a passive and information differed between the dependent and
avoidant stance in which the adolescent does little or independent variables, thereby diminishing the
nothing to change or adapt to the negative experi- problem of shared-method variance. Despite evi-
ence. Avoidance and disengagement from a negative dence that adolescents may be better reporters of
experience may also interrupt exposure and extinc- inner experience than other adults, such as parents
tion processes that help the individual habituate to (Lewinsohn, Hops, Roberts, Seeley, & Andrews,
the experience and its associated affect. 1993), it would be important to replicate the present
As hypothesized, strategy use was also related to findings using alternative sources of information on
adolescent adjustment. Consistent with work on adolescent psychopathology, such as clinician, par-
adolescent depression and coping, adolescents in ent, or teacher reports. Additionally, replication of
this study who more frequently used disengagement these findings among clinically referred samples of
and involuntary engagement strategies reported adolescents would also be a crucial step in extending
more depressive symptoms and problem behaviors the validity and generalizability of these findings.
Adolescent Emotion Regulation 1879

Several limitations with respect to the study References


sample should also be noted. First, although the
participation rate in the current sample was consis- Achenbach, T. M. (1991). Manual for the Youth Self-Report
and 1991 Profile. Burlington: University of Vermont,
tent with that of other ESM studies, it is relatively
Department of Psychiatry.
low compared with other types of social research. It Bradley, S. J. (2000). Affect regulation and the development of
may be that the students who participated in the psychopathology. New York: Guilford.
study differ in systematic ways from those who Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear
declined to participate. Other ESM studies that have models: Applications and data analysis methods. Newbury
addressed this issue have revealed small differences Park, CA: Sage.
in the socioeconomic status of participants versus Buss, K. A., & Goldsmith, H. H. (1998). Fear and anger
nonparticipants, although they have not found regulation in infancy: Effects on the temporal dynamics
differences in other demographic or socioemotional of affective expression. Child Development, 69, 359 – 374.
variables (Larson, 1989; Larson, Richards, Sims, & Cole, P. M., Michel, M. K., & Teti, L. O. D. (1994). The
Dworkin, 2001). We did not have data available to development of emotion regulation and dysregulation:
A clinical perspective. Monographs of the Society for
address this issue in the current study; therefore,
Research in Child Development, 59(2-3, Serial No. 240),
caution should be exercised in generalizing the 73 – 100.
findings to the general adolescent population. Cau- Compas, B. E., Connor, J. K., Saltzman, H., Thomsen, A. H.,
tion is also warranted given the limited ethnic and & Wadsworth, M. (1999). Getting specific about coping:
socioeconomic diversity represented in this primarily Effortful and involuntary engagement responses to
Caucasian working- to middle-class sample. stress in development. In M. Lewis & D. Ramsay
Finally, it is important to recognize that in this (Eds.), Soothing and stress (pp. 229 – 256). Mahwah, NJ:
study we were only able to assess a narrow slice of Erlbaum.
the multifaceted construct of emotion regulation. Connor-Smith, J. K., Compas, B. E., Wadsworth, M. E.,
This study focused only on the down-regulation of Thomsen, A. H., & Saltzman, H. (2000). Response to
negative emotions despite emerging evidence of the stress in adolescence: Measurement of coping and
involuntary engagement stress response. Journal of
importance of regulating states of positive affect
Consulting and Clinical Psychology, 68, 976 – 992.
(Mayne & Bonanno, 2001). Similarly, the study’s Dearing, K. F., Hubbard, J. A., Ramsden, S. R., Parker, E.
focus on the down-regulation of negative emotions H., Relyea, N., & Smithmyer, C. M., et al. (2002).
could be misinterpreted to suggest that negative Children’s self-reports about anger regulation: Direct
emotions are always maladaptive. Indeed, negative and indirect links to social preference and aggression.
emotions serve important communicative and moti- Merrill-Palmer Quarterly, 48, 308 – 336.
vational functions and at times can be instrumental Dodge, K. A., & Garber, J. (1991). Emotion and social
toward an individuals’ goals. In the absence of data information processing. In J. Garber & K. A. Dodge
on participants’ goals with respect to each emotional (Eds.), The development of emotion regulation and dysregu-
experience, however, we relied on the admittedly lation (pp. 159 – 181). Cambridge, England: Cambridge
oversimplified view that individuals typically desire University Press.
Eisenberg, N., Cumberland, A., Spinrad, T. L., Fabes, R. A.,
to decrease their experiences of negative emotions.
Shepard, S. A., & Reiser, M., et al. (2001). The relations of
To advance further our understanding of relations regulation and emotionality to children’s externalizing
between emotion regulation and psychosocial ad- and internalizing problem behavior. Child Development,
justment, it will be essential to address the regula- 72, 1112 – 1134.
tion of positive affect as well as individuals’ goals Eisenberg, N., Fabes, R. A., Murphy, B., Maszk, P., Smith,
with respect to regulating their emotions. M., & Karbon, M. (1995). The role of emotionality and
Perhaps the most important message to be gleaned regulation in children’s social functioning: A long-
from this study is that emotion regulation plays a itudinal study. Child Development, 66, 1360 – 1384.
significant role in both internalizing and externalizing Eisenberg, N., Fabes, R. A., Shepard, S. A., Murphy, B. C.,
forms of psychopathology among early and middle Guthrie, I. K., & Jones, S., et al. (1997). Contempora-
adolescents. Emotion-regulation deficits appear non- neous and longitudinal prediction of children’s social
functioning from regulation and emotionality. Child
specific, involving deficits in regulating multiple
Development, 68, 642 – 664.
emotions and conferring risk for a range of psycho- Garber, J., Braafladt, N., & Weiss, B. (1995). Affect
logical disorders. Although this study cannot demon- regulation in depressed and nondepressed children
strate causality, it shows that emotion regulation is and young adolescents. Development and Psychopathol-
either a risk factor or concomitant feature of disorder ogy, 7, 93 – 115.
that should be targeted in mental health prevention Garber, J., Braafladt, N., & Zeman, J. (1991). The regulation
and intervention programs for adolescent populations. of sad affect: An information-processing perspective. In
1880 Silk, Steinberg, and Morris

J. Garber & K. A. Dodge (Eds.), The development of and companionship. American Journal of Community
emotion regulation and dysregulation (pp. 208 – 240). Cam- Psychology, 29, 565 – 597.
bridge, England: Cambridge University Press. Laurent, J., Catanzaro, S. J., Joiner, T. E. Jr., Rudolph, K. D.,
Gilliom, M., Shaw, D. S., Beck, J. E., Schonberg, M. A., & Potter, K. I., & Lambert, S., et al. (1999). A measure of
Lukon, J. L. (2002). Anger regulation in disadvantaged positive and negative affect for children: Scale develop-
preschool boys: Strategies, antecedents, and the devel- ment and preliminary validation. Psychological Assess-
opment of self-control. Developmental Psychology, 38, ment, 11, 326 – 338.
222 – 235. Lewinsohn, P. M., Hops, H., Roberts, R. E., Seeley, J. R., &
Grolnick, W. S., Bridges, L. J., & Connell, J. P. (1996). Andrews, J. A. (1993). Adolescent psychopathology: I.
Emotion regulation in two-year olds: Strategies and Prevalence and incidence of depression and other
emotional expression in four contexts. Child Develop- DSM – III – R disorders in high school students. Journal
ment, 67, 928 – 941. of Abnormal Psychology, 102, 133 – 144.
Herman-Stahl, M. A., Stemmler, M., & Petersen, A. C. Mayne, T. J., & Bonanno, G. A. (2001). Emotions: Current
(1995). Approach and avoidant coping: Implications for issues and future directions. New York: Guilford.
adolescent mental health. Journal of Youth and Adoles- Nolen-Hoeksema, S., & Morrow, J. (1993). Effects of
cence, 24, 649 – 665. rumination and distraction on naturally occurring
Kovacs, M. (1985). The Children’s Depression Inventory depressed mood. Cognition and Emotion, 7, 561 – 570.
(CDI). Psychopharmacology Bulletin, 21, 995 – 998. Sandler, I. N., Tein, J. Y., & West, S. G. (1994). Coping,
Langrock, A. D., Compas, B. E., Keller, G., Merchant, M. J., stress, and the psychological symptoms of children of
& Copeland, M. E. (2002). Coping with the stress of divorce: A cross-sectional and longitudinal study. Child
parental depression: Parents’ reports of children’s Development, 65, 1744 – 1763.
coping, emotional, and behavioral problems. Journal of Spear, L. P. (2000). The adolescent brain and age-related
Clinical Child and Adolescent Psychology, 31, 312 – 324. behavioral manifestations. Neuroscience and Biobehavioral
Larson, R. (1989). Beeping children and adolescents: A Reviews, 24, 417 – 463.
method for studying time use and daily experience. Steinberg, L., & Avenevoli, S. (2000). The role of context in
Journal of Youth & Adolescence, 18, 511 – 530. the development of psychopathology: A conceptual
Larson, R., Csikszentmihalyi, M., & Graef, R. (1980). Mood framework and some speculative propositions. Child-
variability and the psychosocial adjustment of adoles- Development, 71, 66 – 74.
cents. Journal of Youth and Adolescence, 9, 469 – 490. Stone, A. A., Schwartz, J. E., Marco, C. A., Cruise, L. J.,
Larson, R., & Lampman-Petraitis, C. (1989). Daily emo- Shiffman, S., & Hickox, M., et al. (1998). A comparison
tional states as reported by children and adolescents. of coping assessed by ecological momentary assessment
Child Development, 60, 1250 – 1260. and retrospective recall. Journal of Personality and Social
Larson, R., Raffaelli, M., Richards, M. H., Ham, M., & Psychology, 74, 1670 – 1680.
Jewell, L. (1990). Ecology of depression in late childhood Thompson, R. A. (1994). Emotion regulation: A theme in
and early adolescence: A profile of daily states and search of definition. Monographs of the Society for Research
activities. Journal of Abnormal Psychology, 99, 92 – 102. in Child Development, 59(2-3, Serial No. 240), 25 – 52.
Larson, R. W., & Richards, M. H. (1994). Family emotions: Thurber, C. A., & Weisz, J. R. (1997). ‘‘You can try or you
Do young adolescents and their parents experience the can just give up’’: The impact of perceived control and
same states? Journal of Research on Adolescence, 4, coping style on childhood homesickness. Developmental
567 – 583. Psychology, 33, 508 – 517.
Larson, R. W., Richards, M. H., Sims, B., & Dworkin, J. Wadsworth, M. E., & Compas, B. E. (2002). Coping with family
(2001). How urban African American young adolescents conflict and economic strain: The adolescent perspective.
spend their time: Time budgets for locations, activities, Journal of Research on Adolescence, 12, 243 – 274.

También podría gustarte