This preview shows page 1. Sign up to view the full content.
Unformatted text preview: Child Development, July/August 2002, Volume 73, Number 4, Pages 1151–1165 Continuity, Stability, and Change in Daily Emotional Experience
Reed W. Larson, Giovanni Moneta, Maryse H. Richards, and Suzanne Wilson This longitudinal study examined change in adolescents’ daily range of emotional states between early and
late adolescence. A sample of 220 youth provided reports on their daily emotions at random times during two
1-week periods. At Time 1 they were in the ﬁfth through eighth grades; 4 years later, at Time 2, they were in the
ninth through twelfth grades. Results showed that average emotional states became less positive across early adolescence, but that this downward change in average emotions ceased in grade 10. The results also showed
greatest relative instability between youth in the early adolescent years—correlations over time were lower—with
stability increasing in late adolescence. Lastly, the study found that adolescents’ average emotions had relatively stable relations to life stress and psychological adjustment between early and late adolescence. As a
whole, the ﬁndings suggest that late adolescence is associated with a slowing of the emotional changes of early
Questions of continuity versus change are among the
most basic of developmental life-span psychology
(Block, 1971; Brim & Kagan, 1980; McCrae & Costa,
1990). To what extent is a person’s behavior, personality, and experience stable versus discontinuous over
time? Adolescence is often described as a period of
ﬂux: a time when previously well-adjusted and happy
youth can become distressed (Freud, 1946; Gurian,
1996; Pipher, 1994), as well as a time for “second
chances” and opportunities to reinvent oneself (Csikszentmihalyi & Larson, 1984; Erikson, 1968). Research
suggests that discontinuity may be greatest in early
adolescence, a time when a pile-up of normative life
changes and events is disruptive for many youth
(Petersen, Kennedy, & Sullivan, 1991; Simmons,
Burgeson, Carlton-Ford, & Blyth, 1987), but research
is unresolved as to whether late adolescence is a time
of greater stability.
This study examined questions of constancy versus change in adolescents’ emotional experience—in
their daily range of positive and negative affect. Experiencing greater happiness than unhappiness is important both as an element of mental health and, in its
own right, as a fundamental “good” of human existence. High rates of negative emotion are also related
to problem behavior and lower prosocial behavior
(Eisenberg et al., 1996; Rothbart & Bates, 1998). The
present study examined the extent to which the daily
range of adolescents’ positive-to-negative emotion remained constant versus changed across the period of
adolescence, particularly late adolescence. Constancy
was evaluated, ﬁrst, in terms of similarity of group
means across ages. Does average emotional experi- ence become more negative or positive? Following
the nomenclature used by other authors (McCall,
1977; Moss & Susman, 1980), this is referred to in this
article as “continuity.” Second, constancy was evaluated in the relative ranking of individuals (i.e., degree
of correlation) across time, which has been called
“stability”or “relative stability.” Do happy people remain happy or does reordering take place? Third,
constancy was examined across this age period in the
association of emotional experience with other variables, speciﬁcally life stress and adjustment. Does
emotional experience have a consistent relation to
distress and well-being over time?
To investigate these issues, a cross-sequential
study was performed in which in situ data on hourto-hour emotions were gathered at two points in time
across the span from ﬁfth to twelfth grades. At Time 1,
the sample of 220 included students in the ﬁfth
through eighth grades—representing the pre- and
early adolescent age periods. At Time 2, four years
later, these youth were in the ninth through twelfth
grades, the period of middle and late adolescence. At
both Time 1 and 2, data were collected using the Experience Sampling Method (ESM; Csikszentmihalyi
& Larson, 1987), in which participants carried electronic pagers and provided reports on their immediate emotions at random times when signaled by the
pagers. The sample was limited to European American working- and middle-class Chicago suburban
youth; hence, generalizations are limited to that cultural and socioeconomic population frame. This article
© 2002 by the Society for Research in Child Development, Inc.
All rights reserved. 0009-3920/2002/7304-0012 1152 Child Development is a follow-up to prior reports on a larger sample of
ﬁfth- through ninth-grade students who were studied cross-sectionally at Time 1 (Larson & Ham, 1993;
Larson & Lampman-Petraitis, 1989; Larson & Richards, 1994a).
The contribution of this study is to examine age
changes in daily affect into the high school years. Although past studies have employed global questionnaire reports on affect (particularly depressive affect)
to examine trends into high school, this study went
further by examining data on immediate emotions as
reported during daily experience. Global questionnaire reports on affect are subject to distortions due
to people’s poor memory for affective experiences
and response sets, such as impression management
(Diener, Suh, Lucas, & Smith, 1999; Kahneman, 1999;
Thomas & Diener, 1990), thus it was imperative to
verify patterns in emotional experience using immediate reports. The discrepancy between global and
immediate reports was illustrated in a small 2-year
longitudinal study by Freeman, Csikszentmihalyi,
and Larson (1986), in which, on a global report, high
school juniors and seniors almost unanimously perceived themselves to be happier than they were 2
years earlier, although immediate ESM reports collected at both points in time showed no change in average daily happiness.
Early Adolescence as a Time of Disruption
Past and current theory has focused on early adolescence as a time of greatest changes in emotional experience. Psychoanalytic theorists attributed this to
internal psychological changes. Anna Freud (1946,
1958) presented early adolescence as a time when unpredictable increases in libidinal drive related to puberty and new demands of the superego create emotional upheaval.
Recent literature has given more emphasis to the
role of external factors in driving emotional changes.
Petersen, Kennedy, and Sullivan (1991) described
early adolescence as a time of “developmental transitions,” including the change from elementary to middle
or junior high school, changes in peer expectations,
increased life stress, and changes in relationships and
roles within the family and other contexts. These authors also pointed to internal changes in cognitive
abilities and the physical transformations of puberty;
however, they and others (Brooks-Gunn, Graber, &
Paikoff, 1994) have found evidence that the impact of
puberty is due more to changes in how the adolescent
and others react to the new body than to direct biological impacts on the brain. Like Anna Freud, Petersen and colleagues emphasized the unpredictability of
these transitions; they cited Hamburg’s (1974) comparison of the changes of this age period to a lottery.
Adolescents are subject to a set of transformations
that are, in part, outside of their control (Buchanan,
Eccles, & Becker, 1992; Graber & Brooks-Gunn, 1996;
Petersen & Leffert, 1995).
Evaluation of the role of these internal and external factors in changing adolescents’ emotional experience was beyond the scope of the present study
and not possible with the data available. Thus, the
focus was on the more limited goal of assessing
whether such changes occur, and particularly whether
they continued or diminished in middle and late
Continuity versus Discontinuity in Normative
Patterns across Adolescence
The ﬁrst question was whether, as an aggregate,
adolescents show a trend toward more positive or
dysphoric affect across this age period. The trend
across the adult life span is for people to experience
fewer negative emotions as they get older, and most
studies show less frequent intense positive emotions
(Carstensen, Pasupathi, Mayr, & Nesselroade, 2000;
Diener, Sandvik, & Larsen, 1985; Diener et al., 1999;
Lawton, Kleban, Rajagopal, & Dean, 1992; Mroczek &
Kolarz, 1998).Three studies that employed immediate
reports on experience have conﬁrmed that rates of
negative affect and extreme positive affect are lower
among adults than adolescents (Larson, Csikszentmihalyi, & Graef, 1980; Larson & Richards, 1994b; Verma
& Larson, 1999). Carstensen attributes the trend toward diminished negative affect with age to the developing capacity for emotional regulation and greater
priority given to emotional regulation as people get
older (Carstensen & Charles, 1998; Carstensen, Isaacowitz, & Charles, 1999).
Early adolescence appears to be an anomaly to the
general developmental trend toward diminished frequency of negative emotion. Larson and LampmanPetraitis (1989) analyzed a set of 18,000 ESM reports
on emotions provided by a cross-sectional sample of
483 suburban, European American ﬁfth- through
ninth-grade students (the dataset from which the
present study’s Time 1 data were drawn). They found
a signiﬁcant downward trend in average emotional
states for both boys and girls across this age period,
attributable to both decreased frequency of extreme
positive states and increased frequency of negative
states. Larson and Richards (2000) analyzed a set of
8,500 ESM reports from a sample of 253 urban African
American ﬁfth- through eighth-grade students and Larson et al. found the same trends toward less extreme positive
and more negative affect across this period. Other research that used questionnaire and clinical interviews
also found an increase in reported negative affect, especially depression, across the transition into adolescence (Elliott, Huizinga, & Menard, 1989), particularly
for girls (Magnusson, 1988; Petersen, Sargiani, &
Kennedy, 1991; Rutter, 1980, 1986).
Research on whether this downward trend in emotions continues or reverses in middle and late adolescence is less complete and conclusive. The largest factor associated with increased dysphoria in early
adolescence is stressful life events (Brooks-Gunn &
Warren, 1989; Larson & Ham, 1993; Simmons & Blyth,
1987), and research indicates that the number of
stressful life events levels out or even declines in later
adolescence (Brooks-Gunn, 1991; Compas, Davis, &
Forsythe, 1985; Ge, Lorenz, Conger, Elder, & Simmons, 1994; Graber & Brooks-Gunn, 1996). Perhaps as
a result of this trend for stress, existent research suggests that the rise in negative affect in early adolescence levels out and may even begin to reverse at the
end of adolescence. The one prior ESM study that
spanned early and late adolescence found that average reports of happiness (versus sadness) declined
across early adolescence but then leveled out in the
high school years, with no gender differences (Moneta, Schneider, & Csikszentmihalyi, 2001). This
study, however, relied on a single-item measure of affect and, although longitudinal, only 13% of the sample provided complete data, thus permitting large but
unknown effects of self-selection. Using a questionnaire measure, a recent longitudinal study of Norwegian youth found that girls’ depressive affect increased from age 13 to 15, leveled out until age 18,
and then declined at age 19; whereas boys’ depressive
mood was stable across this period until age 18, and
then declined at age 19 (Holsen, Kraft, & Vittersø,
2000). However, a longitudinal questionnaire study
of New Jersey youth found this inverted-U pattern in
depressed mood to occur for both boys and girls
(Chen, Mechanic, & Hansell, 1998).
The current study employed longitudinal ESM
data to evaluate the trajectory of adolescents’ average
daily, immediate emotional states into and through
the high school years. The Time 1 data for the current
study was a subset of the ﬁfth- through eighth-grade
students used in the report by Larson and LampmanPetraitis (1989); hence, this study examined whether
the cross-sectional downward trend in emotional experience across that age span continued or leveled
out into the ninth- through twelfth-grade period. Also
examined was whether these age trends differed for
boys versus girls. 1153 Relative Stability versus Instability
in Daily Emotional States
This study’s second question was whether individuals change relative to each other. Irrespective of the
group age trends evaluated under the ﬁrst question,
are there shifts among youth in their average level of
affect, or are differences among people stable? Do
speciﬁc individuals maintain the same relative ranking in how happy-to-unhappy their daily lives are?
Discussions of early adolescence have emphasized
that individuals may differ greatly in their experience
of this transitional period. Coleman (1974) theorized
that some adolescents experience the life changes of
this period as spaced out over time, allowing them to
cope and adjust to each one without experiencing major disruption. Other youth, however, experience multiple changes at once, which may overtax their ability
to cope, and have more negative effects on their wellbeing. Simmons and colleagues (1987) obtained support for this theory, showing that young adolescents
who experience a “pile up” of life changes show negative effects on their self-esteem, school grades, and
participation in extracurricular activities. These differences between individuals could be expected to
create instability in emotional experience (i.e., changes
with age in the relative rankings among youth in levels of happiness to unhappiness).
On the other hand, to the extent that daily emotions are a product of underlying temperament, attributable to genes or early experience, greater stability would be expected as adolescents get older.
Questionnaire research establishes a clear role of genetic and early experience in shaping emotional experience (Chorpita & Barlow, 1998; Diener et al., 1999).
Stability in adolescents’ daily environments—for example, in their family environments—might also
contribute to stability in emotional experience (Diener
et al., 1999; Kozma, Stone, & Stones, 2000). In two
studies in the United States and India in which young
adolescents and their parents provided ESM reports,
correlations for average affect between young adolescents and their mothers were r ϭ .53 (United States)
and .43 (India), and between young adolescents and
their fathers were r ϭ .46 and .37, respectively (Larson
& Richards, 1994b; Verma & Larson, 1999). Although
there is no way to parse out the contribution of genetic and environmental factors to these correlations, they do suggest the role of either or both in adolescents’ daily affective experience. It should also
be noted, however, that gene expression can be
timed to occur at any point in development or can
be triggered at any point in the life span as a result
of environmental factors (Brown, 1999). Thus, both 1154 Child Development genes and environments could contribute to instability at any point.
Prior questionnaire research that focused on other
variables provides some evidence that early adolescence may be a life stage of greater relative instability
(Chess & Thomas, 1984; Eccles et al., 1989; Kagan &
Moss, 1962; Schaefer & Bayley, 1963; Simmons &
Blyth, 1987). Little research is available, however, that
indicates whether stability in affect increases in later
adolescence. Holsen et al. (2000) found cross-time
correlations in depressed mood scores to rise slightly
for intervals across older age periods, but reported no
tests of whether this rise was statistically signiﬁcant.
The current study permits assessment of the stability of emotions across the period from ﬁfth to twelfth
grade using hour-to-hour data on emotions. Because
it was cross-sectional, the earlier data did not permit
analyses of within-person trends over time. The new
longitudinal data allow for these analyses, including
comparison of cohorts that were making transitions
across different age spans. A limitation of the current
study is that the 4-year gap between Time 1 and 2 restrained the ability to evaluate narrow time periods.
Nonetheless, instability was expected to be greater
for youth traversing earlier age spans.
An important related question is, how many youth
show negative versus positive changes across the adolescent period? A number of authors have shown
that extreme emotional turmoil is experienced by
only a minority of adolescents (Offer, Ostrov, &
Howard, 1981; Rutter, 1980; Rutter & Rutter, 1993). It
is possible that age changes in the frequency of happy
and dysphoric affect for the sample as a whole could
be due to a subgroup of youth. In the current study,
longitudinal data allowed for the ability to evaluate
what proportion of youth experienced increased negative affect with age.
Stability in the Correlates of Daily Emotions
The third question of the present study was
whether daily emotional experience has stable correlations with life stress and adjustment across this 8year age span. Although the design of this study did
not permit evaluation of causal relations, examination of age differences in correlations could be used to
test whether the concurrent correlations between affect and these other variables differed with age.
Prior research suggests that the negative emotion
of early adolescence is partly a result of life stress
(Brooks-Gunn & Warren, 1989; Merikangas & Angst,
1995; Simmons & Blyth, 1987; Simmons et al., 1987).
Yet, using the Time 1 data from this study, Larson and
Ham (1993) found that the correlation between stress and negative affect emerged between the ﬁfth and
sixth and seventh and ninth grade. For the younger
age period, life stress was not signiﬁcantly related to
negative affect. To explain this ﬁnding, they proposed
that developmental changes across this age span,
such as increased cognitive awareness of the implications of negative events, may make older youth more
sensitive and emotionally vulnerable to negative life
events. The question, then, is does this apparent increased vulnerability continue to increase, stabilize,
or diminish in later adolescence?
Negative affect is also conceptualized as a cause
and a symptom of mental health problems, and it is
important to ask whether this linkage to mental
health may change across the adolescent years. Using
the Time 1 data, Larson, Raffaelli, Richards, Ham, and
Jewell (1990) found that emotional states had a weaker
relation to other depressive symptoms in the preadolescent than the early adolescent age period. One explanation that they suggested was that negative affect
was a less cardinal feature of depression prior to adolescence, an explanation supported by some research
(Ushakov & Girich, 1972). This raises the question of
whether the relation of negative affect with depression and behavioral indicators of psychological maladjustment continues to increase, or stabilizes beyond early adolescence. Another possibility is that
depressed preadolescents may be less able to label
their negative affect (Kovacs, 1986) and thus report it
less often and less reliably, a phenomenon that would
reduce correlations for this age group.
The analyses performed in the present study employed longitudinal data from a subsample of the
youth examined in Larson and Ham (1993) and Larson et al. (1990); hence, the new contribution of this
study is to evaluate whether the strength of the correlations found for the early adolescent period continued to rise or stabilized as youth moved into the high
school years. METHOD
Data were obtained from 220 working- and middleclass European American youth from four suburban
Chicago neighborhoods. At Time 1, these youth were
almost evenly distributed across the ﬁfth through
eighth grades (ages 10–14). At Time 2, which occurred
approximately 4 years later, they were in the ninth
through twelfth grades (ages 13–18). In two of the
neighborhoods, students were in small kindergarten
through eighth-grade schools at Time 1 and a large 4year high school at Time 2. In the two other neighbor- Larson et al. hoods, students were spread across small elementary
and middle schools at Time 1 (one school had a ﬁfth–
sixth grade transition, the other a sixth–seventh grade
transition), and were in a large 4-year high school at
Time 2. Thus, all students went through the transition
to a large high school between Time 1 and Time 2.
This longitudinal sample of 220 represented a subsample of the ﬁfth through eighth grade students
who were studied at Time 1. The original sample was
selected randomly with stratiﬁcation by time of year,
grade, gender, and neighborhood. Data collection at
Time 1 was conducted in eight “waves” so that different students participated at different seasons of the
year (see the Appendix). To reduce researcher labor,
these eight waves were collapsed at Time 2 into three
“rounds” of data collection. These three rounds were
timed so that students would participate in approximately the same season at Time 2 as they did at Time
1, and so that the interval between Time 1 and 2
would be as close as possible to 4 years. The interval
between Time 1 and 2 varied from 42 to 56 months.
Two groups of students (those in Wave 1 and 2 at
Time 2) were in a school grade that was 5 years
greater than their grade at Time 1 (for all others it was
4 years, except for 4 students, including 1 in the 5-year
group, who had been held back a grade). These students did not differ substantially or signiﬁcantly from
other students on any of the measures used in this
study. The collapsing of groups at Time 2 also meant
that some students provided data at different times of
the year at Time 1 and 2; however, analysis showed
that scores for the central variable—affect—did not
vary signiﬁcantly by time of year. It should be also
noted that students who were beyond twelfth grade at
Time 2 (i.e., students who were in the eighth grade
and in Waves 1 and 2 at Time 1, and a subsample of 73
ninth-grade students who were studied at Time 1)
were not followed at Time 2 and thus were not included in the sample examined in the present study.
The 220 students represented 50.2% of the randomly selected youth who were invited to participate
and would have been eligible to be in the longitudinal
sample. From an initial pool of 438 invitees, 328 obtained parental consent and participated at Time 1.
Among these, 67.1% participated at Time 2. Sixtyseven had moved and were unreachable at Time 2, 3
had died, and 38 participated but provided data at
either Time 1 or 2 that did not meet minimum quality
criteria. Nonparticipation at Time 2 was not related to
parents’ education level, employment, or the SES ranking of their jobs, nor to adolescents’ grade at Time 1.
Nonparticipation at Time 2 was somewhat higher
among boys, 2(1, N ϭ 328) ϭ 8.60, p ϭ .002; thus,
there were more girls (N ϭ 123) than boys (N ϭ 97) in 1155 the ﬁnal sample. Nonparticipation was also somewhat higher among students who at Time 1 reported
lower average affect on the ESM, t(322) ϭ 2.05, p ϭ .041,
reported more negative life events, t(324) ϭ 2.18, p ϭ
.030, had higher depression scores, t(271) ϭ 2.29,
p ϭ .032, had lower self-esteem, t(327) ϭ 2.43, p ϭ .016,
had lower school grade point averages (GPAs), t(321) ϭ
4.28, p Ͻ .001, and were rated by their parents as having more behavioral problems, t(320) ϭ 2.05, p ϭ .041
(the speciﬁc measures used in these analyses are identiﬁed below). Therefore, attrition was greater among
youth who were more distressed and had higher
problem behavior at Time 1. However, it should be
noted that the majority of distressed youth did participate at Time 2. The participation rate among youth
whose average affect was in the lowest third of the
Time 1 sample was 62.0% (as compared with 69.6%
for the remaining youth). It should also be emphasized that although the differences just reported were
signiﬁcant, they were not of great magnitude and not
likely to have had major effects on the ﬁndings. The
students who did not participate at Time 2 had somewhat lower average Time 1 affect (M ϭ .98 versus
1.16), but their average was well within the range of
positive affect, and on a scale that spanned from Ϫ3 to
ϩ3, this difference was not very great.
Participants provided data on their emotional
states for 1 week at Time 1 and Time 2, via the ESM
(Csikszentmihalyi & Larson, 1987). For each 1-week
period, they carried electronic pagers and self-report
booklets and were instructed to ﬁll out one self-report
form each time they were signaled. One signal was
sent at a random time within each 2-hour block of
time. The self-report asked them to provide information about the situation they were in at the time of the
signal and to rate their emotional state, as well as
other dimensions of their experience on scaled items.
At Time 1, signals occurred between 7:30 am and 9:30
pm for the entire week. To accommodate the later bedtimes of older youth, at Time 2 signals occurred between 7:30 am and 10:30 pm on weekdays and 8:00 am
and 12:00 am on weekends. The minimum criteria
for including students in the ﬁnal sample was response to at least 15 signals at both Time 1 and 2, and
response to at least 50% of the signals between their
last report at each time (these criteria excluded students
who provided reports sporadically over the week, but
included a few students who did a good job of responding for 3 or 4 contiguous days and then stopped).
Self-reports were obtained in response to most signals. At Time 1, participants provided reports for a 1156 Child Development mean of 85% of the signals. At Time 2, they provided
reports for a mean of 76% of the signals. Estimates
based on a subset of youth indicated that about 6% of
signals were missed due to mechanical failure of the
pager. Students reported that other signals were
missed due to forgetting the pager or booklet at home
or in their bedroom or being involved in an activity
that could not be interrupted, such as taking a test.
The students provided an average of 40.2 reports per
person at Time 1 and 34.7 per person at Time 2. The
complete data set included 16,477 ESM reports across
the two data collections.
At both Time 1 and Time 2, participants also completed sets of questionnaires, and data were collected
from the school and from one parent, typically the
Affect. This study employed a three-item measure
of immediate affect that has been used in much prior
ESM research, and has a demonstrated record of validity and reliability (Csikszentmihalyi & Larson,
1987; Larson, 1989). At the moment of each signal,
participants rated their emotional state on 7-point semantic differential items (happy–unhappy, cheerful–
irritable, friendly–angry). Scores for responses to these
three items were averaged to create a scale from Ϫ3 to
ϩ3, where negative scores represent negative affect,
Cronbach’s ␣ ϭ .75. Although responses to this scale
ﬂuctuated from one report to the next, the stability of
underlying central tendency was evident in split-half
correlations. At Time 1, individuals’ mean responses
for the ﬁrst half of the week correlated with their
mean response for the second half, r ϭ .67; for Time 2
this correlation was r ϭ .55. These correlations did not
differ signiﬁcantly by grade or gender. Standard deviations were also stable for individuals at both Time 1,
r ϭ .53, and Time 2, r ϭ .66, with no signiﬁcant difference by grade or gender. Past research shows that adolescents’ average scores on this scale are correlated
with teachers’ ratings of emotional state (Larson &
Ham, 1993), parents’ ratings of internalizing symptoms (Verma & Larson, 1999), depression, and other
conceptually related variables (Larson, 1989).
It should be noted that this scale places positive
and negative affect at opposites ends on a continuous
scale, a positioning that has been challenged by some
researchers who have conceptualized positive and
negative affect as independent dimensions. In a careful factor analytic study, however, Tellegen, Watson,
and Clark (1999) showed that a general happiness-tounhappiness dimension accommodates much of the
variance across different emotions. Stressful life events. On the questionnaire completed
at the end of each week of ESM reporting, students
completed a 51-item checklist of major life events
(Larson & Ham, 1993), based on a measure originally
developed by Coddington (1972). They were asked to
check those events that had happened to them in the
past 6 months. The score for this report was the totaled number of the 33 negative events that each student checked. At Time 1, these counts were found to
be correlated with parents’ reports of the adolescents’
experience for the same set of events, r ϭ .47.
Measures of adjustment. The following measures of
adjustment were obtained at both Time 1 and Time 2.
Depression was measured using the 27-item self-report
Child Depression Inventory (Kovacs, 1986). Selfesteem was measured using the 10-item scale developed by Rosenberg (1965). Students’ current GPA for
academic classes was obtained from the schools. Parents completed the Child Behavior Checklist (Achenbach, 1991), which was used to compute a score for total behavioral problems.
Analytic procedures were selected that were suited
to the two-level, hierarchical structure of the ESM
data. These data included 16,477 reports on moments
in time (Level 1) that were provided by 220 individuals (Level 2). Techniques were chosen that took into
account the variations in emotional states that occurred both across moments and across individuals.
When feasible, multilevel modeling (ML), a regression procedure speciﬁcally designed for data
with this type of hierarchical structure (Goldstein,
1987, 1995; see also Bryk & Raudenbush, 1992; Longford, 1993) has been used. For the present analysis,
this modeling proceeded by ﬁtting separate regressions for each individual to obtain an average regression model that was valid for the entire population from which the individuals were sampled. The
estimation procedure is iterative and, at each iteration, provides improved estimates of both personspeciﬁc and population average regression coefﬁcients until convergence is achieved. Thus, in the
present study it made full use of the information provided by the 16,477 observations, while taking into
account the unique patterns for each person. The ﬁnal
solution of the iterative process provides regression
coefﬁcients for the population, plus their standard
errors—which allows for testing their statistical signiﬁcance. Multilevel modeling is especially suited to
the analysis of ESM data (Larson, Richards, Moneta,
Holmbeck, & Duckett, 1996; Moneta & Csikszentmihalyi, 1996, 1999). Larson et al. Multilevel modeling regression was used to evaluate the question of continuity in adolescents’ average
affect across developmental periods. Based on past
research (Larson & Lampman-Petraitis, 1989), we
chose to use school grade (rather than age) as the index of developmental level. Given the appearance of
gender differences in some past studies, it was essential to evaluate how grade trends might vary by gender. Thus, a multilevel model was tested with the following form:
Affect ϭ ␤0 ϩ ␤1 ϫ Grade (linear) ϩ
␤2 ϫ Grade (quadratic) ϩ ␤3 ϫ Time ϩ
␤4 ϫ Gender ϩ ␤5 ϫ (Grade ϫ Gender),
in which Affect was a Level 1 variable measure that
varied within person across each ESM report; Gender
was a Level 2 variable that remained constant within
person across waves; and Grade and Time were variables that varied within person across the two data
collection periods and thus were both Level 1 and
Level 2 variables. Grade was school grade at the respective data collection period and Time was the indicator for the two data collection periods (0 and 1).
The intercept and the term for Time were deﬁned as
random effects; this allowed each individual to have a
unique intercept and Time 1 to Time 2 slope, and permitted the model to adjust for individual differences
in the scaling of affect. The inclusion of terms for both
Grade and Time allowed for estimation of the mean
grade trend for affect across the whole sample, independent of any test–retest effect. A signiﬁcant Time
effect would indicate that the grade trend in affect differed between the ﬁrst and second administration
due to a test–retest method bias, whereas a nonsignificant effect would rule out such a possibility. The inclusion of Time, whether signiﬁcant or not, made it
possible to estimate the grade trend controlling for a
possible test–retest method bias.
This model was estimated by means of the program ML3 (Prosser, Rasbash, & Goldstein, 1991). The
signiﬁcance of each term in the equation was evaluated by computing the ratio of the point estimate (the
␤s in the equation) to its standard error, then comparing these against the standardized normal distribution. Multilevel modeling also has the capability to
test differences in between-person and within-person
variance, which was employed to evaluate whether
these changed between Time 1 and Time 2.
For analyses that were not suited to this ML
model, other appropriate analytic techniques were
employed, following guidelines suggested by Larson
and Delespaul (1992) for analyzing ESM data. In all
these analyses the within-person variation in responses was controlled by using the person, rather 1157 than the single ESM observation, as the unit of analysis. This was achieved by computing aggregate scores
for each person (e.g., the person’s average affect) at
one point of data collection. In several instances, ML
was used to evaluate these scores. In other instances,
traditional techniques such as paired-samples t tests
and repeated-measures analysis of variance (ANOVA)
were used. For several analyses, the sample was divided into grade cohorts, deﬁned by their school
grade at Time 1. These included those who started the
study in grades 5 (n ϭ 57), 6 (n ϭ 62), 7 (n ϭ 53), and
8 (n ϭ 48). Descriptions of the speciﬁcs of these analyses are provided in the Results section.
Continuity versus Discontinuity in Group Means
The ﬁrst question asked was whether the downward grade trend in average affect that was evident
across the ﬁfth- through eighth-grade period would
continue through the high school years. To test this,
we ﬁrst evaluated a multilevel regression model with
linear and quadratic terms for grade, along with gender and time (Time 1 versus Time 2).
These analyses showed a nonlinear grade trend in
average affect (Table 1). The linear and quadratic
terms for grade were signiﬁcant. The quadratic curve
was concave, with the grade trend toward less positive affect diminishing in the higher grades. As Figure
1 shows, the decline in average affect stopped at the
tenth grade. A separate ML regression that employed
only the Time 2 data found no signiﬁcant grade trend
for the high school period. The nonsigniﬁcance of the
time effect in Table 1 indicates that the grade trend in
Table 1 Final Multilevel Regression Model of Affect on Time,
Gender, and Grade
Mean regression coefficients (␤)
Within-persons variance components
Covariance (intercept, time)
Between-persons variance components
Covariance (intercept, time)
* p Ͻ .05; ** p Ͻ .01; *** p Ͻ .001. Standard
.063 1158 Child Development Figure 1 Grade trend in average affect. affect estimated on the Time 1 data did not differ from
that estimated on the Time 2 data; in other words,
there was no test–retest effect.
It must be noted that despite the downward grade
trend across early and late adolescence, throughout
the entire period covered by the study, the average
score for affect remained consistently on the positive
side, above the neutral value of 0.0. This ﬁnding
raises the issue of whether adolescents experienced
increased negative affect, decreased positive affect, or
both. To clarify this, an ordinary paired-sample t test
was performed on the relative frequencies of positive
and negative affect experienced at Time 1 and 2. For
each participant, the percentage of times the composite affect score was greater than 0.0 (positive affect)
and less than 0.0 (negative affect) were computed.
The average percentage of positive affect declined
from 73.9% at Time 1 to 70.7% at Time 2, t(219) ϭ 1.93,
p ϭ .056, whereas the average percentage of negative
affect increased from 12.7% to 19.6%, t(219) ϭ Ϫ6.20,
p Ͻ .001. Thus, the decline in average affect was attributable to decreased frequency of positive affect
and, to a larger extent, increased frequency of negative affect.
Further permutations of the basic ML regression
tested for interactions among the independent variables; none were found. As shown in Table 1, there
was a main effect for gender, with girls reporting
more positive average affect. However, when terms
were added to the equation for the interaction between gender and both the linear and quadratic grade
trends, neither was found to be signiﬁcant. This
meant that the concave grade trends for girls and
boys were parallel, as shown in Figure 2. The evaluation of terms for the interaction of time and the linear
and quadratic terms for grade showed that they did
not account for signiﬁcant additional variance. Figure 2 Estimated time trends of mean affect across grades
for girls and boys. The ML analyses, however, did indicate changes in
variance in affect between Time 1 and Time 2. The estimated within-person variance components, shown
in Table 1, indicated that the within-person variance
signiﬁcantly increased between Time 1 and 2, rising
from 1.154 to 1.584, and the between-person variance
signiﬁcantly diminished between Time 1 and Time 2,
contracting from .500 to .371. This indicates that individuals reported a wider range of positive and negative emotions at Time 2 and that there was less difference among individuals in their average affect at
Time 2. To more closely examine the within-person
change, an ordinary repeated-measures ANOVA was
conducted that used the person as the unit of analysis
and had each person’s standard deviation as the dependent variable, with time as the within-person factor and gender as the between-person factor. The
interaction between time and gender was highly signiﬁcant, F(1, 218) ϭ 11.4, p Ͻ .001, as was the effect for
gender, F(1, 218) ϭ 25.1, p Ͻ .001. A plot of the estimated marginal means indicated that the age change
in within-person variance was almost entirely attributable to girls. Furthermore, when separate pairedsample t tests were performed, boys’ mean standard
deviation in affect showed no signiﬁcant increase between Time 1 and Time 2, whereas girls’ mean standard deviation signiﬁcantly increased from 1.01 to 1.32.
In contrast to this pattern for within-person variance,
follow-up tests of the change from Time 1 to Time 2 in
between-person variance did not suggest that this
change differed markedly by gender.
In conclusion, these ﬁndings indicate that the
downward trend in affect found across the transition
into early adolescence did not continue into late adolescence. Rather, late adolescence appeared to be associated with little change in the average level of positiveto-negative affect, with girls showing an increase in
the variance of affect in the high school years. Larson et al. 1159 Stability versus Instability in Comparative Ranking
The second question of the present study was how
stable individuals were in comparison with each
other over the 4-year period between Time 1 and Time
2. To evaluate this question, the person was used as
the unit of analysis. Mean affect scores were computed separately for each person for Time 1 and Time
2, and then the correlations between these values
were calculated. Findings showed substantial instability. For the entire sample, the correlation between
Time 1 and Time 2 was modest, r(210) ϭ .35, p Ͻ .001,
indicating that for many individuals, mean affect
changed relative to other individuals over the 4 years.
Further analyses indicated differences in stability
between cohorts. We had predicted that the cohorts
who were younger at Time 1—and for whom the 4year span included the ﬁfth and sixth grade—would
show lesser stability. To test this hypothesis, regressions were computed that predicted Time 2 affect
from Time 1 variables. The independent variables included Time 1 mean affect, Time 1 grade, gender, the
interaction of Time 1 affect and Time 1 grade, and the interaction of Time 1 affect and gender.
This regression showed differences in stability
across the four grade cohorts. As in the prior analyses,
Time 1 affect and gender were signiﬁcant predictors
of Time 2 affect. The interaction term for gender and
for Gender ϫ Grade were not signiﬁcant predictors of
Time 2 affect. The important ﬁnding was that the term
for the interaction of Time 1 affect and Time 1 grade
(cohort) was a signiﬁcant predictor of Time 2 affect, ␤ ϭ
.112, SE ϭ .038, p ϭ .027. To interpret these differences
between cohorts, separate regression lines were computed for each (see Figure 3). As is apparent, the
slopes of these regressions lines are progressively
steeper for each older cohort. This indicates that there
was greater stability for the older grade cohorts.
The same procedures were used to examine stability in the within-person variability in affect between
Time 1 and 2. Standard deviations were computed for
each individual’s reports at Time 1 and reports at
Time 2. The correlations between these scores across
the 4-year period was small, r ϭ .20. In regression
analyses, the strength of this association did not vary
as a function of cohort or gender.
The ﬁndings from these and the previous analyses
led to the question of how many youth showed downward change in their average emotional state between Time 1 and 2. Findings from the ﬁrst set of analyses (Figure 1) indicated that emotional states were, on
average, less positive at Time 2. The question, however, was whether this was attributable to most youth
or just to a minority who experienced a large down- Figure 3 Regression lines for predicting affect at Time 1 from
affect at Time 2, by grade cohort. ward change. When the entire sample of 220 students
was examined, it was found that 63%—approximately
ﬁve eighths—reported less positive average affect at
Time 2 than at Time 1, and the remaining 37% reported more positive affect (no student had exactly
the same mean affect at both times). This rate of 63%
was identical for boys and girls and did not differ signiﬁcantly between grade cohorts. To estimate how
many youth showed a major change, each student’s
Time 2 mean was evaluated relative to his or her Time
1 mean and standard deviation. Thirty-four percent
reported a mean at Time 2 that was below their Time
1 mean by more than .5 standard deviation units, and
16% reported a Time 2 mean that was .5 standard deviation units above their Time 1 mean. Thus, one third
of students showed a major downward change in
their average emotional state, and one sixth showed a
major upward change.
Stability in the Correlates of Daily Emotions
The ﬁnal question asked in the current study was
whether the correlations between mean affect and
measures of life stress and adjustment changed between early and late adolescence. To test this question, the individual was used as the unit of analysis. Table 2 displays the correlations between scores for
mean affect at Time 1 and 2, with concurrent measures of stressful events and adjustment.
The ﬁndings showed that affect was associated
with life stress at both age periods, but the strength of
this relation did not change between periods. To test
whether this relation differed between age periods,
an ML regression was performed in which the composite affect scores for Time 1 and Time 2 were repeated measures within individuals. The dependent 1160 Child Development Table 2 Concurrent Correlations of Average Affect with Stress
Events and Adjustment
Stressful life events
Grade point average
Behavior problems (Child Behavior Checklist) Time 2 Ϫ.28***
Ϫ.14ϩ * p Ͻ .05; *** p Ͻ .001.; ϩ p Ͻ .10 variable in this regression was average affect at the respective time of measurement. The independent variables were time, stress, and the interaction term for
Time ϫ Stress. The Time ϫ Stress term was the test
of the age difference: Was the strength of the association different between Time 1 and Time 2? The term
for this interaction was not signiﬁcant, which indicated an absence of difference. To summarize the relation between life stress and daily affective experience appeared to be consistent across early and late
The relations between affect and the adjustment
variables were also found to be stable across early and
late adolescence (Table 2). As with the analyses of life
stress, an ML regression was tested in which the individual was the unit of analysis, with time as a repeated within-person variable. For all dependent
variables the interaction term for time was not significant, indicating stability in the strength of association
between affect and these variables across this age
span. There were also no signiﬁcant two-way interactions with gender, or three-way interactions with gender and time for any of these dependent variables.
The ﬁndings of this study indicate that change in
young people’s daily range of emotions slows between early and late adolescence. Past research has
shown that early adolescence is a time when the average youth experiences a downward shift in this
range—in the direction of more negative and fewer
extreme positive states. It also has shown that early
adolescence is a period of low stability among youth
in their range of emotions: there is ﬂux in which the relative happiness of different individuals shifts. The
ﬁndings of this longitudinal research suggest that this
change and ﬂux slows in late adolescence. The downward shift in average states was not found to continue
into late adolescence and there was greater stability
among youth in their relative levels of happiness.
Many of the present study’s results reinforce ﬁndings from prior studies, but they also go beyond prior
studies in demonstrating these age trends with intensive data on emotional experience obtained during
random moments in adolescents’ daily lives.
It should be emphasized that these results came
from a sample of European American suburban
youth from the United States, and may not generalize
to youth living in differing life situations or cultural
worlds. Early adolescence may not be a peak time of
change among other groups, or this slowing of change
in late adolescence may not occur, or may occur earlier or later. It should also be noted that the longitudinal sample used in this study had a higher rate of
attrition among more distressed youth. Although
most youth with lower average affect, as well as less
positive scores on measures of stress and adjustment,
remained in the sample, these youth were somewhat
underrepresented. This underrepresentation is likely
to have created slightly more positive values for average affect in the sample and to have slightly reduced
the strength of correlations (as a result of more restricted ranges) than would have been expected for a
perfectly representative sample.
Before discussing the differences that were found
between early and late adolescence, it is useful to note
the similarities, as revealed in the third set of analyses. This set found stable relations across these two
periods—ﬁrst, between adolescents’ average daily
emotions and their experience of stressful events.
Stressful events in past longitudinal research appear
to have a causal relation to negative daily emotion
(Merikangas & Angst, 1995). The stability of the
present study’s correlation for stressful events suggests that the contribution of stress to negative emotion does not change markedly from early to late adolescence, and that adolescents’ degree of emotional
sensitivity or vulnerability to stressful events remains
the same. Second, stable relations were found across
the two periods between average emotional experience and adjustment variables such as self-esteem
and problem behaviors. Adjustment is generally conceptualized as being affected by and affecting emotional experience. The ﬁnding of cross-age stability in
the strengths of correlations between adjustment and
emotions suggests another constancy between early
and late adolescence in the underlying causes and
conceptual signiﬁcance of daily emotional experience. These underlying constancies provide a backdrop to the age changes found in the ﬁrst and second
sets of analyses.
The ﬁrst set of ﬁndings concerned age changes in
the average emotions experienced across all youth,
a dimension of change referred to as continuity/
discontinuity. The downward shift in average emo- Larson et al. tions in early adolescence—toward more dysphoric
and less extreme positive affect—was found to level
out in late adolescence. It should be emphasized that
the average emotional state after this bottoming out
was still in the positive range: .9 on a scale from ϩ3 to
Ϫ3. Negative affect was reported for only one ﬁfth of
random moments sampled during the high school
age period. Nonetheless, late adolescents appeared to
experience a higher rate of negative affect and an average level of happiness that was somewhat less positive than that experienced in the happier days before
adolescence began. The good news was that the decline in this average level of happiness stopped at
around grade 10, with no further downward change.
This leveling off of downward change in late adolescence is consistent with and reinforces the ﬁndings
of questionnaire studies. The one point of difference
is that some questionnaire research has found the
early adolescent age increase in depressive affect to
occur only for girls (Holsen et al., 2000; Petersen, Sargiani, & Kennedy, 1991), whereas the current research
found the downward trend to occur for both boys and
girls. This study also found girls’ affect to be more
positive than boys’ affect across this age period (see
Figure 2). This discrepancy in ﬁndings may be partly
attributable to differences in what was being measured.
The present study measured only the experience of
immediate affect, whereas questionnaire studies often
measure a wide range of symptoms associated with
clinical depression, such as sleep disturbance. Possibly the discrepancy in ﬁndings was because girls, but
not boys, show age increases in these other symptoms.
Indeed, it is widely accepted that girls, but not boys,
show an increase in clinical depression during this
age period (Angold & Rutter, 1992; Petersen et al.,
1993). A second related explanation is suggested by
the ﬁnding that the within-person variability in girls’
(but not boys’) emotions increased into late adolescence. With age, girls reported wider daily variations
on the scale of negative-to-positive emotions. As a result, a study that focuses only on rates of depressive
affect (as many questionnaire studies do) would
show an exaggerated age trend for girls, because it
would show only one tail of a widening emotional
distribution. Yet for most girls, the age increase in
negative affect appears to be partly counterbalanced
by their frequent experience of positive affect.
Looking across the life span, early adolescence for
both genders is an exception in the long-term trend
toward reduced negative emotion. As youth enter
and move through adolescence they experience
greater negative emotion, possibly as a result of the
increased stressful events of this transitional period
and their heightened sensitivity to this stress due to 1161 cognitive change (Larson & Asmussen, 1991). The
present study found that late adolescents’ average
daily emotions were not higher than those of younger
adolescents—suggesting that they were still affected
by stress. The age trend toward increased negative affect stopped, but it did not reverse. However, questionnaire research suggests that negative affect begins
to diminish after age 18, in the period immediately
after this study’s coverage (Chen et al., 1998; Holsen
et al., 2000). Studies that compared adults and adolescents indicate that intense positive affect also diminishes in the years after adolescence (Diener et al.,
1999; Larson & Richards, 1994b), so that the longer
term trend is toward less variable daily emotions
around a mildly positive baseline. Thus, middle and
late adolescence is a low point—a nadir—in emotional experience.
The second set of ﬁndings was congruent with the
ﬁrst in suggesting that change in affective experience
slows down between early and late adolescence. The
second set of ﬁndings dealt with changes in the relative ranking between individuals—what has been
called relative stability. Stability is evidenced when
there is a correlation in scores over a time interval: a
higher correlation indicates stability. The present
study found that these correlations differed for cohorts in the sample who made a transition across different age periods. The youth who went from ﬁfth to
ninth grades across the study showed less correlation
than those who went from eighth to twelfth. This suggests that the early years, grades 5 through 7, are associated with the largest instability. A shortcoming of
this study is that the 4-year span between Time 1 and
2 and the fact that all cohorts traversed the transition
into high school limited the ability to isolate speciﬁc
periods. Research that examines shorter time spans
would better isolate transitions of greatest instability.
Nonetheless, the present ﬁndings clearly suggest
that stability in emotional experience increases with
age. One explanation for this increased stability is
methodological (i.e., with age, adolescents become
better reporters of their emotional states), resulting in
more stable scores. It is conceivable that cognitive
changes in early adolescence alter response style or
sensitivity to emotions, creating changes in emotional
experience that are more apparent than real. Using
Time 1 data from the same participants, Larson and
Lampman-Petraitis (1989) partly addressed this possibility. They showed that when asked to rate the
emotions depicted in a series of drawings of faces,
there was no age difference between ﬁfth to ninth
grade in how these emotions were rated. This suggests no change in response tendency. Holsen et al.
(2000) also argued that measurement is stable across 1162 Child Development this age period, based on their ﬁnding that the internal reliability of their scale of depressed affect did not
change with age. For these reasons, we think that the
increased stability in adolescents’ daily emotions is
not an artifact of measurement, but rather a real
The more likely explanation for the increased stability in later adolescence is that this is a more stable
period, and there are fewer changes in the conditions
that affect youths’ daily lives. In interpreting similar
data for changes in self-esteem, Alsaker and Olweus
(1992) theorized that later adolescence is associated
with the gradual consolidation of unspeciﬁed underlying “structures.” However, it is not possible to determine from these data whether the increased stability
was due to consolidation of endogenous or exogenous structures, or both. It is possible that the slowing
of change is due to slowing of internal psychological
or physiological changes, or to greater constancy in
gene expression. It is also possible that higher relative
stability in late adolescence is due to reduced change
in adolescents’ daily environments, or in how they
experience these daily environments. There is much
to be learned about the factors underlying change and
stability across this age period.
Thus far, the issues of the present study’s ﬁrst and
second sets of analyses have been separated; however, they need to be thought of together. An analysis
that combined the two issues found that many, but by
no means all, youth showed a downward change in
their average daily emotions from the ﬁfth- through
eighth-grade period to the ninth- through twelfthgrade period. Five eighths of the students (63%) reported lower average affect in the later age period. By pure chance, 50% would have been expected to report
lower affect at Time 2; thus, the striking ﬁnding may
be that 37% reported higher average affect at this
time. Perhaps the more meaningful ﬁgures are that
one third of the youth showed a downward change
that was greater than half of their Time 1 standard deviation, and one sixth showed an upward change of
more than half a standard deviation. Of course, any
individual’s scores are likely to have been affected by
week-to-week and month-to-month variations at both
times. The important point, however, is that the trend
was not universally downward; there were many
youth who did not show this trend. Future research
needs to continue to consider differing individual trajectories as a topic of study.
For those individuals who experience major changes
from early to late adolescence, this may be a signiﬁcant turning point. The present ﬁndings showed that
as adolescents get older, their average emotional level
becomes more stable relative to other teens. This suggests that individuals’ baseline emotional states may
be less easily changed after they pass early adolescence. Thus, future efforts to understand what inﬂuences stability and change in adolescents’ emotions
need to focus on what factors account for the largest
downturns in early adolescence and what can be
done to avert them, as well as look for what factors
promote lasting upturns in baseline emotional state.
This research was supported by National Institute of
Mental Health grant R01 MH53846 awarded to
Maryse H. Richards. APPENDIX
8b Time 2
Date May 1985
October, November 1985
April, May 1986
March 1987 Round
3b Date November 1989
Not studied at Time 2
September 1990 Months
from Time 1
to Time 2 Number of
16 Note: The Time 1 study consisted of eight waves of students, each consisting of approximately equal numbers of boys and girls from two
separate schools (one working class and one middle class). These eight waves were studied in three “rounds” of participation at Time 2. Larson et al. ADDRESSES AND AFFILIATIONS
Corresponding author: Reed W. Larson, Department
of Human and Community Development, University
of Illinois, 1105 W. Nevada Street, Urbana, IL 61801;
e-mail: firstname.lastname@example.org. Giovanni Moneta is at Harvard Business School, Boston, MA; Maryse H. Richards is at Loyola University of Chicago, Chicago, IL;
and Suzanne Wilson is also at the University of Illinois. REFERENCES
Achenbach, T. M. (1991). Manual for the Child Behavior Checklist/4–18. Burlington: University of Vermont, Department of Psychiatry.
Alsaker, F. D., & Olweus, D. (1992). Stability of global selfevaluations in early adolescence: A cohort longitudinal
study. Journal of Research on Adolescence, 2, 123–145.
Angold, A., & Rutter, M. (1992). Effects of age and pubertal
status on depression in a large clinical sample. Development and Psychopathology, 4, 5–28.
Block, J. (1971). Lives through time. Berkeley, CA: Bancroft
Brim, O. G., Jr., & Kagan, J. (1980). Constancy and change: A
view of the issues. In O. Brim, Jr. & J. Kagan (Eds.), Constancy and change in human development (pp. 1–25). Cambridge, MA: Harvard University Press.
Brooks-Gunn, J. (1991). How stressful is the transition to adolescence in girls? In M. E. Colten & S. Gore (Eds.), Adolescent stress: Causes and consequences (pp. 131–149). Hawthorne, NY: Aldine de Gruyter.
Brooks-Gunn, J., Graber, J. A., & Paikoff, R. L. (1994). Studying links between hormones and negative affect: Models
and measures. Journal of Research on Adolescence, 4, 469–
Brooks-Gunn, J., & Warren, M. P. (1989). Biological and social contributions to negative affect in young adolescent
girls. Child Development, 60, 40–55.
Brown, B. (1999). Optimizing expression of the common human genome for child development. Current Directions
in Psychological Science, 8, 37–41.
Bryk, A., & Raudenbush, S. W. (1992). Hierarchical linear
models: Applications and data analysis methods. Newbury
Park, CA: Sage.
Buchanan, C. M., Eccles, J. S., & Becker, J. B. (1992). Are adolescents the victims of raging hormones: Evidence for
activational effects of hormones on moods and behavior
at adolescence. Psychological Bulletin, 111, 62–107.
Carstensen, L., & Charles, S. T. (1998). Emotion in the second half of life. Current Directions in Psychological Science,
Carstensen, L., Isaacowitz, D., & Charles, S. (1999). Taking
time seriously: A theory of socioemotional selectivity.
American Psychologist, 54, 165–181.
Carstensen, L., Pasupathi, M., Mayr, U., & Nesselroade, J.
(2000). Emotional experience in everyday life across the
adult life span. Journal of Personality and Social Psychology,
79, 644–655. 1163 Chen, H., Mechanic, D., & Hansell, S. (1998). A longitudinal
study of self-awareness and depressed mood in adolescence. Journal of Youth and Adolescence, 27, 719–734.
Chess, S., & Thomas, A. (1984). Origins and evolution of behavior disorders from infancy to early adult life. New York:
Chorpita, B. F., & Barlow, D. H. (1998). The development of
anxiety: The role of control in the early environment.
Psychological Bulletin, 124, 3–21.
Coddington, R. D. (1972). The signiﬁcance of life events as
etiologic factors in the diseases of children: II. A study of
a normal population. Journal of Psychosomatic Research,
Coleman, J. C. (1974). Relationships in adolescence. London:
Compas, B. E., Davis, G. E., & Forsythe, C. J. (1985). Characteristics of life events during adolescence. American Journal of Community Psychology, 13, 677–691.
Csikszentmihalyi, M., & Larson, R. (1984). Being adolescent.
New York: Basic Books.
Csikszentmihalyi, M., & Larson, R. (1987). The Experience
Sampling Method. Journal of Nervous and Mental Disease,
Diener, E., Sandvik, E., & Larsen, R. J. (1985). Age and sex effects for emotional intensity. Developmental Psychology,
Diener, E., Suh, E. M., Lucas, R. E., & Smith, H. L. (1999).
Subjective well-being: Three decades of progress. Psychological Bulletin, 125, 276–302.
Eccles, J. S., Wigﬁeld, A., Flanagan, C. A., Miller, C., Reuman,
D. A., & Yee, D. (1989). Self concepts, domain values, and
self-esteem: Relations and changes at early adolescence.
Journal of Personality, 57, 283–310.
Eisenberg, N., Fabes, R. A., Karbon, M., Murphy, B. C.,
Wosinski, M., Polazzi, L., Carlo, G., & Juhnke, C. (1996).
The relations of children’s dispositional prosocial behavior to emotionality, regulation, and social functioning.
Child Development, 67, 974–992.
Elliott, D., Huizinga, D., & Menard, S. (1989). Multiple problem youth: Delinquency, substance abuse and mental health
problems. New York: Springer-Verlag.
Erikson, E. (1968). Identity youth and crisis. New York: W.W.
Freeman, M., Csikszentmihalyi, M., & Larson, R. (1986).
Adolescence and its recollection: Towards an interpretive model of development. Merrill-Palmer Quarterly, 32,
Freud, A. (1946). The ego and the mechanisms of defense. New
York: International Universities Press.
Freud, A. (1958). Adolescence. Psychoanalytic Study of the
Child, 15, 255–278.
Ge, X., Lorenz, F. O., Conger, R. D., Elder, G. O., Jr., & Simmons, R. L. (1994). Trajectories of stressful life events and
depressive symptoms during adolescence. Developmental Psychology, 30, 467–483.
Goldstein, H. (1987). Multilevel models in educational and social research. London: Grifﬁn.
Goldstein, H. (1995). Multilevel statistical models. New York:
Halsted. 1164 Child Development Graber, J. A., & Brooks-Gunn, J. (1996). Transitions and turning
points: Navigating the passage from childhood through
adolescence. Developmental Psychology, 32, 768–776.
Gurian, M. (1996). The good son: Shaping the moral development of our boys and young men. New York: Putnam.
Hamburg, B. (1974). Early adolescence: A speciﬁc and
stressful stage of the life cycle. In G. V. Coelho, D. A.
Hamburg, & J. E. Adams (Eds.), Coping and adaptation
(pp. 101–124). New York: Basic Books.
Holsen, I., Kraft, P., & Vittersø, J. (2000). Stability in depressed
mood in adolescence: Results from a 6-year longitudinal
panel study. Journal of Youth and Adolescence, 29, 61–78.
Kagan, J., & Moss, H. A. (1962). Birth to maturity. New York:
Kahneman, D. (1999). Objective happiness. In D. Kahneman, E. Diener, & N. Schwarz (Eds.), Well-being: The
foundations of hedonic psychology. New York: Russell Sage
Kovacs, M. (1986). A developmental perspective on methods and measures in the assessment of depressive disorders: The clinical interview. In M. Rutter, C. Izard, & P. Read
(Eds.), Depression in young people: Developmental and clinical
perspectives (pp. 435–465). New York: Guilford Press.
Kozma, A., Stone, S., & Stones, M. J. (2000). Stability in components and predictors of subjective well-being (SWB):
Implications for SWB structure. In E. Diener & D. R.
Rahz (Eds.), Advances in quality of life theory and research
(pp. 13–30). London: Kluwer.
Larson, R. (1989). Beeping children and adolescents: A
method for studying time use and daily experience. Journal of Youth and Adolescence, 18, 511–530.
Larson, R., & Asmussen, L. (1991). Anger, worry, and hurt in
early adolescence: An enlarging world of negative emotions. In M. E. Colton & S. Gore (Eds.), Adolescent stress:
Causes and consequences (pp. 21–41). New York: Aldine de
Larson, R., Csikszentmihalyi, M., & Graef, R. (1980). Mood
variability and the psycho-social adjustment of adolescents. Journal of Youth and Adolescence, 9, 469–490.
Larson, R., & Delespaul, P. (1992). Analyzing experience
sampling data: A guidebook for the perplexed. In M.
DeVries (Ed.), The experience of psychopathology (pp. 58–
78). Cambridge, U.K.: Cambridge University Press.
Larson, R., & Ham, M. (1993). Stress and “storm and stress”
in early adolescence: The relationship of negative events
with dysphoric affect. Developmental Psychology, 29, 130–
Larson, R., & Lampman-Petraitis, C. (1989). Daily emotional states reported by children and adolescents. Child
Development, 60, 1250–1260.
Larson, R., Raffaelli, M., Richards, M. H., Ham, M., & Jewell, L. (1990). The ecology of depression in late childhood
and early adolescence: A proﬁle of daily states and activities. Journal of Abnormal Psychology, 99, 92–102.
Larson, R. W., & Richards, M. H. (1994a). Divergent realities:
The emotional lives of mothers, fathers, and adolescents. New
York: Basic Books.
Larson, R. W., & Richards, M. H. (1994b). Family emotions:
Do young adolescents and their parents experience the same states? Journal of Research on Adolescence, 4, 567–
Larson, R. W., & Richards, M. H. (2000, July). Changes in
daily emotions associated with entry into adolescence for urban African Americans. Paper presented at the biannual
meeting of the International Society for the Study of Behavioral Development, Beijing, China.
Larson, R. W., Richards, M. H., Moneta, G. B., Holmbeck,
G., & Duckett, E. (1996). Changes in adolescents’ daily
interactions with their families from ages 10 to 18: Disengagement versus transformation. Developmental Psychology, 32, 744–754.
Lawton, M. P., Kleban, M. H., Rajagopal, D., & Dean, J.
(1992). Dimensions of affective experience in three age
groups. Psychology and Aging, 7, 171–184.
Longford, N. T. (1993). Random coefﬁcient models. Oxford,
U.K.: Oxford University Press.
Magnusson, D. (1988). Individual development from an interactional perspective: A longitudinal study. Hillsdale, NJ:
McCall, R. B. (1977). Challenges to a science of developmental psychology. Child Development, 48, 333–344.
McCrae, R., & Costa, P. T. (1990). Personality in adulthood.
New York: Guilford Press.
Merikangas, K. R., & Angst, J. (1995). The challenge of depressive disorders in adolescence. In M. Rutter (Ed.),
Psychosocial disturbances in young people: Challenges for
prevention (pp. 131–165). New York: Cambridge University Press.
Moneta, G. B., & Csikszentmihalyi, M. (1996). The effect of
perceived challenges and skills on the quality of subjective experience. Journal of Personality, 64, 275–310.
Moneta, G. B., & Csikszentmihalyi, M. (1999). Models of
concentration in natural environments: A comparative
approach based on streams of experiential data. Social
Behavior and Personality, 27, 603–637.
Moneta, G. B., Schneider, B., & Csikszentmihalyi, M. (2001).
A longitudinal study of self-concept and experiential
components of self-worth and affect across adolescence.
Applied Developmental Science, 5, 125–157.
Moss, H. A., & Susman, E. J. (1980). Longitudinal study of
personality development. In O. Brim, Jr. & J. Kagan
(Eds.), Constancy and change in human development (pp.
530–595). Cambridge, MA: Harvard University Press.
Mroczek, D., & Kolarz, C. (1998). The effect of age on positive and negative affect: A developmental perspective
on happiness. Journal of Personality and Social Psychology,
Offer, D., Ostrov, E., & Howard, K. (1981). The adolescent: A
psychological self-portrait. New York: Basic Books.
Petersen, A. C., Compas, B. E., Brooks-Gunn, J., Stemmler,
M., Ey, S., & Grant, K. E. (1993). Depression in adolescence. American Psychologist, 48, 155–168.
Petersen, A. C., Kennedy, R. E., & Sullivan, P. (1991). Coping
with adolescence. In M. E. Colten & S. Gore (Eds.), Adolescent stress: Causes and consequences (pp. 93–110). Hawthorne, NY: Aldine de Gruyter.
Petersen, A. C., & Leffert, N. (1995). What is special about
adolescence? In M. Rutter (Ed.), Psychosocial disturbances Larson et al. in young people: Challenges for prevention (pp. 3–36). New
York: Cambridge University Press.
Petersen, A. C., Sargiani, P. A., & Kennedy, R. E. (1991). Adolescent depression: Why more girls? Journal of Youth and
Adolescence, 20, 247–271.
Pipher, M. (1994). Reviving Ophelia: Saving the selves of adolescent girls. New York: Putnam.
Prosser, R., Rasbash, G., & Goldstein, H. (1991). ML3 software for three-level analysis: Users’ guide for V.2. London:
Institute of Education, University of London.
Rosenberg, M. (1965). Society and the adolescent self-image.
Princeton, NJ: Princeton University Press.
Rothbart, M. K., & Bates, J. E. (1998). Temperament. In N.
Eisenberg (Ed.), W. Damon (Series Ed.), Handbook of child
psychology: Vol. 3. Social, emotional, and personality development (5th ed., pp. 105–176). New York: Wiley.
Rutter, M. (1980). Changing youth in a changing society: Patterns of adolescent development and disorder. Cambridge,
MA: Harvard University Press.
Rutter, M. (1986). The developmental psychopathology of depression: Issues and perspectives. In M. Rutter, C. Izard, &
P. Read (Eds.), Depression in young people: Developmental and
clinical perspectives (pp. 3–30). New York: Guilford Press.
Rutter, M., & Rutter, M. (1993). Developing minds: Challenge
and continuity across the life span. New York: Basic Books. 1165 Schaefer, E. S., & Bayley, N. (1963). Maternal behavior, child
behavior, and their intercorrelations from infancy through
adolescence. Monographs of the Society for Research in
Child Development, 28(3, Serial No. 87).
Simmons, R. G., & Blyth, D. A. (1987). Moving into adolescence: The impact of pubertal change and school context.
Hawthorne, NY: Aldine de Gruyter.
Simmons, R. G., Burgeson, R., Carlton-Ford, S., & Blyth, D. A.
(1987). The impact of cumulative change in early adolescence. Child Development, 58, 1220–1234.
Tellegen, A., Watson, D., & Clark, L. A. (1999). On the dimensional and hierarchical structure of affect. Psychological Science, 10, 297–303.
Thomas, D. L., & Diener, E. (1990). Memory accuracy in the
recall of emotions. Journal of Personality and Social Psychology, 59, 291–297.
Ushakov, G. K., & Girich, Y. P. (1972). Special features of
psychogenic depression in children and adolescents. In
A. Annell (Ed.), Depressive states in childhood and adolescence (pp. 510–516). Stockholm, Sweden: Almquist &
Verma, S., & Larson, R. (1999). Are adolescents more emotional? A study of the daily emotions of middle class Indian adolescents. Psychology and Developing Societies, 11,
View Full Document
This note was uploaded on 02/21/2012 for the course PYSC 521 taught by Professor Dr.zarrett during the Spring '11 term at South Carolina.
- Spring '11