Download as pdf or txt
Download as pdf or txt
You are on page 1of 34

020 kiuru (23-55) 2/4/08 1:14 PM Page 23

M E R R I L L -P A L M E R Q U A R T E R LY, V O L . 54, N O . 1

Peer Group Influence and Selection


in Adolescents’ School Burnout
A Longitudinal Study
Noona Kiuru, Kaisa Aunola, Jari-Erik Nurmi, Esko Leskinen, and Katariina
Salmela-Aro, University of Jyväskylä

The present study investigated the extent to which peer group similarity in school
burnout is due to peer group influence and the extent to which it is due to peer
group selection. Moreover, the roles of academic achievement and gender in
school burnout were examined. A total of 611 ninth graders were examined at
the beginning of the final term of comprehensive school, and 614 were exam-
ined at the end of the final term. The results of the Multilevel Latent Growth Mod-
eling showed that peer group influence was responsible for peer group similarity,
but no evidence was found for peer group selection. The results showed further
that high academic achievement protected group members against an increase
in school burnout.

Peer groups are among the most significant social contexts in adolescence
(Magnusson & Stattin, 1998; Rubin, Bukowski, & Parker, 1998). Previous
research has shown that the members of adolescents’ peer groups are similar
with respect to many characteristics and behaviors, such as internalizing and
externalizing problem behavior (Cairns & Cairns, 1994; Hogue & Steinberg,
1995). Research in academic settings has shown that the members of peer

Noona Kiuru, Kaisa Aunola, Jari-Erik Nurmi, and Katariina Salmela-Aro, Department of
Psychology, University of Jyväskylä, Finland. Esko Leskinen, Department of Mathematics and
Statistics, University of Jyväskylä, Finland.
This study is a part of the ongoing Kuopio School Transition (KST) study and was funded by
grants from the Finnish Academy (720 1822, 720 7421, 121 0319 213 486) and Finnish Cultural
Foundation. We are grateful to Christina Salmivalli for her valuable comments on the earlier draft
of the manuscript and for Asko Tolvanen for statistical consultation.
Correspondence concerning this article should be addressed to Noona Kiuru, Department of
Psychology, University of Jyväskylä, P.O. Box 35, 40014 Jyväskylä, Finland. Phone: (358)-14-260-
2817. Fax: (358)-14-260-4400. E-mail: noona.kiuru@psyka.jyu.fi.
Merrill-Palmer Quarterly, January 2008, Vol. 54, No. 1, pp. 23–55. Copyright © 2008 by Wayne
State University Press, Detroit, MI 48201.

23
020 kiuru (23-55) 2/4/08 1:14 PM Page 24

24 Merrill-Palmer Quarterly
groups are similar also with respect to academic achievement and learning
motivation (Chen, Chang, & He, 2003; Kindermann, McCollam, & Gibson,
1996; Ryan, 2001). These similarities have been explained by reference to
two processes: peer influence and selection (e.g., Cohen, 1977; Urberg, Luo,
Pilgrim, & Degirmencioglu, 2003). Although several studies have examined
the role of these two processes (e.g., Espelage, Holt, & Henkel, 2003; Kinder-
mann et al., 1996; Ryan, 2001), previous research on this topic has two major
limitations. First, only a few longitudinal studies (Cohen, 1977; Ennett &
Bauman, 1994; Kindermann et al., 1996) have examined peer group influ-
ence and selection in the same study. Consequently, little is known about the
relative importance of these processes in peer group similarity. Second, no
prior studies have examined the roles of peer group influence and selection in
adolescents’ burnout in school context. Consequently, the present study
investigated to what extent peer influence and selection contribute to peer
group similarity in school burnout at the end of comprehensive school. More-
over, the roles of academic achievement and gender in regard to school
burnout were investigated.

School Burnout among Adolescents


School provides an important developmental context for adolescents (Eccles,
2004). It has been found, for example, that adolescents’ perceptions of and
experiences in school are related to various adjustment outcomes, such as
psychological stress and negative affect (Cole, Peeke, Dolezal, Murray, &
Canzoniero, 1999; Crystal et al., 1994), self-esteem (Harter, 1996; Masi,
Sbrana, Poli, Tomaiuolo, Favilla, & Marcheschi, 2000), and subjective health
complaints (Aro, Paronen, & Aro, 1987; Hurrelmann, Engel, & Weidman,
1992). Students who dislike school are more likely to experience internal and
external problem behaviors, psychosomatic problems, and reduced quality of
life (e.g., Jessor, 1991; Kasen, Johnson, & Cohen, 1990). Although many
studies have examined the role of school experiences in adolescent adjust-
ment, less is known about the extent to which adolescents connect stress and
overload directly with school.
Burnout has been defined as emotional and mental exhaustion and as
chronic stress syndrome in a certain life context that is caused by context-
related overload, time pressure, and lack of resources (e.g., Demerouti,
Bakker, Nachreiner, & Schaufeli, 2001; Iacovides, Fountoulakis, Kaprinis,
& Kaprinis, 2003; Lee & Ashforth, 1996; Maslach, Schaufeli, & Leiter,
2001; Toppinen-Tanner, Ojajärvi, Väänänen, Kalimo, & Jäppinen, 2005).
The majority of research in burnout has been carried out in work contexts;
however, it also may be usefully applied in the school setting. On the basis
020 kiuru (23-55) 2/4/08 1:14 PM Page 25

Peer Group Socialization and Selection 25

of research conducted in work contexts, it can be assumed that burnout in


the school context consists of three dimensions: exhaustion due to study
demands, cynical and detached attitude toward one’s studies, and feelings
of incompetence as a student (Salmela-Aro & Näätänen, 2005; Schaufeli et
al., 2002). School burnout can be assumed to be due to a lack of fit between
the student’s internal resources for schoolwork and his or her own expecta-
tions for success in school or those held by other people such as teachers,
peers, and parents. Previous research in the work context has shown that
burnout leads to many negative consequences such as absenteeism, poor
health, and higher risk for depression (e.g., Toppinen-Tanner et al., 2005).
The same might be assumed to hold for burnout at school, although only a
few studies have been conducted in academic settings (Schaufeli, Martìnez,
Pinto, Salanova, & Bakker, 2002). The present study uses the concept of
school burnout to examine adolescents’ low well-being at school.
School burnout might be assumed to relate to many concepts that have
been used previously to describe adolescents’ school experiences. For exam-
ple, the exhaustion dimension of school burnout comes close to academic
stress (Murberg & Bru, 2004) and internal problem behavior (Nolen-
Hoeksema & Girgus, 1994), cynicism and detached attitudes toward school
resemble low learning motivation (Deci, Vallerand, Pelletier, & Ryan, 1991)
and low task-value (Eccles, 1983), and feelings of incompetence at school
has similarities with low academic self-concept (e.g., Harter, 1996; Saunders,
Davis, Williams, & Williams, 2004). Although such similarities would sug-
gest the use of the different subcomponents of school burnout as separate
indicators, there are also reasons for using burnout as an overall scale. First,
the different burnout dimensions have been found to be closely associated
(e.g., Toppinen-Tanner et al., 2005). Second, it has been suggested that no
single burnout factor alone can properly indicate the global nature of the con-
cept. Third, Kalimo, Pahkin, Mutanen, and Toppinen-Tanner (2003) have
suggested that it is difficult to draw unanimous conclusions on the causes of
overall burnout on the basis of the separate correlates of exhaustion, cyni-
cism, and lack of efficacy.
In the present study, two antecedents of school burnout were also
investigated. First, because burnout has been assumed to be caused by work
overload and lack of resources, we assumed that how well or badly adoles-
cents are doing at school may be related to their school burnout (see also
Schaufeli et al., 2002). Previous studies have shown, for example, that poor
academic performance and academic failures are associated with psycho-
logical stress and negative affect (Cole et al., 1999; Crystal et al., 1994),
low self-esteem (Harter, 1996; Masi et al., 2000), and subjective health
complaints (Aro et al., 1987; Hurrelmann et al., 1992).
020 kiuru (23-55) 2/4/08 1:14 PM Page 26

26 Merrill-Palmer Quarterly
Second, because previous research has also shown gender differences
in academic achievement and adjustment, gender was included as one
antecedent of school burnout. For example, girls tend to perform better at
school than boys (e.g., Dwyer & Johnson, 1997; Pomerantz, Altermatt, &
Saxon, 2002) and to attribute greater importance to academic achievement
compared to boys (Berndt & Miller, 1990; Murberg & Bru, 2004). How-
ever, girls have also been found to experience higher levels of stress (e.g.,
Ge, Lorenz, Conger, Elder, & Simons, 1994; Jose & Ratcliffe, 2004) and
internalized symptoms (e.g., Hoffmann, Powlishta, & White, 2004; Nolen-
Hoeksema & Girgus, 1994; Pomerantz et al., 2002).

Peer Group Influence and Selection


During their adolescent years, individuals begin to spend more time in peer
groups (Brown 1990; Rubin et al., 1998). Through interactions with their
peers, adolescents acquire a wide range of skills, attitudes, and experiences
(Brown, 1990; Bukowski, Newcomb, & Hartup, 1996; Rubin et al., 1998).
Previous research carried out on adolescents’ peer groups suggests that peer
interactions take place at multiple levels (Brown, 1990; Hinde, 1987; Rubin
et al., 1998). Brown (1989), for example, described peer interactions as oper-
ating on three levels: dyads, cliques, and crowds. A dyad refers to a reciprocal
relationship between two individuals. A clique consists of a small number of
adolescents who hang around together and develop close relationships.
Crowds, in turn, are reputation-based peer groups and larger collectives of
similarly stereotyped individuals. Gender differences have also been found in
peer group composition. For example, girls’ peer groups are typically more
intimate and tightly connected than those of boys (Benenson, 1990; Urberg,
Degbirmenciogblu, Tolson, & Halliday-Scher, 1995).
Previous research has shown that the members of adolescents’ peer
groups resemble each other in many respects, such as propinquity, age, sex,
and race (e.g., Cairns & Cairns, 1994); academic achievement and motiva-
tion (Chen et al., 2003; Ryan, 2001); internalizing distress (Hogue & Stein-
berg, 1995); and various external problem behaviors (Espelage et al., 2003;
Kiesner, Poulin, & Nicotra, 2003; Urberg, Degbirmenciogblu, & Pilgrim,
1997). Similarity between peer group members has been explained by peer
group influence and selection (e.g., Cairns & Cairns, 1994; Cohen, 1977;
Ennett & Bauman, 1994). Peer group selection (cf. selective association)
refers to the tendency of individuals to seek the company of like-minded
peer groups. In turn, peer group influence (cf. reciprocal socialization)
refers to the tendency of group members to shape and reinforce each other’s
shared attributes and behaviors over time. Urberg et al. (2003) presented a
020 kiuru (23-55) 2/4/08 1:14 PM Page 27

Peer Group Socialization and Selection 27

two-stage model of peer group processes in which the first phase is the
acquisition of a peer group (i.e., peer group selection). By choosing to asso-
ciate with a particular peer group, adolescents select a social context that
exposes them to a particular set of values, behaviors, and opportunities. The
second phase consists of the socialization process in which peer group
members either conform or do not conform to each other’s behavior.
The processes of peer group influence and selection can be separated
only within longitudinal studies (e.g., Kandel, 1978). It has been proposed,
on the one hand, that peer group influence operates if the behavior of a peer
group at Time 1 predicts changes in adolescent behaviors between Times 1
and 2 (e.g., Degbirmenciogblu, Urberg, Tolson, & Richard, 1998). Peer influ-
ence, according to this definition, has been found to operate in adolescents’
peer groups with respect to academic achievement and motivation (Kinder-
mann et al., 1996; Ryan, 2001); smoking, drinking and, drug use (Cairns &
Cairns, 1994; Ennett & Bauman, 1994; Urberg et al., 1997); aggressive
behavior (Cairns & Cairns, 1994; Espelage et al., 2003); and internalizing
distress (Hogue & Steinberg, 1995). Peer group influence has been ex-
plained by observational learning, identification, and conformity desires or
pressures (Bandura, 1977; Berndt, 1999; Suls & Wheeler, 2000). In the
present study we assumed that peer influence may contribute to peer group
similarity in school burnout in many ways. For example, previous research
has shown that the members of peer groups typically discuss important
matters in their lives (Malmberg, 1996). Feelings and attitudes concerning
school might be assumed to be among such experiences. In addition, ado-
lescents may coruminate about their school-related stress and negative feel-
ings in their peer groups. For example, Rose (2002) found that girls have a
higher tendency to coruminate about their problems and negative feelings
than boys and that corumination was related to high-quality close friend-
ships as well as to depression and anxiety. Such corumination may then
lead to higher levels of school burnout, including cynical attitudes toward
school and a sense of inadequacy as a student. Also, peer group norms con-
cerning the importance of academic achievement may contribute to peer
group members’ school burnout.
On the other hand, peer group selection can be inferred if individuals
are similar to their new peer groups prior to group formation (e.g., Urberg,
Degbirmenciogblu, & Tolson, 1998) or if they resemble their new peer group
more than the group they left (e.g., Kandel, 1978). Evidence for peer group
selection has been found for many characteristics such as age, gender, and
ethnicity (Cairns & Cairns, 1994); academic achievement and motivation
(Kindermann et al., 1996); smoking (Ennett & Bauman, 1994); and inter-
nalizing distress (Hogue & Steinberg, 1995). It has been suggested that the
020 kiuru (23-55) 2/4/08 1:14 PM Page 28

28 Merrill-Palmer Quarterly
impact of peer group selection is based on shared interests and reciprocal
liking (Brown, 1989; Urberg et al., 2003). In the present study we assumed
that selection may play a role in peer group similarity in school burnout.
For example, feelings of exhaustion, cynicism, and incompetence associ-
ated with school as well as enthusiasm, optimism, and competence are
important attitudes that may play a role in the selection of peer groups.
However, they may play an even more important role in the exclusion of
some adolescents from a peer group by other group members. For example,
peer groups having a number of members with high levels of burnout may
be likely to exclude individuals with optimistic view of school.
Previous research on peer influence and selection in the peer group
context among adolescents has two major limitations. First, only a few
studies (Cohen, 1977; Ennett & Bauman, 1994; Kindermann et al., 1996)
have examined peer group influence and selection in the same study. Many
studies have examined only peer group influence and have controlled for
peer group selection (e.g., Espelage et al., 2003; Ryan, 2001). Second, few
previous studies have examined the roles of peer group influence and selec-
tion in the lack of well-being at school. In the present study, low well-being
at school was operationalized as school burnout.

Aims
The present study analyzed adolescents’ peer group membership and school
burnout across two measurements during their final term of comprehensive
school by examining the following research questions:

1. Do the members of adolescents’ peer groups show similar levels of


school burnout during the final term of comprehensive school?
2. To what extent does peer influence explain peer group similarity in
school burnout? We assumed that if the members of stable peer
groups (i.e., groups that either remained unchanged or to which new
members were joined between Time 1 and Time 2) showed similar
changes in their school burnout from Time 1 to Time 2, this would
provide evidence for peer group influence. Moreover, we assumed
that if the members of stable peer groups displayed more similarity
with each other at Time 2 than at Time 1 with respect to school
burnout (i.e., become more homogeneous across time), this would
also provide evidence of peer group influence.
3. To what extent does peer group selection explain peer group simi-
larity in school burnout? We assumed that if the members of new
peer groups at Time 2 were already similar to each other at Time 1
020 kiuru (23-55) 2/4/08 1:14 PM Page 29

Peer Group Socialization and Selection 29

with regard to school burnout or if they changed in a similar direc-


tion, this would provide evidence for peer group selection. Simi-
larly, we assumed that if the adolescents who changed their peer
group between Time 1 and Time 2 showed greater similarity with
their peer groups at Time 2 than with their peer groups at Time 1,
this again would provide evidence for peer group selection. More-
over, we assumed that if the adolescents who changed their peer
group between Time 1 and Time 2 differed more from their peer
group at Time 1 compared to those adolescents who stayed in the
same group, this would also provide evidence for selection.
4. Do academic achievement and gender predict school burnout and
changes in burnout? To what extent do these associations operate at
the peer group level, and to what extent do they occur at the individ-
ual level?

Schooling in Finland
Finnish children start their education at kindergarten during the year of
their 6th birthday. One year later, at the age of 7, they move to comprehen-
sive school, where they remain for the next nine years. Comprehensive
school divides into a lower level (first through sixth grades) and an upper
level (seventh through ninth grades). Upper comprehensive school students
spend approximately half of their school hours with their classmates study-
ing obligatory subjects and about half in classes with other students study-
ing mainly selective subjects.
Up to age 16, all Finnish adolescents have a similar basic education.
After comprehensive school, a total of 55% of adolescents enter senior sec-
ondary schools and 37% enter vocational schools, 2% go to a special 10th
grade, and 6% exit formal education (School statistics, Central Statistical
Office of Finland, 2003). The ninth grade of comprehensive school can be
assumed to be stressful for students because high academic achievement is
required to gain entry to the senior secondary schools and to some vocational
schools.

Method
Participants
The present study is part of the Kuopio School Transition (KST) study
(Salmela-Aro, Niemivirta, & Nurmi, 2003). The KST is an ongoing study
020 kiuru (23-55) 2/4/08 1:14 PM Page 30

30 Merrill-Palmer Quarterly
with the aim of examining adolescents’ life-planning and well-being in
middle and late adolescence. The participants of the present study were
ninth graders facing the transition to secondary education. All the ninth-
grade students (N = 773) from eight schools in one medium-sized town
(population = 88,000) in Central Finland were recruited for the study. The
participants were asked to participate twice during their final term of com-
prehensive school: first at the beginning of the spring term (end of January)
and second at the end of the spring term (end of May). The interval between
measurements was four months. At Time 1, 611 adolescents (297 girls, 314
boys) from eight comprehensive schools participated in the study. At Time
2, the number of participants was 614 (308 girls, 306 boys). A total of 517
participants (265 girls, 252 boys) filled in the questionnaires at both time
points. The median age of the participants was 15 (M = 15; SD = 0.34). The
number of ninth-grade students in schools (N = 8) varied between 55 and
186, with an average of 97 students.
The participation rate of the different schools selected for the present
study varied between 70% and 95%. Attrition at Time 1 was due to absence
from school on the day when the data were gathered (N = 144, 19.35%), or
to inappropriate answers or incomplete personal data (N = 19, 2.45%). At
the second measurement, 161 students (16.92%) were absent during the
measurement day, and 30 questionnaires (3.88%) were excluded from the
study due to inappropriate answers or incomplete personal data. A total of
96 adolescents who did not answer at Time 1 were reached at Time 2. By
contrast, 94 adolescents who participated in the study at Time 1 did not fill
in the questionnaire at Time 2. Attrition analyses were carried out, and the
adolescents who participated in both measurements were compared to
those who participated only at Time 1 or at Time 2. The results showed that
the adolescents who participated in both measurements had higher aca-
demic achievement at school at the previous semester (measured at Time 1;
M = 8.06, SD = 0.81) than those who participated in only one measurement
(M = 7.77, SD = 0.87); t(565) = 3.75. No selection effects were found
according to gender or school burnout.
The majority of the participants (99%) were Finnish-speaking, 1% of
them having some other mother tongue. A total of 63.2% of participants lived
with both of their biological parents, 10% lived in families consisting of their
mother or father living with her or his new spouse, 23.3% lived with their
lone mother, 2.7% lived with their lone father, and 0.8% lived with somebody
else. Moreover, 27% of the fathers and 21.9% of the mothers of the partici-
pants worked in higher white-collar occupations (e.g., physicians, teachers,
lawyers, managers), 16% of the fathers and 50.6% of the mothers were in
lower white-collar occupations (e.g., secretaries, salespersons, nurses),
020 kiuru (23-55) 2/4/08 1:14 PM Page 31

Peer Group Socialization and Selection 31

35.9% of fathers and 14.4% of mothers were in blue-collar occupations (e.g.,


welders, construction workers, mechanics), 7% of fathers and 1.2% of moth-
ers were private entrepreneurs, 1.2% of fathers and 1.6% of mothers were
students, 1.7% of fathers and 1.4% of mothers were retired, and 11.2% of
fathers and 8.9% of mothers had other status (e.g., unemployed).
The questionnaires were group-administered to the students in their
classrooms during regular school hours. The adolescents answered ques-
tions concerning school burnout and peer relations at both time points. Aca-
demic achievement and gender were measured at Time 1.

Measures
School burnout. School burnout was examined with the School Burnout
Scale (BBI-10) developed by Salmela-Aro and Näätänen (2005) and accord-
ing to adolescents’ estimations from the previous month. The scale is based
on the Burnout Inventory for Working Life (BBI-15) (Näätänen, Aro,
Matthiesen, & Salmela-Aro, 2003; Salmela-Aro, Näätänen, & Nurmi, 2004).
The scale consists of 9 unidimensional items measuring school burnout (e.g.,
I feel overwhelmed by schoolwork; I have become less interested in school-
work and I often think of dropping out of school; I expected to do better aca-
demically than I have done), which were rated on a 6-point scale (1 = strongly
disagree, 6 = strongly agree). For the purpose of this study, a sum score was
formed on the basis of the 9 items to indicate the level of adolescents’ school
burnout. The Cronbach’s alpha reliability for the scale was .86 at Time 1 and
.89 at Time 2. In the present data, school burnout was found to correlate posi-
tively with depression (r = .55) and negatively with self-esteem (r = –.42).
Academic achievement. Academic achievement was measured at Time
1 by asking the participants to report their grade point average (GPA) from
the preceding spring term (i.e., Time 0). GPA ranged from 4 (lowest) to 10
(highest). This self-reported GPA has shown to correlate .96 with the actual
GPA (Holopainen & Savolainen, 2005).
Gender. Gender was coded by asking adolescents to circle the correct
alternative (1 = girl, 2 = boy).
Self-esteem. Participants’ self-esteem was measured with the abbrevi-
ated version of Rosenberg’s Self-Esteem Scale (Rosenberg, 1965). The
scale consisted of 5 unidimensional items that the adolescents rated on a 7-
point scale (1 = strongly disagree, 7 = strongly agree). A sum score was cal-
culated from all 5 items. The Cronbach’s alpha for the scale was .76 at Time
1 and .74 at Time 2.
Depression. Participants’ depression was measured by using the
Depression Scale (DEPS) developed by Salokangas, Stengård, and Putanen
020 kiuru (23-55) 2/4/08 1:14 PM Page 32

32 Merrill-Palmer Quarterly
(1994). The scale included 10 questions concerning participants’ mood dur-
ing the last month (e.g., “I felt myself sad; I felt like my future is hopeless”),
which were rated with on a 4-point scale (1 = not at all, 4 = very much). A
sum score was calculated from all 10 items. The Cronbach’s alpha reliabil-
ity for the scale was .92 at Time 1 and .92 at Time 2.
Peer group identification. Participants’ peer groups were identified by
using the sociometric procedure developed by Coie, Dodge, and Coppotelli
(1982). The participants were asked to nominate up to three same-age school-
mates from their school with whom they most liked to spend their time and
three adolescents with whom they least liked to spend time (positive and neg-
ative nominations). Cross-gender nominations were allowed in order to gain
a realistic picture of adolescents’ peer relations. According to Terry and Coie
(1991), allowing cross-gender nominations enlarges the pool of nominations,
which in turn increases the stability of the measurement.
To identify peer groups, only positive peer nominations were used. At
both time points, approximately 90% of positive nominations were of
same-sex peers. The number of positive peer nominations ranged from 0 to
3, with an average of 2.5 nominations. No nominations were given by 10%
of the participants. Girls gave more nominations than boys and also had
more reciprocal nominations than boys at both measurements. Approxi-
mately 66% of nominations were for peers from the same class, 27% for
peers from other classes in the same school, 3% for the peers from other
schools, and 4% for unknown peers.
Sociograms based on the positive nominations were drawn for all eight
schools. Peer group membership was defined on the basis of reciprocal,
unilateral, and indirect links. A reciprocal link is a reciprocal positive nom-
ination between two individuals. A unilateral link is a unilateral positive
nomination from one adolescent to another. An indirect link (i.e., common
peer) is a link between two individuals via a third person.
The following criteria were used to determine group membership. To be
categorized as a member of a particular group, (1) at least 50% of a person’s
reciprocal and unilateral links have to be within the peer group, and (2) either
a reciprocal, a unilateral, or an indirect link has to exist from each member to
every other member of the peer group. Participants were assigned to groups if
both the group membership criteria were met. If participants had links to mul-
tiple peer groups, they were assigned to the peer group in which they had
most of their peer links. Also, those adolescents who were absent during the
day of measurement (Time 1: N = 25; Time 2: N = 27) were included in peer
groups if 50% of the peer nominations they received were in a particular peer
group. Peer groups consisted typically of at least three members. However,
groups consisting of only two members were also considered peer groups if
020 kiuru (23-55) 2/4/08 1:14 PM Page 33

Peer Group Socialization and Selection 33

the members shared a reciprocal tie and did not belong to any larger peer
groups. The criteria resembled those used in other studies using social net-
work analysis to identify peer groups (e.g., Ennett & Bauman, 1994;
Espelage et al., 2003; Ryan, 2001; Urberg et al., 1997).

Analysis Strategy
The research questions were investigated by using intraclass correlations
and Multilevel Latent Growth Modeling (MLGM) (Muthén & Muthén,
1998–2006; see also Duncan, Alpert, Hops, Stoolmiller, & Muthén, 1997).
Intraclass correlations (ICC) were calculated to determine what pro-
portion of the variance in the observed variables is due to the peer-group
level (between-peer group variation) and what is due to the individual level
(within-peer group variation) (Heck, 2001; Muthén, 1991). The statistical
significance of the intraclass correlations and the variance estimates at the
peer group level and at the individual level were also examined.
The MLGM method combines two more frequently used statistical
analyses. The first is Latent Growth Modeling (LGM), which estimates dif-
ferent growth components such as level (intercept) and linear change
(slope) from data consisting of at least two measurements. The simplest
form of the LGM involves one variable measured the same way at two time
points. This allows both the initial level and the rate of change between the
two time points to be estimated (Duncan, Duncan, Strycker, Li, & Alpert,
1999; Rogosa, Brandt, & Zimowski, 1982). The second is multilevel (hier-
archical) analysis, which provides a tool to differentiate between the varia-
tion in the level and change components that is due to similarities among
peer group members (i.e., variance between peer groups, between-peer
group level) and which is due to individual differences (i.e., variation
between individuals within peer groups, within-peer group level). By com-
bining these two statistical procedures, the MLGM allowed us not only to
estimate the level and change components in school burnout but also to
examine which parts of these growth components are shared by peer group
members and which parts are typical of individual adolescents. Moreover,
the MLGM analyses enabled both group-level and individual-level predic-
tors of level and rate of change to be entered into the analysis. The present
data included two measurements of school burnout, and adolescents’ peer
groups were also identified at both measurements.
The first aim of the present study was to examine the extent to which
the members of adolescents’ peer groups would be similar in terms of their
school burnout. In order to answer this research question, we examined the
intraclass correlation for school burnout measured at Time 1 by using the
020 kiuru (23-55) 2/4/08 1:14 PM Page 34

34 Merrill-Palmer Quarterly
peer grouping at Time 1 as a clustering variable and for burnout measured
at Time 2 by using the peer grouping at Time 2 as a clustering variable.
Our second research question was to examine to what extent peer influ-
ence would explain peer group similarity in school burnout. To answer this
question, we first conducted MLGM analyses in which peer group influence
was examined among adolescents in stable peer groups by using the peer
groupings at Time 1 as a clustering variable. Stable peer groups included only
those peer groups that remained unchanged or to which new members were
joined between the two measurements. In this set of analyses, we investigated
whether the members of stable peer groups identified at Time 1 would be sim-
ilar with respect to their level of school burnout and whether they would also
show a similar rate of change in burnout from Time 1 to Time 2. We assumed
that if the adolescents belonging to same peer groups changed in the same
direction across time with regard to school burnout (i.e., there is statistically
significant between-peer group variation in the rate of change), this would
provide evidence for peer group influence. If an effect of peer group influence
was found, the predictors of level of and rate of change in school burnout—
that is, academic achievement and gender—would be added to the model both
at the peer group and individual levels. As a second way of analyzing the role
of peer influence in peer group similarity, an additional analysis was carried
out by comparing peer group homogeneity between Time 1 and Time 2 among
stable peer groups. We assumed that if peer group members displayed more
similarity to each other at Time 2 than at Time 1 with respect to school burnout
(i.e., intraclass correlation of school burnout increases over time), this would
provide further support for peer group influence.
Our third research question was to examine the extent to which peer
selection would explain the peer group similarity in school burnout. We pre-
sumed that peer group selection can operate only when adolescents form new
peer groups, when move from one group to another, or when nonmembers
become peer group members. Two kinds of analyses were carried out in order
to examine peer group selection. First, peer group similarity in school
burnout at both measurements was examined among new peer groups by
using the peer groupings at Time 2 as a clustering variable. Only new peer
groups that were formed between Time 1 and Time 2 were included in these
analyses. The analyses were started by calculating intraclass correlations for
school burnout at Time 1 and Time 2 on the basis of peer groupings measured
at Time 2. If statistically significant between-peer group-level variance was
found in the burnout variables at Time 1 or Time 2, only then would MLGM
analyses be carried out. We assumed that if the peer groupings at Time 2
among new peer groups accounted for the variance at the initial level (Time
1) or in the rate of change in school burnout between Time 1 and Time 2, this
would provide evidence for peer group selection. In other words, if the
020 kiuru (23-55) 2/4/08 1:14 PM Page 35

Peer Group Socialization and Selection 35

adolescents in the new peer groups at Time 2 were already similar at Time 1
or if they changed in a similar direction, this would provide evidence for peer
group selection. By contrast, if no statistically significant variance in the level
or rate of change in school burnout emerged when using the groupings at
Time 2, this would provide evidence against the role of selection in the peer
group similarity. The role of predictors would be examined only if a peer
group selection effect had been found.
Second, to examine further the role of peer group selection, another set
of analyses were carried out for those adolescents who changed their peer
group between measurements. The aim of these analysis were to examine
(1) whether the adolescents who changed their peer groups would differ
more from their peer group at Time 1 compared to those adolescents who
stayed in the same group and (2) whether the adolescents who changed
their peer group resembled their peer groups more at Time 2 than their peer
groups at Time 1. These analyses were carried out by conducting independ-
ent sample t-test (question 1) and paired-sample t-test (question 2) by using
the absolute values of the scores standardized within the peer groupings.
The multilevel analyses were performed by using the Mplus statistical
package Version 4.1 (Muthén & Muthén, 1998–2006). By using the miss-
ing data method, we were able to utilize all the available observations in the
data set when estimating the parameters of the models. Because the vari-
ables were skewed, the parameters of the models were estimated using the
Maximum Likelihood Robust (MLR) estimator (Muthén & Muthén,
1998–2006). The MLR produces robust standard errors and a χ2-test sta-
tistic for missing data with nonnormal outcomes by means of a sandwich
estimator (Muthén & Muthén, 1998–2006). The goodness-of-fit of the esti-
mated models was evaluated by four indicators: χ2-test, Comparative Fit
Index (CFI), Root Mean Square Error of Approximation (RMSEA), and
Standardized Root Mean Square Residual (SRMR). The within-level
(below the diagonal) and between-level (above the diagonal) correlations
and means and variances for the observed variables for stable peer groups
are presented in Table 1, and those for new peer groups formed between
measurements are presented in Table 2.

Results

Peer Group Composition


A total of 148 peer groups were identified at Time 1. The majority (N = 139)
of the peer groups were composed of same-sex adolescents. The number of
girls’ peer groups was 72, and the number of boys’ peer groups was 67.
Only 9 peer groups consisted of members of both sexes. The proportion of
020 kiuru (23-55)

36
2/4/08

Table 1. Sample Correlation Matrix and Intraclass Correlations of Observed Variables according to the Peer Grouping
1:14 PM

at Time 1 among Stable Peer Groups

Variables 1. 2. 3. 4. Means ICC Variance (Standard Error)


1. School burnout (T1) 1.00 .46a –.15 –.65a 2.44 .18c 0.11 (0.06)d
Page 36

2. School burnout (T2) .48a 1.00 –.42a –.11 2.42 .33a 0.34 (0.09)a
3. Academic achievement (T0) –.17b –.06 1.00 –.32b 8.10 .49a 0.32 (0.08)a
4. Gender –.00 –.03 –.11 1.00 1.40 .96a 0.23 (0.01)a
Variance (Standard error) 0.50(0.06)a 0.67(0.08)a 0.34 (0.04)a 0.01(0.01)
Note. Between-level above the diagonal, and within-level below the diagonal; NWithin = 248, NBetween = 75; a p < .001; b p < .01; c p < .05.; d p < .10.
Merrill-Palmer Quarterly
020 kiuru (23-55)
2/4/08

Table 2. Sample Correlation Matrix and Intraclass Correlations of School Burnout according to the Peer Grouping
1:14 PM

at Time 2 among New Peer Groups

Variables 1. 2. 3. 4. Means ICC Variance (Standard Error)


1. School burnout (T1) 1.00 .98a .83a –.53b 2.48 .00 —
Page 37

2. School burnout (T2) .58a 1.00 .02 –.55b 2.38 .17 0.19 (0.14)
Peer Group Socialization and Selection

3. Academic achievement (T0) –.04 –.03 1.00 –.47b 8.07 .24c 0.14(0.07)a
4. Gender –.11 –.03 –.17 1.00 1.42 .97a 0.23(0.02)a
Variance (Standard error) 0.77(0.17)a 0.94(0.20)a 0.47(0.09)c 0.01(0.01)
Note. Between-level above the diagonal, and within-level below the diagonal; NWithin = 96, NBetween = 30; a p < .001; b p < .01; c p < .05.
37
020 kiuru (23-55) 2/4/08 1:14 PM Page 38

38 Merrill-Palmer Quarterly
adolescents who did not belong to any peer group was 17% (N = 107). The
size of the peer groups ranged from 2 to 7 (M = 3.51).
At Time 2, a total of 140 peer groups were identified. Again, most of
the peer groups (N = 131) consisted of members of the same sex, and 17%
(N = 102) of the adolescents did not belong to any peer group. The number
of girls’ peer groups was 72, boys’ peer groups 59, and mixed peer groups
9. The size of a peer group ranged from 2 to 8 (M = 3.50).
At both measurements, girls (Time 1: M = 0.75, SD = 0.29; Time 2: M =
0.68, SD = 0.32) had more reciprocal peer relations in their peer groups
than boys (Time 1: M = 0.60, SD = 0.37, t[508] = 5.11; Time 2: M = 0.52,
SD = 0.37, t[497] = 5.10). Boys were also more likely (Time 1: Adjusted
standardized residual = 2.7; Time 2: Adjusted standardized residual = 3.0)
than girls not to belong to any peer group (T1: χ2[1, N = 652] = 7.32, p <
.01; T2: χ2[1, N = 619] = 8.86, p < .01). Gender differences in group size
were not statistically significant.
A total of 38 peer groups (26%) remained unchanged from Time 1 and
Time 2, 37 (25%) peer groups had new members at Time 2, and 73 (49%)
peer groups experienced substantial changes in group membership between
the two measurements. These results are consistent with previous research on
peer group stability (Cairns & Cairns, 1994; Degbirmenciogblu et al., 1998;
Hogue & Steinberg, 1995; Ryan, 2001). Of the 73 peer groups that experi-
enced substantial changes between the two measurements, 41 groups were
dissolved, and the number of peer groups that were either divided into two
parts or in which two or more former peer groups had merged was 32. A total
of 30 new peer groups were formed from the dissolved peer groups at Time 2.

Peer Group Similarity


Our first research question was to examine whether the members of adoles-
cents’ peer groups would show similar levels of school burnout during the
final term of comprehensive school. The results of the intraclass correla-
tions at Time 1 for the whole sample, using the peer groupings at Time 1,
showed that 10% (p < .05) of the variance in school burnout was due to sim-
ilarity among the peer group members (between variance estimate = 0.07,
standard error = 0.03, p < .05). The intraclass correlations at Time 2 for the
whole sample, using the peer grouping of Time 2, showed that 16% (p <
.001) of the variance of school burnout was due to similarity among the
peer group members (between variance estimate = 0.30, standard error =
0.08, p < .001). These results showed that the members of adolescents’ peer
groups were somewhat similar in terms of their school burnout, particularly
at Time 2.
020 kiuru (23-55) 2/4/08 1:14 PM Page 39

Peer Group Socialization and Selection 39

Peer Group Influence in School Burnout


To examine the role of peer influence in the similarity found among the
members of adolescents’ peer groups with regard to school burnout, two
groups of adolescents were included in the analyses: those who belonged to
unchanged peer groups between Time 1 and Time 2 and those who
belonged to peer groups that had gained new members. These two types of
peer groups were named stable peer groups (N = 75). The analyses concern-
ing peer group influence commenced by examining how the variance in
school burnout at Time 1 and Time 2 and its predictor variables at Time 1,
that is, academic achievement and gender, divided into the between peer
group and within peer group components by using the peer groupings at
Time 1 as a clustering variable. Examination of the intraclass correlations
showed that 18% (p < .05) of the variance in school burnout at Time 1 and
33% (p < .001) at Time 2, was due to similarities among peer group mem-
bers (between peer group effect), while the rest was due to individual dif-
ferences (within peer group effect). Moreover, 49% (p < .001) of the
variance in academic achievement and 96% (p < .001) in gender (i.e., peer
group homogeneity in gender) was due to similarities among peer group
members, and the rest was due to individual differences. The results
showed further that in all the variables, the between level variance and
within level variance were statistically significant: peer group members
were similar to each other not only in school burnout but also in the predic-
tor variables (see Table 1).
Consequently, multilevel analyses for peer group influence among the
stable groups were carried out in the following steps. First, MLGM was
carried out for school burnout by using the peer groupings at Time 1 as a
clustering variable. In this model, two growth components, that is, the ini-
tial level and the rate of change, were specified for school burnout. The
growth components were allowed to correlate with each other. Because we
had only two measurement points, the measurement error variances of the
observed variables were fixed to zero. The estimates of the between peer
group variance (variation due to peer group similarity) and the within peer
group variance (variation due to individual differences) in the level and
change components and their covariances at different levels are shown in
Table 3. Next, in order to examine whether the level of school burnout pre-
dicted the rate of change in burnout, the path from the level to the rate of
change was estimated at both levels. The final model, which includes only
statistically significant paths (χ2 [2, NWithin = 248, NBetween = 75] = 0.19, p =
0.91; CFI = 1.00, TLI = 1.07; RMSEA = 0.00; SRMRBetween = 0.04, SRMR-
Within = 0.00), is presented in Figure 1.
020 kiuru (23-55) 2/4/08 1:14 PM Page 40

40 Merrill-Palmer Quarterly
Table 3. Unstandardized Parameter Estimates (Standard Errors) for Multilevel
Latent Growth Modeling for School Burnout according to the Peer Groupings
at Time 1 for Stable Peer Groups

Growth Parameters of School Burnout Peer Influence Effect


Between Level Within Level
Means
Level 2.44(0.06)a
Slope –0.02(0.08)
Variances
Level 0.11(0.06)c 0.50(0.06)a
Slope 0.28(0.09)b 0.64(0.12)a
Covariance (level, slope) –0.02(0.05) 0.23(0.06)a
Note 1. NWithin = 248, NBetween = 75; p < .001; p < .01; p < .10.
a b c

The results showed that at the peer group level there was statistically
significant between peer group variance both at the initial level of school
burnout and in the rate of change in school burnout (see Figure 1). In other
words, the members of stable peer groups identified at Time 1 were similar
in their level of school burnout, and they also showed similar changes in
burnout during the final term of comprehensive school, suggesting evi-
dence for peer group influence. The initial level of school burnout shared
by a peer group did not predict the rate of change in the peer group’s school
burnout. There was no change in school burnout at the mean level.
At the individual level, there was also statistically significant variance
both in the level and rate of change in school burnout. Moreover, the initial
level of school burnout negatively predicted the rate of change in school
burnout: the lower the initial level of burnout, the more it increased from
Time 1 to Time 2.
In order to make sure that the finding for peer group influence in school
burnout was not due to confounding effects of the variables closely related
to school burnout, that is, depression and low self-esteem, we also carried
out the same analyses described above by adding depression and self-
esteem as control variables. The results of these analyses showed, however,
that the effect of peer group influence on school burnout remained even
after controlling the effects of depression and self-esteem.
Second, we investigated the extent to which academic achievement
and gender would predict the level and rate of change in school burnout at
the peer group level and at the individual level among stable peer groups.
The paths from the predictors to the level and rate of change in school
020 kiuru (23-55) 2/4/08 1:14 PM Page 41

Peer Group Socialization and Selection 41

L evel C hange

0
1 1 1

B etween School School


B ur nout (T 1) B ur nout (T 2)

School School
Within
B ur nout (T 1) B ur nout (T 2)

1 1 0 1

L evel C hange
R 2 = 0.18
-.49***
Figure 1. MLGM for peer group influence (grouping at Time 1) among stable groups in
school burnout. The paths between variables are based on standardized estimates. Note 1. * p
< .05; ** p < .01; *** p < .001. Note 2. T1 = Time 1, T2 = Time 2.

burnout were estimated at both the peer group and individual levels. The
final model, which includes only statistically significant paths (χ2 [7, NWithin
= 248, NBetween = 75] = 8.41, p = 0.30; CFI = 0.98, TLI = 0.97; RMSEA =
0.03; SRMRBetween = 0.13, SRMRWithin = 0.04), is presented in Figure 2.
The results at the peer group level showed that academic achievement
predicted the rate of change in school burnout: the lower the level of aca-
demic achievement typical of a peer group, the greater the increase in
school burnout reported by peer group members (see Figure 2). Moreover,
gender predicted the level of school burnout typical of the peer group: peer
groups composed mainly of girls showed more school burnout than peer
groups composed mainly of boys. Gender also predicted the level of aca-
demic achievement: peer groups composed mainly of girls showed higher
academic achievement than those of boys.
The results at the individual level showed that academic achievement
negatively predicted the level of school burnout: the lower the adolescents’
academic achievement, the more they suffered from school burnout inde-
pendently of their peer group (see Figure 2). No effect of gender was found.
It is noteworthy here that adolescents’ peer groups were almost homoge-
neous with regard to gender, that is, 96% of the variation in it was explained
by peer group level. Consequently, gender can be considered mainly as a
peer group level predictor.
020 kiuru (23-55) 2/4/08 1:14 PM Page 42

42 Merrill-Palmer Quarterly
-.47*

G ender 1

L evel C hange
-.68*** R 2=.46 R 2=.22
0
-.34*
1 1 1

B etween G PA
School School
(T 0) R 2 = 0.11
B ur nout (T 1) B ur nout (T 2)

G PA School School
Within
(T 0) B ur nout (T 1) B ur nout (T 2)

-.19**
1 1 0 1

L evel C hange
R 2=.17
R 2=.04
-.42***

Figure 2. MLGM for school burnout with predictors (grouping at Time 1) among stable
peer groups. The paths between variables are based on standardized estimates. Note 1. * p <
.05; ** p < .01; *** p < .001. Note 2. T0 = Time 0, T1 = Time 1, T2 = Time 2. Note 3. 1 girl =
1; boy = 2.

Besides examining whether the members of stable peer groups shared


similar changes in their school burnout, we also examined whether peer
group homogeneity increased over time among the stable groups. This was
done by comparing peer group similarity (i.e., intraclass correlations) in
school burnout between Time 1 and Time 2 by using the peer groupings at
Time 1 as a clustering variable. The comparison was done using the Mplus
program (Muthén & Muthén, 1998–2006) by constructing a model for the
observed variables at Time 1 and Time 2 in which the statistical signifi-
cance of the intraclass correlations as well as the difference between the
intraclass correlations at the two timepoints were tested (H0: Difference
between ICCs at Time 1 and Time 2 is zero; H1: Difference between ICCs is
larger than zero). The results showed that the intraclass correlation for
school burnout was statistically significant both at Time 1 (intraclass corre-
lation estimate = 0.18, standard error = 0.09, p < .05) and at Time 2 (intra-
class correlation estimate = 0.34, standard error = 0.11, p < .001). The
results concerning the difference between the intraclass correlations for
school burnout at Time 1 and Time 2 showed that the difference between
the intraclass correlations was marginally significant (estimate of differ-
ence = 0.16, standard error = 0.11, p < .10). In other words, these results
020 kiuru (23-55) 2/4/08 1:14 PM Page 43

Peer Group Socialization and Selection 43

suggest that the members of peer groups increasingly tend to resemble each
other across time (i.e., an increase in peer group homogeneity). The results
provide further evidence for peer group influence.

Peer Group Selection


Our next research question was to examine whether peer group selection
would explain the similarity between peer group members in their school
burnout. We assumed that peer group selection would operate mainly when
adolescents form new peer groups.
The first set of analyses investigating the role of peer selection in peer
group similarity were carried out for new peer groups (N= 30) formed
between the two measurements. The analyses were started by examining
how the variance in school burnout at Time 1 and Time 2 and its predictors
at Time 1, that is, academic achievement and gender, divided into the
between peer group and within peer group components by using the peer
grouping at Time 2 as a clustering variable. Examination of the intraclass
correlations showed that 0% of the variance in school burnout at Time 1 and
17% at Time 2 was due to the similarity between peer group members. Nei-
ther of these between peer group level variances in school burnout were sta-
tistically significant (see Table 2). Because using the peer grouping
measured at Time 2 yielded no significant variance at the peer group level
at either of the measurements of school burnout, no further MLGM analy-
ses were carried out for peer group selection. The fact that there was no sig-
nificant peer group similarity (i.e., between-peer group variation) in school
burnout among the new peer groups either at Time 2 or at Time 1 prior to
the formation of the new peer groups suggested that peer selection was not
responsible for peer group similarity in school burnout. In other words,
adolescents did not select their peer groups on the basis of similarity in
school burnout. Because no evidence of peer group selection effect was
found in the case of school burnout, the role of predictors, that is, academic
achievement and gender, in peer groups’ school burnout was not examined.
To analyze peer group selection further, the additional analyses were
carried out by comparing the adolescents belonging to stable groups (N =
248) to those adolescents who changed their peer groups between the two
measurements (N = 128). These analyses were carried out in the following
steps. First, the scores for school burnout (Time 1) were standardized
within the peer groupings at Time 1. Then, absolute values were taken
from these standardized scores. Next, by using the t-test for independent
samples, we examined whether the adolescents who changed their peer
group between two measurements differed more from their initial group
020 kiuru (23-55) 2/4/08 1:14 PM Page 44

44 Merrill-Palmer Quarterly
than those adolescents who stayed in the same group. No significant dif-
ferences in similarity with the initial peer group were found between the
adolescents who remained in the same group and the adolescents who
changed their group.
Next, we examined whether the adolescents who changed their peer
group resembled the peer groups at Time 2 more than their peer groups at
Time 1, that is, whether they had moved to new peer groups to which they
had greater similarity than they had with their initial peer group. To exam-
ine this, the scores for school burnout at Time 2 were also standardized
within the peer groupings at Time 2. By using the t-test for paired samples
we examined whether the absolute values of the standardized scores for
school burnout within peer groups at Time 1 differed from those standard-
ized within peer groups at Time 2 among those adolescents who had
changed their peer groups between measurements. No statistically signifi-
cant results were found.

Discussion
Research on adolescents’ peer groups has shown that peer group members
are similar with respect to many characteristics and behaviors in academic
settings, such as academic achievement and learning motivation (e.g., Chen
et al., 2003; Kindermann et al., 1996; Ryan, 2001). This similarity has been
explained by peer influence and selection (e.g., Ryan, 2001). However, no
prior studies have examined the role of peer group influence and that of
peer selection in the same study in relation to feelings of burnout at school.
This was the aim of the present study. The results showed that it was peer
influence that contributed to the peer group similarity in school burnout. No
evidence was found for the effect of peer group selection, although this may
be due to the relatively short follow-up period and small sample size. The
results showed further that peer groups characterized by high academic
achievement protected group members against an increase in school
burnout during the final term of comprehensive school.

Peer Group Influence and Selection in Adolescents’ School Burnout


The results of the present study showed, first, that the members of adoles-
cents’ peer groups were somewhat similar in terms of their school burnout.
These results expand previous research by suggesting that alongside aca-
demic achievement and motivation (Kindermann et al., 1996; Ryan, 2001)
and externalizing (Cairns & Cairns, 1994; Kiesner et al. 2003; Espelage et
al., 2003) and internalizing problem behaviors (Hogue & Steinberg, 1995),
020 kiuru (23-55) 2/4/08 1:14 PM Page 45

Peer Group Socialization and Selection 45

adolescents’ peer groups are also similar in terms of how their members
think and feel about school, that is, about their school burnout.
The main aim of the present study was, however, to examine to what
extent peer group similarity in school burnout is due to peer group influence
and what extent to peer group selection. The results of the study suggested
that it was peer group influence that is responsible for peer group similarity in
school burnout. The results showed, for example, that peer group members
shared similar changes in school burnout over time. They also became more
similar across time. These results are in accordance with those found in some
other characteristics, such as academic achievement and motivation (Ryan,
2001), initiation of smoking (Ennett & Bauman, 1994), aggressive behavior
(Cairns & Cairns, 1994; Espelage et al., 2003), and internalizing distress
(Hogue & Steinberg, 1995). There are several processes, such as corumina-
tion (Rose, 2002), observational learning, and desires or pressures to conform
(e.g., Bandura, 1977; Berndt, 1999; Suls & Wheeler, 2000), that may be
responsible for peer group influence on adolescents’ school burnout. Previ-
ous research has shown, for example, that the members of peer groups typi-
cally discuss important matters in their lives (Malmberg, 1996). Attitudes and
feelings concerning school might be assumed to be among such experiences.
Similarly, adolescents may coruminate (Rose, 2002) about their school-
related stress and workload in their peer groups. This may contribute to
higher levels of school burnout, including cynical attitudes toward school and
a sense of inadequacy as a student. Similarly, if some members of a peer
group feel overwhelmed by school demands, they may communicate their
school burnout to other group members, thereby increasing the latter’s likeli-
hood of feeling the same. Peer group members may also reciprocally rein-
force and increase each other’s burnout in their mutual communication.
Conformity pressures may operate less directly in the case of school burnout.
Although it is unlikely that a peer group overtly expects school burnout from
its members, it is possible that an increase or a decrease in burnout shared in
the peer group results from unconscious tendencies to conform to the atti-
tudes and feelings typical of a peer group.
The results of the present study did not, however, provide evidence for
the importance of peer selection in the peer group similarity in school
burnout. These results are different from some previous findings according
to which peer group selection operates in academic outcomes (Cairns &
Cairns, 1994; Cohen, 1977; Kindermann et al., 1996) and in internalizing
distress (Hogue & Steinberg, 1995). There are, however, at least three possi-
ble explanations for the differences between the findings of the present study
concerning school burnout and those of prior research concerning academic
outcomes and distress. The first possible explanation for the lack of a peer
020 kiuru (23-55) 2/4/08 1:14 PM Page 46

46 Merrill-Palmer Quarterly
group selection effect in school burnout found in the present study is that the
time period over which adolescents were followed up was relatively short.
Had the follow-up been longer, the selection effect may have appeared. The
fact that peer group members showed some similarity in their school
burnout already at Time 1 suggests that some peer group selection might
have taken place before the present study was carried out. Second, the
design of the present study only allowed for the examination of initial selec-
tion among new peer groups formed between measurements. The fact that
the number of new peer groups was relatively small decreased the power of
detecting a peer group selection effect. Consequently, there is an evident
need for future studies to examine initial peer group selection among adoles-
cents during the re-formation of peer groups, for example, after a school
transition. It is also possible that initial peer group selection may be a more
powerful form of selection than the kind of selection that directs changes in
group memberships among existing peer groups. Adolescents who change
their peer groups may also have different motives for changing their peer
group compared to adolescents facing initial peer group selection: some of
them may join a group that they admire, and some may be forced to join a
less preferred peer group if they are no longer tolerated in their original peer
group. Third, the results on peer group selection may vary substantially
across attributes and behaviors (e.g., Hartup & Stevens, 1997). For example,
school burnout, which represents a lack of fit between the student’s internal
resources for schoolwork and external demands, differs in many ways from
academic achievement and problem behaviors (Hogue & Steinberg, 1995;
Kindermann et al., 1996; Ryan, 2001). It is possible that peer group selection
is more likely to occur on the basis of characteristics that are evident in indi-
vidual behaviors, whereas it is less likely to take place on the basis of their
subjective evaluations, such as school burnout.
Overall, there is an evident need to focus future research on a variety of
microprocesses that are responsible for peer group influence and selection.
Such research may require observational studies in adolescents’ peer groups,
in-depth interviews to gain adolescents’ views of the operant processes, the
modeling of changes in peer networks after school transitions, and even
intervention studies aiming at finding the ways to include isolated individu-
als in peer groups.

Academic Achievement, Gender, and School Burnout


The next aim of the present study was to examine whether academic
achievement would predict the levels of and changes in school burnout and
the extent to which these associations would operate at the peer group level
020 kiuru (23-55) 2/4/08 1:14 PM Page 47

Peer Group Socialization and Selection 47

and at the individual level. The results showed that belonging to a high-
achieving peer group protected group members against the increase in
school burnout during the final term of comprehensive school: peer groups’
high overall academic achievement predicted a decrease in school burnout
among its members. This result is consistent with the studies showing that
high academic achievement predicts high emotional well-being and pro-
tects against maladjustment (Gerard & Buehler, 2004; Seroczynski, Cole,
& Maxwell, 1997; Crystal et al., 1994). The present study adds to the previ-
ous literature by showing that these processes also occur at the peer group
level. There are several possible explanations for this. First, high-achieving
adolescents belonging to the same peer group may support and reinforce
each other’s academic success. For example, group members may encour-
age each other’s academic achievement by providing positive feedback
after academic success, marked, for example, by good grades (Ryan, 2001).
This may then contribute to the positive attitudes of the group toward
school. Second, adolescents belonging to a high-achieving peer group may
observe and imitate each other’s high commitment to schoolwork, which
may decrease their school burnout (Bandura, 1977; Suls & Wheeler, 2000;
see also Wentzel, Barry, & Caldwell, 2004). Finally, a group atmosphere
encouraging high academic ambitions can also be strengthened via group
discussions about the meaning of schoolwork. Berndt, Laychak, and Park
(1990) have shown that group negotiation and discussion influence the
extent to which adolescents spend time pursuing academic activities.
Conversely, the results of the present study also showed that a peer
group characterized by low academic achievement promoted an increase in
school burnout among the group members during the final term of compre-
hensive school. There are at least three possible explanations for this result.
First, spending time in a peer group whose members suffer from academic
problems and yet feel pressure to perform better may promote an increase
in school burnout. For example, previous research has shown that problem
behaviors have a tendency to increase when problem-behaving adolescents
spend time together (e.g., Dishion, Spracklen, Andrews, & Patterson, 1996;
Kiesner, Cadinu, Poulin, & Bucci, 2002; Urberg et al., 1997). This may also
be the case in school burnout: as a result of frequent interactions between
group members suffering from academic difficulties and distress, overall
school burnout may increase in the peer group (see also Hogue & Stein-
berg, 1995). Different group processes, such as group communication and
conformity tendencies, are likely to play a contributory role. Second, previ-
ous research has shown that problem behaviors tend to cluster or co-occur
among adolescents (e.g., Bryant, Schulenberg, O’Malley, Bachman, &
Johnston, 2003; Donovan & Jessor, 1985). It may be that adolescents with
020 kiuru (23-55) 2/4/08 1:14 PM Page 48

48 Merrill-Palmer Quarterly
multiple problems are particularly vulnerable to school burnout because
other problems subtract from the resources available to their schoolwork.
Finally, it is also possible that cross-pressures to perform better from par-
ents and teachers are particularly strong in adolescents who are members of
low-achieving groups. This may then contribute to increased school
burnout (see also Brittain, 1963; Brody & Shaffer, 1982).
The results at the individual level showed that academic achievement
was associated with school burnout: adolescents who showed low aca-
demic achievement reported more burnout over their schoolwork irrespec-
tive of their peer group. These findings are in line with those of previous
studies showing that failure at school is associated with various negative
adjustment outcomes, such as psychological stress and negative affect
(Cole et al., 1999; Crystal et al., 1994; Seroczynski et al., 1997), and sub-
jective health complaints (Aro et al., 1987; Hurrelmann et al., 1992).
Overall, the results of the present study suggested that in addition to
directing prevention programs at individual adolescents, peer groups
should also be taken into account. Our results suggest that low-achieving
peer groups are particularly at risk for an increase in school burnout. There-
fore, identifying adolescents’ peer groups may provide a starting point
toward improving their well-being at school.
The final aim of the present study was to examine whether gender
would predict school burnout and changes in it. The results at the peer
group level showed that peer groups composed mainly of girls showed
more burnout than groups composed mainly of boys. Academic achieve-
ment was also higher among girls’ peer groups. These results are in accor-
dance with previous studies that have shown that compared to boys, girls
typically perform better at school (Dwyer & Johnson, 1997; Pomerantz et
al., 2002) but are also more vulnerable to experiencing stress and having
internalizing symptoms (Hoffmann et al., 2004; Leadbeater, Blatt, & Quin-
lan, 1995; Nolen-Hoeksema & Girgus, 1994). Girls have also been found to
attribute more importance to academic success than boys (Berndt & Miller,
1990; Murberg & Bru, 2004). Consequently, girls may fear more academic
failure and worry more about not meeting high-achievement standards.
These tendencies may have lead to higher levels of school burnout in girls’
peer groups compared to boys’ groups. In her recent work, Rose (2002)
found that girls have a higher tendency to coruminate about their problems
and negative feelings compared to boys. This may also provide an explana-
tion for relatively high levels of school burnout in girls’ peer groups com-
pared those of boys evidenced in the present study.
The results of the present study also show indirect effect from gender to
the rate of change in burnout via academic achievement: peer groups com-
020 kiuru (23-55) 2/4/08 1:14 PM Page 49

Peer Group Socialization and Selection 49

posed mainly of boys tend to have lower academic achievement than peer
groups composed mainly of girls, which in turn predicts the risk for an
increase in school burnout during the final term of comprehensive school.
Girls’ peer groups, in turn, have higher levels of both academic achieve-
ment and school burnout, and their burnout levels did not increase as much
as that of boys at the end of comprehensive school. One possible explana-
tion for this result is that a ceiling effect operated among the girls: since
girls’ school burnout levels were already relatively high at the beginning of
the present study, an increase in school burnout over the course of the final
term of comprehensive school was less likely among them than among
boys whose burnout levels were lower at the beginning of the follow-up.

Limitations
At least six limitations should be taken into account in any effort to general-
ize the findings of the present study. First, peer group influence and selec-
tion were examined over a relatively short time period. Consequently, there
is an evident need in the future for studies on the role of peer group influ-
ence and selection in adolescents’ academic adjustment across longer time
periods. Second, peer groups were studied only within schools. It has been
suggested, however, that the sole use of school-based data may underesti-
mate the impact of peer relations, as adolescents are likely to have friends
and peer groups outside school as well (e.g., Kiesner et al., 2003). It can be
assumed, however, that in-school peers are those most influential for school
burnout because they share similar school experiences. Third, participants
were restricted to belonging to only one peer group. There is evidence,
however, that adolescents’ can belong to several peer groups simultane-
ously (e.g., Brown, 2004). In the present study, we examined only the peer
groups that contained most of the adolescents’ peer nominations (i.e., the
primary peer groups). It is possible that other peer groups are also influen-
tial in relation to adolescents’ school burnout. Fourth, the method used to
measure peer groups allowed only three peer nominations, which may have
artificially restricted the size of the peer groups. It is possible that due to
this procedure, some of the adolescents’ peers were not nominated. How-
ever, allowing only three peer nominations emphasizes close peer choices
and therefore captures the most important peer group members. Fifth, mul-
tilevel latent growth models were estimated using only two time points.
The use of only two measurement points in latent growth modeling did not
allow us to study the shape of change in school burnout over time; however,
it allowed us to estimate the amount of change in school burnout (Duncan et
al., 1999; Rogosa et al., 1982). In other words, we were able to examine
020 kiuru (23-55) 2/4/08 1:14 PM Page 50

50 Merrill-Palmer Quarterly
whether peer group members share similar changes in school burnout, that
is, are influenced by each other. Finally, the number of peer groups in which
we were able to examine peer selection was substantially smaller than the
number of peer groups in which peer influence was investigated. This may
have impacted some of our interpretations.

Conclusions
The present study showed that the members of adolescents’ peer groups
were similar in terms of school burnout. The results provided also evidence
for the effect of peer influence, but not enough new peer groups were avail-
able to allow conclusions about peer selection. High achievement in the
peer group was found to protect the group members against an increase in
school burnout. Our results suggest that special attention should be paid to
the role of peer groups in adolescents’ lack of well-being at school.

References
Aro, H., Paronen, O., & Aro, S. (1987). Psychosomatic symptoms among 14–26
year old Finnish Adolescents. Social Psychiatry, 22, 171–176.
Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice Hall.
Benenson, J. F. (1990). Gender differences in social networks. Journal of Early
Adolescence, 10, 472–495.
Berndt, T. J. (1999). Friends’ influence on students’ adjustment to school. Educa-
tional Psychologist, 34, 15–28.
Berndt, T. J., Laychak, A. E., & Park, K. (1990). Friends’ influence on adolescents’
academic achievement motivation: An experimental study. Journal of Educa-
tional Psychology, 82, 664–670.
Berndt, T. J., & Miller, K. E. (1990). Expectancies, values, and achievement in jun-
ior high school. Journal of Educational Psychology, 82, 319–326.
Brittain, C. V. (1963). Adolescent choices and parent-peer cross-pressures. Ameri-
can Sociological Review, 28, 385–391.
Brody, G. H., Shaffer, D. R. (1982). Contributions of parents and peers to children’s
moral socialization. Developmental Review, 2, 31–75.
Brown, B. B. (1989). The role of peer groups in adolescents’ adjustment to second-
ary school. In T. J. Berndt & G. W. Ladd (Eds.), Peer relationships in child
development (pp. 188–215). Oxford, UK: Wiley.
Brown, B. B. (1990). Peer groups and peer cultures. In S. S. Feldman & G. R.
Elliott (Eds.), At the threshold: The developing adolescent (pp. 171–196).
Cambridge: Harvard University Press.
020 kiuru (23-55) 2/4/08 1:14 PM Page 51

Peer Group Socialization and Selection 51

Brown, B. B. (2004). Adolescents’ relationships with peers. In R. M. Lerner &


L. D. Steinberg (Eds.), Handbook of adolescent psychology (2nd ed., pp.
363–394). Hoboken, NJ: Wiley.
Bryant, A. L., Schulenberg, J. E., O’Malley, P. M., Bachman, J. G., & Johnston,
L. D. (2003). How academic achievement, attitudes, and behaviours relate to
the course of substance use during adolescence: A 6-year, multiwave national
longitudinal study. Journal of Research on Adolescence, 13, 361–397.
Bukowski, W. M., Newcomb, A. F., & Hartup, W. W. (Eds.). (1996). The company
they keep: Friendship in childhood and adolescence. Cambridge: Cambridge
University Press.
Cairns, R. B., & Cairns, D. B. (1994). Lifelines and risks: Pathways of youth in our
time (pp. 90–129). Cambridge: Cambridge University Press.
Chen X., Chang L., & He Y. (2003). The peer group as a context: Mediating and
moderating effects on relations between academic achievement and social
functioning in Chinese children. Child Development, 74, 710–727.
Cohen, J. M. (1977). Sources of peer group homogeneity. Sociology of Education,
50, 227–241.
Coie, J. D., Dodge, K. A., & Coppotelli, H. (1982). Dimensions and types of social
status: A cross-age perspective. Developmental Psychology, 18, 557–570.
Cole, D. A., Peeke, L., Dolezal, S., Murray, N., & Canzoniero, A. (1999). A longitu-
dinal study of negative affect and self-perceived competence in young adoles-
cents. Journal of Personality and Social Psychology, 77, 851–862.
Crystal, D. S., Chen, C., Fuligni, A. J., Stevenson, H. W., Hsu, C.-C., Ko, H.-J., et
al. (1994). Psychological maladjustment and academic achievement: A cross-
cultural study of Japanese, Chinese, and American high school students. Child
Development, 65, 738–753.
Deci, E. L., Vallerand, R. J., Pelletier, L. G., & Ryan, R. M. (1991). Motivation and
education: The self-determination perspective. Educational Psychologist, 26,
325–346.
Degbirmenciogblu, S. M., Urberg, K. A., Tolson, J. M., & Richard, P. (1998). Adoles-
cent friendship networks: Continuity and change over the school year. Merrill-
Palmer Quarterly, 44, 313–337.
Demerouti, E., Bakker, A. B., Nachreiner, F., & Schaufeli, W. B. (2001). The job
demands-resources model of burnout. Journal of Applied Psychology, 86,
499–512.
Dishion, T. J., Spracklen, K. M., Andrews, D. W., & Patterson, G. R. (1996).
Deviancy training in male adolescent friendships. Behavior Therapy, 27,
373–390.
Donovan, J. E., & Jessor, R. (1985). Structure of problem behavior in adolescence
and young adulthood. Journal of Consulting and Clinical Psychology, 53,
890–904.
020 kiuru (23-55) 2/4/08 1:14 PM Page 52

52 Merrill-Palmer Quarterly
Duncan, T. E., Duncan, S. C., Alpert, A., Hops, H., Stoolmiller, M., Muthén, B.
(1997). Latent variable modeling of longitudinal and multilevel substance use
data. Multivariate Behavioral Research, 32, 275–318.
Duncan, T. E., Duncan, S. C., Stycker, L. A., Li, F., & Alpert, A. (1999). An intro-
duction to latent variable growth modelling. Mahwah, NJ: Erlbaum.
Dwyer, C. A., & Johnson, L. M. (1997). Grades, accomplishments, and correlates.
In W. W. Willingham & N. S. Cole (Eds.), Gender and fair assessment (pp.
127–156). Mahwah, NJ: Erlbaum.
Eccles, J. S. (1983). Expectancies, values, and academic behaviours. In J. T. Spence
(Ed.), Achievement and achievement motives (pp. 75–146). San Francisco:
Freeman.
Eccles, J. S. (2004). Schools, academic motivation, and stage-environment fit. In
R. M. Lerner & L. D. Steinberg (Eds.), Handbook of adolescent psychology
(2nd ed., pp. 125–153). Hoboken, NJ: Wiley.
Ennett, S. T., & Bauman, K. E. (1994). The contribution of influence and selection
to adolescent peer group homogeneity: The case of adolescent cigarette smok-
ing. Journal of Personality and Social Psychology, 67, 653–663.
Espelage D. L., Holt M. K., & Henkel R. R. (2003). Examination of peer-group
contextual effects on aggression during early adolescence. Child Develop-
ment, 74, 205–220.
Ge, X., Lorenz, F. O., Conger, R. D., Elder, G. H., & Simons, R. L. (1994). Trajec-
tories of stressful life events and depressive symptoms during adolescence.
Developmental Psychology, 30, 467–483.
Gerard, J. M., & Buehler, C. (2004). Cumulative environmental risk and youth mal-
adjustment: The role of youth attributes. Child Development, 6, 1832–1849.
Harter, S. (1996). Teacher and classmate influences on scholastic motivation, self-
esteem, and level of voice in adolescents. In J. Juvonen & K. R. Wentzel
(Eds.), Social motivation: Understanding children’s school adjustment (pp.
11–42). New York: Cambridge University Press.
Hartup, W. W., & Stevens, N. (1997). Friendships and adaptation in the life course.
Psychological Bulletin, 121, 355–370.
Heck, R. H. (2001). Multilevel modeling with SEM. In G. A. Marcoulides & R. E.
Schumacker (Eds.), New developments and techniques in structural equation
modelling (pp. 89–127). Mahwah, NJ: Erlbaum.
Hinde, R. A. (1987). Individuals, relationships and culture: Links between ethology
and the social sciences. Cambridge: Cambridge University Press.
Hoffmann, M. L., Powlishta, K. K., & White, K. J. (2004). An examination of gen-
der differences in adolescent adjustment: The effect of competence on gender
role differences in symptoms of psychopathology. Sex Roles, 50, 795–810.
Hogue, A.,& Steinberg, L. (1995). Homophily of internalized distress in adolescent
peer groups. Developmental Psychology, 31, 897–906.
020 kiuru (23-55) 2/4/08 1:14 PM Page 53

Peer Group Socialization and Selection 53

Holopainen, L., & Savolainen, H. (2005). Unpublished raw data. University of


Joensuu and University of Jyväskylä, Finland.
Hurrelmann, K., Engel, U., & Weidman, J. C. (1992). Impacts of school pressure,
conflict with parents, and career uncertainty on adolescents stress in the Fed-
eral Republic of Germany. International Journal of Adolescence and Youth, 4,
33–50.
Iacovides, A., Fountoulakis, K. N., Kaprinis, St., & Kaprinis, G. (2003). The rela-
tionship between job stress, burnout and clinical depression. Journal of Affec-
tive Disorders, 75, 209–221.
Jessor, R. (1991). Risk behaviour in adolescence: A psychosocial framework for
understanding and action. Journal of Adolescent Health, 12, 597–605.
Jose, P. E., & Ratcliffe, V. (2004). Stressor frequency and perceived intensity as
predictors of internalizing symptoms: Gender and age differences in adoles-
cence. New Zealand Journal of Psychology, 33, 145–154.
Kalimo, R., Pahkin, K., Mutanen, P., & Toppinen-Tanner, S. (2003). Staying well or
burning out at work: Work characteristics and personal resources as long-term
predictors. Work & Stress, 17, 109–122.
Kandel, D. B. (1978). Homophily, selection, and socialization in adolescent friend-
ships. American Journal of Sociology, 84, 427–436.
Kasen, S., Johnson, J., & Cohen, P. (1990). The impact of school emotional climate
on student psychopathology. Journal of Abnormal Child Psychology, 18,
165–177.
Kiesner, J., Cadinu, M., Poulin, F., & Bucci, M. (2002). Group identification in
early adolescence: Its relations with peer adjustment and its moderator effect
on peer influence. Child Development, 73, 196–208.
Kiesner J., Poulin, F., & Nicotra E. (2003). Peer relations across contexts: Individual-
network homophily and network inclusion in and after school. Child Develop-
ment, 74, 1328–1343.
Kindermann, T. A., McCollam, T. L., & Gibson, E., Jr. (1996). Peer networks and
students’ classroom engagement during childhood and adolescence. In J. Juvo-
nen & K. R. Wentzel (Eds.), Social motivation: Understanding children’s
school adjustment (pp. 279–312). New York: Cambridge University Press.
Leadbeater, B. J., Blatt, S. J., & Quinlan, D. M. (1995). Gender-linked vulnerabili-
ties to depressive symptoms, stress, and problem behaviours in adolescents.
Journal of Research on Adolescence, 5, 1–29.
Lee, R., & Ashforth, B. (1996). A meta-analytic examination of the correlates of the
three dimensions of job burnout. Journal of Applied Psychology, 81, 123–133.
Magnusson, D., & Stattin, H. (1998). Person-context interaction theories. In
R. Lerner (Ed.) & W. Damon (Series Ed.). Handbook of child psychology: Vol
1. Theoretical models of human development (pp. 685–740). New York: Wiley.
020 kiuru (23-55) 2/4/08 1:14 PM Page 54

54 Merrill-Palmer Quarterly
Malmberg, L.-E. (1996). How do Finnish students prepare for their future in three
school types? The relation between content of plans, information gathering,
and self-evaluations. British Journal of Educational Psychology, 66, 457–469.
Masi, G., Sbrana, B., Poli, P., Tomaiuolo, F., Favilla, L., & Marcheschi, M. (2000).
Depression and school functioning in non-referred adolescents: A pilot study.
Child Psychiatry and Human Development, 30, 161–171.
Maslach, C., Schaufeli, W. B., & Leiter, M. P. (2001). Job burnout. Annual Review
of Psychology, 52, 397–422.
Murberg, T. A., & Bru, E. (2004). School-related stress and psychosomatic symp-
toms among Norwegian adolescents. School Psychology International, 25,
317–332.
Muthén, B. O. (1991). Multilevel factor analysis of class and student achievement
components. Journal of Educational Measurement, 28, 338–354.
Muthén, L., & Muthén, B. O. (1998–2006). Mplus Version 4.1. http://www
.statmodel.com
Nolen-Hoeksema, S., & Girgus, J. S. (1994). The emergence of gender differences
in depression during adolescence. Psychological Bulletin, 115, 424–443.
Näätänen, P., Aro, A., Matthiesen, S., & Salmela-Aro, K. (2003). Bergen Burnout
Indicator 15. Helsinki, Finland: Edita.
Pomerantz, E. M., Altermatt, E. R., Saxon, J. L. (2002). Making the grade but feel-
ing distressed: Gender differences in academic performance and internal dis-
tress. Journal of Educational Psychology, 94, 396–404.
Rogosa, D., Brandt, D., & Zimowski, M. (1982). A growth curve approach to the
measurement of change. Psychological Bulletin, 92, 726–748.
Rose, A. J. (2002). Co-rumination in the friendships of girls and boys. Child Devel-
opment, 73, 1830–1843.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ:
Princeton University Press.
Rubin, K. H., Bukowski, W., & Parker, J. G. (1998). Peer interactions, relation-
ships, and groups. In N. Eisenberg (Ed.) & W. Damon (Series Ed.), Handbook
of child psychology: Vol 3. Social, emotional, and personality development
(pp. 619–700). New York: Wiley.
Ryan, A. M. (2001). The peer group as a context for the development of young ado-
lescent motivation and achievement. Child Development, 72, 1135–1150.
Salmela-Aro, K., Niemivirta, M., & Nurmi, J.-E. (2003). Kuopio School Transition
(KST) study (ongoing). University of Helsinki and University of Jyväskylä,
Finland.
Salmela-Aro, K., & Näätänen, P. (2005). BBI-10: Nuorten koulu-uupumus-
menetelmä [BBI-10: Adolescents’ school burnout method]. Helsinki, Finland:
Edita.
020 kiuru (23-55) 2/4/08 1:14 PM Page 55

Peer Group Socialization and Selection 55

Salmela-Aro, K., Näätänen, P., & Nurmi, J.-E. (2004). The role of work-related per-
sonal projects during two burnout interventions: A longitudinal study. Work &
Stress, 18, 208–230.
Salokangas, R. K. R., Stengård, E., & Poutanen, O. (1994). DEPS—Uusi väline
depression Seulontaan [DEPS—An instrument for screening depression].
Duodecim, 110, 1141–1148.
Saunders, J., Davis, L., Williams, T., & Williams, J. H. (2004). Gender differences
in self-perceptions and academic outcomes: A study of African American high
school students. Journal of Youth and Adolescence, 33, 81–90.
Schaufeli, W. B., Martìnez, I. M., Pinto, A. M., Salanova, M., & Bakker, A. B.
(2002). Burnout and engagement in university students: A cross-national
study. Journal of Cross-cultural Psychology, 33, 464–481.
School statistics (2003). http://www.stat.fi/til/khak/2005/khak_2005_2006–12-21
_tie_001.html. Central Statistical Office of Finland. Helsinki, Finland.
Seroczynski, A. D., Cole, D. A., & Maxwell, S. E. (1997). Cumulative and compen-
satory effects of competence and incompetence on depressive symptoms in
children. Journal of Abnormal Psychology, 106, 586–597.
Suls, J., & Wheeler, L. (2000). Handbook of social comparison: Theory and
research. New York: Plenum.
Terry, R., Coie, J. D. (1991). A comparison of methods for defining sociometric sta-
tus among children. Developmental Psychology, 27, 867–880.
Toppinen-Tanner, S., Ojajärvi, A., Väänänen, A., Kalimo, R., & Jäppinen, P.
(2005). Burnout as a predictor of medically certified sick-leave absences and
their diagnosed causes. Behavioral Medicine, 31, 18–27.
Urberg, K. A., Degbirmenciogblu, S. M., & Pilgrim, C. (1997). Close friend and
group influence on adolescent cigarette smoking and alcohol use. Develop-
mental Psychology, 33, 834–844.
Urberg, K. A., Degbirmenciogblu, S. M., Tolson, J. M. (1998). Adolescent friendship
selection and termination: The role of similarity. Journal of Social and Per-
sonal Relationships, 15, 703–710.
Urberg, K. A., Degbirmenciogblu, S. M., Tolson, J. M., & Halliday-Scher, K. (1995).
The structure of adolescent peer networks. Developmental Psychology, 31,
540–547.
Urberg, K. A., Luo, Q., Pilgrim, C., & Degbirmenciogblu, S. M. (2003). A two-stage
model of peer influence in adolescent substance use: Individual and relation-
ship-specific differences in susceptibility to influence. Addictive Behavior, 28,
1243–1256.
Wentzel, K. R., Barry C. M., & Caldwell, K. A. (2004). Friendships in middle
school: Influences on motivation and school adjustment. Journal of Educa-
tional Psychology, 96, 195–203.

You might also like