Content uploaded by Niobe Way
Author content
All content in this area was uploaded by Niobe Way
Content may be subject to copyright.
ORIGINAL PAPER
Students’ Perceptions of School Climate During the Middle School
Years: Associations with Trajectories of Psychological and
Behavioral Adjustment
Niobe Way Æ Ranjini Reddy Æ Jean Rhodes
Published online: 30 October 2007
! Springer Science+Business Media, LLC 2007
Abstract A cross-domain latent growth curve model was
used to examine the trajectories of change in student per-
ceptions of four critical dimensions of school climate (i.e.,
teacher support, peer support, student autonomy in the
classroom, and clarity and consistency in school rules and
regulations) among 1,451 early adolescents from the
beginning of sixth through the end of eighth grade; and the
effects of such trajectories on the rate of change in psy-
chological and behavioral adjustment. Findings indicated
that all of the dimensions of perceived school climate
declined over the 3 years of middle school. Furthermore,
declines in each of the dimensions of perceived school
climate were associated with declines over time in psy-
chological and behavioral adjustment. Moreover, the
direction of effects between each dimension of perceived
school climate and psychological or behavioral adjustment
were often unidirectional rather than bi-directional,
underscoring the role of perceived school climate in the
psychological and behavioral health of early adolescents.
Gender and socioeconomic class differences in these pat-
terns are noted.
Keywords School climate ! Adolescent adjustment !
Middle school ! Latent growth curve modeling
Introduction
Middle schools play an important role in the development
of adolescents. They provide a context in which students
learn, implicitly and explicitly, about themselves and
relationships, and about how to navigate the turbulent
waters of early adolescence. There is considerable evi-
dence that the interpersonal, organizational, and
instructional ‘‘climate’’ of middle schools strongly influ-
ences students’ adjustment across multiple domains
(Eccles and Roeser 1999; Roeser et al. 1998, 2000 ; Ku-
perminc et al. 1997; Skinner and Wellborn 1997). Mid dle
school climates often challenge early adolescents’ adap-
tive capacities and account for at least some of the
difficulties that arise during this developmental stage
(Roeser et al. 1998). Despite growing interest in students’
perceptions of their middle schools, and evidence for the
influence of such perceptions on psychological and
behavioral adjustment, our understanding of this construct
and its association with adjustmen t rests on a relatively
small base of empirical findings (Bachman and O’Malley
1986; Brand et al. 2003; Hoge et al. 1990; Kuperminc
et al. 1997; Roeser and Eccles 1998; Roeser et al. 1998;
Way and Robinson 2003). Most studies have focused on
how perceptions of school climate shape academic
adjustment (Bryk and Driscoll 1988; Griffith 1995;
Skinner et al. 1990), giving far less consideration to its
emotional and behavioral consequences. This is surprising
given that the emergence of depressive symptoms, low-
ered self-esteem, and behavioral difficulties often
coincides with the middle-school years (Blyth et al. 1983;
Eccles et al. 1999; Seidman et al. 1994).
Moreover, studies of school climate have predominantly
focused on how students’ perceptions change during the
transition from elementary to middle school, as opposed to
N. Way (&)
Department of Applied Psychology, New York University,
Steinhardt School of Education, East Building, Suite 400, 239
Greene Street, New York, NY 10003, USA
e-mail: nw4@nyu.edu
R. Reddy ! J. Rhodes
University of Massachusetts, Boston, MA, USA
123
Am J Community Psychol (2007) 40:194–213
DOI 10.1007/s10464-007-9143-y
the changi ng patterns of students’ perceptions during
middle school itself. Such research has also typically
treated students’ perceptions of school climate as static,
baseline predictors of other outcomes variables (see Brand
et al. 2003; Roeser and Eccles 1998; Kupermin c et al.
1997; Way and Robinson 2003). Yet subtle factors, such as
shifts in student–teacher relationships or peer relationships,
can easily influence adolescents’ perceptions of their
schools. These changes in perceptions of school climate, in
turn, are likely to have various psychological and behav-
ioral implications.
In this study, we used cross-domain latent growth curve
models to examine trajectories of change in students’
perceptions of four critical components of school climate
(i.e., teacher support, peer support, student autonomy in the
classroom, and clarity and consistency in school rules)
among early adolescents from the beginning of sixth
through the end of eighth grade, and the effects of such
trajectories on the rate of change in psychological and
behavioral adjustment. This investigation is a secondary
analysis of a data set drawn from the Illinois Center for
Prevention Research (see Felner et al. 1997).
Theoretical and Empirical Background
Theorists and researchers have underscored the importance
for middle school students’ wellbeing of four aspects of
school climate: (1) the nature of relations hips between
teachers and students; (2) the nature of relationships
between students; (3) the extent to which student autonomy
is allowed in the decision-making process; and (4) the
extent to which the school provides clear, consistent, and
fair rules and regulations (Bachman and O’Malley 1986;
Connell and Wellborn 1991; Epstein and Karweit 1983;
Kuperminc et al. 1997; Haynes and Emmons 1994; Roeser
and Eccles 1998; Roeser et al. 1998; Skinner and Wellborn
1997). These elements of the school climate (i.e., related-
ness, opportunities for autonomy, and clarity/consistency)
are considered particularly important for students during
the middle school years because they match with the
developmental needs of early adolescents (Eccles and
Midgley 1989; Eccles et al. 1993; Roeser et al. 1998;
Connell and Wellborn 1991). Researchers have noted that
if there is not a ‘‘person-environment fit’’ or the school is
not perceived by the students as supporting their need for
relatedness, autonomy, and consistency,
1
their psycholog-
ical and behavioral health will be at risk (Connell and
Wellborn 1991; Eccles et al. 1993). An important compo-
nent of these models linking school climate with
psychological and behavioral adjustment is that it is the
adolescents’ perceptions or experiences of the school
environment rather than a more objective account of the
school that is most relevant for understanding adolescents’
adjustment and wellbeing (Connell and Ryan 1987; Con-
nell and Wellborn 1991; Eccles et al. 1993).
The empirical research examining the association
between students’ perceptions of the teacher support, peer
support, student autonomy, and clarify and consistency in
school rules on psychological or behavior al adjustment has
suggested that they are strongly associated with one
another (Brand et al. 2003; Kuperminc et al. 1997; Roeser
et al. 1998; Hoge et al. 1990; Way and Robinson 2003).
For example, Roeser et al. (1998) found that students’
perceptions of their schools’ learning environment (i.e.,
teacher support, and support for student autonomy) during
seventh grade significantly predicted change over time in
emotional functioning from seventh to eighth grade, even
after accounting for demographic characteristics. In other
words, improvements in emotional functioning from sev-
enth to eighth grade was significantly predicted by higher
reports of teacher support and support for student auton-
omy during the seven th grade. Similarly, Hoge et al.
(1990) found that more positive perceptions of school cli-
mate, including perceptions of teacher support, predicted
increases in global self-esteem from fall to spring during
the sixth and seventh grade, after accounting for IQ, sex,
race, single parent family, and socioeconomic status.
Others have also found significant associations between
teacher and/or peer support, opportunities for autonomy,
and clarity and consistency in school rules and students’
emotional adjustment (see Brand et al. 2003; Bachman and
O’Malley 1986; Roeser et al. 1998; Kuperminc et al. 1997;
Way and Robinson 2003).
In addition, linkages between teacher and/or peer sup-
port, opportunities for autonomy, and clarity and
consistency in school rules and behavior problems have
been reported (Brand et al. 2003; Bryk and Driscoll 1988;
Kuperminc et al. 1997; Rutter et al. 1979; Sommer 1985).
For example, Kasen et al. (1990) found that an increase in
students’ perceptions of autonomy in the classroom in
middle school over one-year period predicted decreases in
alcohol use, conduct problems, and oppositional behavior.
In their study of 188 middle schools, Brand et al. (2003)
found that students’ perceptions of different dimensions of
school climate, including teacher and peer support (and
negative peer interaction), was significantly associated with
self-esteem, depressive symptoms, delinquency, and sub-
stance use.
Taken together, these studies shed light on the range of
psychological and behavioral outcomes associated with
1
Connell and Wellborn (1991) refer to a psychological need for
sense of relatedness, autonomy, and competency among students
rather than relatedness, autonomy, and consistency. However, they
note that a sense of competency will often emerge in an environment
that is fair and consistent.
Am J Community Psychol (2007) 40:194–213 195
123
students’ perceptions of middle schools environments and
underscore the importance of schools addressing the psy-
chological needs of early adolescence (Connell and
Wellborn 1991). To a certain extent, however, the statis-
tical methods that have been most commonly employed
have masked patterns of change and adjustment over time.
These techniques, which have spanned from correlational
and mean level analyses of groups of students (e.g.,
ANOVA’s, MANOVA’s), to more complex grouping and
structural models, rest on a more static view of a period
that is typically characterized by heterogeneity and flux.
Snapshot views of this developmental phase do not ade-
quately describe the different patterns over time of
perceived school climate nor do they explain the changing
nature of psycho logical and behavioral adjustment. For
some students, the middle schools years may be marked by
ongoing patterns of success and emotional wellbeing. Ini-
tial positive reactions to the school environment might
portend ongoing opportunities for positive growth and
development. For others, however, a honeymoon period
might gradually give way to disappointment with their
middle school’s limitations. Still others might experience
transitional struggles that forecast either continued nega-
tive perceptions of the school or gradual adaptation to the
setting. Each of these scenarios is likely to predict different
levels of emotional and behavioral adjustment. The goal in
this study is to extend the research on school climate by
tracking the effects of various patterns of perceptions of
school climate over time.
Our work is not only guided by developmental theories
such as those by Eccles and colleagues (e.g., Eccles et al.
1993; Roeser et al. 1998) and motivational theories such as
those by Connell and colleagues (Connell and Wellborn
1991), but also by an integrated systems perspective that
recognizes that perceptions of school climate are closely
tied to various demographic variables such as gender and
socioeconomic status (SES). Research, for example, has
suggested that adolescent girls may experience school as a
less supportive environment than boys, stemming from
institutional gender biases, and a greater likelihood of
sexual harassment from male peers (AAUW 1992; Brown
and Gilligan 1992; Orenstein 1994; Sadker and Sadker
1994). These difficult experiences may translate into more
negative perceptions of school . In addition, since the
harassment from peers often increases during middle
school as girls reach puberty, girls’ perceptions of school
climate may be more likely than boys to decline over time
(Orenstein 1994). Furthermore, increasing levels of rela-
tional aggression among girls in middle schools has been
noted by researchers (Crick and Rose 2000; Rys and Bear
1997) and may lead girls to grow increasingly dissatisfied
with their peers—a key component of the interpersonal
dimension of school climate.
Along similar lines, lower SES students may hold more
negative views of their schools than their higher SES
counterparts, as the former groups are more likely to attend
under-funded and understaffed schools that have few
resources to create positive learning environments (Alvi-
drez and Weinstein 1999; Conchas and Noguera 2004;
Kuperminc et al. 1997; Way 1998). Researchers over the
past three decades have documented the negative percep-
tions of school evident among many low SES students
(Anyon and Wilson 1997; Fine 1991; Kozol 1992; Way
1998). Furthermore, the perceptions of school climate
among lower SES middle school students may become
more negative over time, in comparison with higher SES
students, as their ability to recognize the limits of their
school grows stronger (Conchas and Noguera 2004).
Current Study
In this study, we sought to examine four questions: (1)
How do students’ perceptions of teacher support, peer
support, opportunities for autonomy in the cla ssroom, and
clarity and consistency of school rules and regulations
change during the middle school years? (2) Does g ender or
SES explain variations in perceptions of these four
dimensions of school climate over time? (3) Are changes in
perceptions of thes e four dimensions associated with
changes in psychological and behavioral adjustment? (4)
What is the direction of effect between each dimension of
perceived school climate and psychological and behavior
adjustment? Although researchers have tested the hypoth-
esis that students’ perceptions of school climate influence
their psychological and behavioral adjustment, it is also
plausible that students’ adjustment may influence their
perceptions of their school environments. Researchers have
not, however, tested both possibilities in one study.
Based on previous research (e.g., Eccles et al. 1993), we
hypothesized that stud ents’ perceptions of the teacher and
student support, opportunities for autonomy, and consis-
tency and clarity of school rules would become more
negative during the middle school years. We also expected
that boys and higher SES students would report more
positive perceptions of the school environment and less
sharp declines over time than girls and lower SES students,
respectively. Finally, we hypothesized that perceptions of
school climate woul d influence adjustment, rather than the
reverse (see Fig. 1).
196 Am J Community Psychol (2007) 40:194–213
123
Method
Source of Data
Data were drawn from a larger ongoing longitudinal study
of the role of educational environments and state-wide
change efforts on early adolescents in middle schools.
2
The
schools that participated in this study were all members of
a statewide middle school association, which serves stu-
dents from geographically and socio-economically diverse
backgrounds. Administrators in the schools agreed to par-
ticipate in the study as part of statewide school assessment
plan. The assessment was not tied to a particular
intervention.
The original study included 2,860 students from 30
schools. The response rate ranged from 70 to 90%. We
eliminated those students who had missing information and
those who did not have complete data for all 3 years of the
study. Since we sought to understand changes during the
3 years of middle school, we also eliminated schools that
did not have grades six, seven, and eight. The current study
focused on the 1,451 middl e school students (from 22
middle schools) who had complete data from the three
consecutive years of the study, 1995–1997, when the stu-
dents were in sixth, seventh, or eighth grade.
Participants
At Time 1, the sample consisted of 1,451 early adolescents
who had begun sixth grade in 1995. Slightly over half the
sample was female (54.2%) and the majority was European
American (91%). Eligibility to receive free or reduced-
price lunches was taken as a marker of low-income status.
At the beginning of middle school, 20% of the sample
reported receiving free or reduced-price lunches. A little
over one-third of the students reported that their mothers
had com pleted college (37.4%) and over half reported that
their fathers had two or more years of college (17%) or had
completed college (41.6%). The majority of the adoles-
cents (85%) also reported living in two-parent families.
The students were drawn from 22 Midwestern schools all
housing grades six through eight. The schools represented
diverse geographical locales with half of the schools in
small town or rural areas and 40.8% situated in urban areas.
Six grade enrollment in each school ranged from 68 to 414
students.
Procedures
Data collection occurred in partic ipating schools during the
Spring of 1995, 1996, and 1997. The IR B at the partici-
pating University reviewed and approved Human Subject
procedures.
3
The informed consent of students and their
parents was obtained and confidentiality was assured.
Teachers distributed the surveys in classrooms to students,
and read them aloud to reduce disparities in reading level.
Teachers were always present during survey administra-
tion. The surveys were typically completed in two class
lo
o
h
cS
lanosrepretnI
e
ta
m
il
C
rehcaeT
tropp
uS
reeP
troppuS
rofseitinutropp
O
y
m
o
n
otuAtn
e
dut
S
fo
y
t
i
ra
lC
sel
uRloohcS
ycnetsisnoC
ssenriafdna
loohcS
l
ano
i
tcurtsn
I
etami
l
C
loohcS
la
no
i
tazinagrO
etamilC
sm
e
lborP
r
oivaheB meetsE-fleSlar
e
neG Depression
Fig. 1 Perceived middle school
environment: Theoretical model
2
We gratefully acknowledge the contribution of the researchers at
the Illinois Center for Prevention Research and of Robert D. Felner,
who served as Principal Investigator of the Alliance of Illinois Middle
Schools network evaluation. The evaluation was funded by a
Carnegie Foundation grant to the University of Illinois.
3
A two tiered consent procedure developed and approved by the
Institutional Review Board at the participating university allowed the
investigators to use passive consent procedures. The first step required
each school to create a Parent Advisory Team (PAT) that was
reflective of the children and families attending that school. The PAT
was asked to review the surveys and consent form that would
subsequently be sent home to parents for their approval. The PAT was
to consider community norms and values when reviewing the topics
covered in the surveys, and they were to determine if the consent form
accurately represent the goals and purposes of the study. Parents
serving on PAT team were also asked to serve as resources to other
parents who might have questions supporting the surveys, and copies
were kept in the school from office for parental review. Once the PAT
signed the advisory form, schools were required to send home letters
to parents and were asked to return the form if they did not want their
child to participate.
Am J Community Psychol (2007) 40:194–213 197
123
periods over a 2-day period. The schools determin ed the
class to administer the survey. Some schools administered
the survey during homeroom while others used a class
within the school day. Once surveys were completed,
surveys were collected and returned to the university for
processing.
Measures
School Climate
In order to examine school climate, we examined teacher
support, peer support, student autonomy and clarity and
consistency in school rules (see Appendix A). These four
dimensions were assessed through subscales of the Per-
ceived School Climate Scale (1993) developed by the
Center for Prevention Research and Development (CPRD)
(see Felner et al. 1997). The questionnaire is based on the
Classroom Environment Scal e (Trickett and Moos 1973), a
widely used and well-validated measure. Changes to the
Classroom Environment Scale involved rewording items to
eliminate double negatives and providing more familiar
language for colloquial terms. The Perceived School Cli-
mate Scale has been found to be a valid and reliable
measure of school climate (see Brand et al. 1994, 2003).
Teacher Support
The six items on this subscale of the Perceived School
Climate Scale indexed students’ perceptions of teacher
support in the school. Students rated on a 5-point scale
(1 = never to 5 = always) how true statements were for
their school such as teachers go out of their way to help
students and teachers take a personal interest in students.
Higher scores indicate higher levels of teacher support.
Internal consistency of the scale for the 3 years of the
present study was .74, .74, and .79, respectively.
Peer Support
Two subscales of the Perceived School Climate Scale
assessed adolescent’s perceptions of peer support: negative
peer interactions and affiliation. The negative peer inter-
actions subscale (Time 1 a = .77, Time 2 a = .77, Time 3
a = .82) and the affiliation subscale (Time 1 a = .69, Time
2 a = .70, Time 3 a = .78) assessed on a 5-point scale
(1 = never to 5 = always) how well stud ents in the school
got along with each other, enjoyed working together,
helped each other, or were mean to each other. The items
on the negative peer interaction subscale were recoded and
the composite score of the two subscales was used. Higher
scores on this dimension indicate a positive peer climate.
Internal consistency for the combined scale for the 3 years
of the present study was .77, .72, and .75, respectively.
Opportunities for Student Autonomy
This construct was assessed through the 5-item student
decision-making subscale of the Perceived School Climate
Scale. Rated on a 5-point scale (1 = never to 5 = always),
the scale assessed whether students felt that they had a say
in how things worked in school, helped decide how class
time was spent, given a chance to help make decisions, and
asked what they would like to learn about. Higher scores
imply a greater degree of student decision-making. Internal
consistency for the 3 years of the present study was .70,
.72, and .75, respectively.
Clarity and Consistency in School Rules
Two subscales of the Perceived School Climate Scale,
school structure and school harshness, were combined to
assess clarity and consistency in school rules and regula-
tions. School structure (Time 1 a = .68, Time 2 a = .75,
Time 3 a = .82) and School harshness (Time 1 a = .65,
Time 2 a = .69, Time 3 a = .75), assessed on a 5-point
scale (1 = never to 5 = always), indexed the degree to
which students p erceived clarity in the school rule structure
and their perceptions of arbitrary or punitive discipline
practices in the school. The harshness subscale was reco-
ded and then a composite subscale score was obtained.
Higher scores on this dimension implied perceptions of
clarity, consistency, and fairness in the school. Internal
consistency for the combined scale for the 3 years of the
present study was .62, .65, and .59, respectively.
Indices of Wellbeing
Behavior Problems
The Behavior Problems Scale was based on the Delin-
quency Scale of the Youth Self-Report (Achenbach and
Edelbrock 1987). The scale is a shortened version of the
original scale indexing frequency rather than severity of
problem behavio r. The scale was shortened by The CPRD
(Felner et al. 1997). The 10-item measure indexed on a 5-
point scale (1 = never to 5 = more than 12 times) how
often students, over the past 6 months, reporte d having
committed a problem behavior, including telling lies,
breaking rules at school, cutting classes or skipping school,
198 Am J Community Psychol (2007) 40:194–213
123
hitting other people, and acting mean toward others.
Internal consi stency over 3 year s for the current stud y was
a = .85, a = .88, and a = .90 respectively.
Self-esteem
Self-esteem was indexed by the six-item general self-
esteem subscale of the Self-Evaluation Questionnaire
(SEQ, DuBois and Felner 1991; DuBois et al. 1996).
DuBois et al. (1996) have reported convergent and dis-
criminant validity of the scale. Students rated statements
such as ‘‘I like being the way I am,’’ ‘‘I am happy with
myself as a person,’’ ‘‘I am as good a person as I want to
be,’’ on a 4-point scale (strongly disagree to strongly agree)
with higher ratings indicating higher levels of self-esteem.
Internal consistency of the subscale for the 3 years was
a = .83, a = .84, and a = .86, respectively.
Depressive Symptoms
Depressive symptoms were indexed with a shortened ver-
sion of the Children’s Depressive Inventory (Kovacs 1980/
1981). The sixteen-item self-report scale measures cogni-
tive, affective, and behavioral symptoms of depressive
symptoms. For each item, students selected one of three
statements (increasing in symptom severity) that best
described their feelings and ideas during the previous
2 weeks. The scale yields a single aggregate measure of
depressive symptomology reflecting an underlying unidi-
mensional construct. Lower scores indexed lower levels of
reported depressive symptoms. Reliability for the 3 years
was a = .85, a = .80, and a = .89, respectively.
Demographic Characteristics
Single items were used to assess students’ grade level,
gender (female coded 1), whether they received free lunch
or reduced price lunch at school (free or reduced-price
lunch coded 1), race or ethnicity (minority status coded 1),
father and mother’s educational level, and with whom they
have lived in the past year.
Analytic Strategy
Cross-domain growth modeling was used to test the
hypotheses of the current study. Technical discussions
(e.g., Sayer and Willett 1998; Willet and Keiley 2000) as
well as longitudinal applications (e.g., Willett and Sayer
1996) of this methodology are available in the literature.
Cross-domain modeling capitalizes on the richness of two
methodologies, individual growth modeling, and covari-
ance structure analysis. The former permits the assessment
of systematic differences in change within and between
individuals, while the latter allows a more flexible system
of testing inter-relationships among these changes. For
instance, one can investigate not only whether individual
differences in change in a particular domain are related to a
specific set of predictors but also if change in one doma in is
related to change in another.
Individual growth models specify ‘‘intercept’’ and
‘‘slope’’ growth parameters and define within-person or
level-1 models. The intercept parame ter represents initial
status as defined by the investigator whereas the slope
parameter represents the rate of change over time.
Covariance structure analyses then allows the specification
and test of relationships among different individual growth
models and thereby defines the between-person or level-2
models. In cross-domain modeling, level-1 models are first
mapped onto a covariance structure matrix. In structural
equation modeling terms, each level-1 model can be treated
as an exogenous or endogenous measurement models. The
repeated observations of each construct form indicators
that delineate two latent growth parameters, the intercept
and slope. An important step in defining both growth
parameters is choosing a suitable mathematical model to
represent change over time. Since we were constrained by
only three waves of data, a linear trajectory in growth was
selected for the current model. As the observations were
taken during each academic year, this defined a time
interval of t = 1, 2, 3. In order for the intercept to represent
status in the sixth grade, we rescaled the time metric by
subtracting one (so that t = 0, 1, 2). Apart from ensuring
that the intercept represented initial status, this rescaling
also implied that the slope represents linear rate of change
over time. The first growth parameter, the intercept, is
defined by the first latent factor and represents the initial
levels or status of the measures when the adolescent is in
the sixth grade. The second growth parameter, slope, is
represented by the equal spaced loadings (i.e., 0, 1, and 2)
of the observed variables and defines the linear rate of
change of growth over time. Note that all individuals of the
sample are assumed to have the same linear form of growth
but different individuals may have different values for the
two growth parameters. The means of these latent intercept
and slope factors represent group growth factors and are
the estimate of the average true intercept and slope in the
population. Heterogeneity in growth is captured by the
variances of the intercept and slope latent factors. Predic-
tors of this variability can thus be easily introduced using
the structural equation modeling (SEM) framework by
regressing the predictors on the intercept and slope latent
factors.
Am J Community Psychol (2007) 40:194–213 199
123
The power of the cross-domain analyses is obtained
when such individual latent growth models are combined
together. In SEM terminology, these combine d models
represent structura l models that speci fy relationships
between the indivi dual growth models. These combined
models therefore define between-person or level-2 models.
These models can further be expanded by adding specific
person-level characteristics that are used as predictors of
initial status and growth.
In the case of the current study, we were interested in (a)
unconditional models that would highlight the average
trajectories of change over time for each dimension of
school climate as well as each outcome variable, (b) the
effects of student gender and socioeconomic status on both
the initial levels and slopes of each dimension of perceived
school climate, (c) the associations among changes in the
different dimensions of school climate perceptions and the
concomitant change s in self-esteem, depressive symptoms,
and behavior problems, and (d) models that examined the
alternative hypothesis, namely, that the direction of effects
between each dimension of perceived school climate and
adjustment is bi-directional rather than unidirectional.
Secondary to these analyses, we also report the nature of
the relation between initial levels of the school climate
dimensions and initial levels of depressive symptoms,
general self-esteem, and behavior problems, and the rela-
tion between initial levels of school climate dimensions
and rates of change in depressive symptoms, self-esteem,
and behavior problems. Cross-domain growth modeling
was selected for the current study because it allows us to
fully examine each of our research questions. All models
were estimated using LISREL 8.50 (Jo
¨
reskog and So
¨
rbom
2001). In the LISREL mode l, the standardized regression
coefficient is the effect size (Kline 1998). As suggeste d by
Kline, the interpretation of the absolute magnitudes of path
coefficients can follow the general recommendations by
Cohen (1988). Standardized path coefficients with absolute
values less than .10 indicate a ‘‘small’’ effect; values
around .30 a ‘‘medium’’ effect; and values of .50 and above
reflect a ‘‘large’’ effect.
Results
Treatment of Missing Data
Comparison of the retained and eliminated sample revealed
that students who were retained for analysis were more
likely to be female, white, and have mothers who were
college educated. Furthermore these students were also
more likely to come from two-parent families. The
list-wise deletion of the student data also resulted in the
elimination of one school from the sample. The eliminated
school had a greater percentage of African American stu-
dents than the retained school s in the sample. Ethnic
variability in school composition was controlled for by
controlling for minority status in the combined models.
To test the nature of any probable bias in the list-wise
elimination process, multiple imputation procedures were
conducted using NORM software (Schafer 2000). Pre-
liminary data screening revealed percent of missing data
for each variable in the analyses, which ranged from .80 to
7.87%. Four imputed data sets were created using a two-
step process under the multivariate normal model. These
data sets were then utilized to obtain parameter estimates
for the unconditional models, the conditional models for
the school climate dimensions, and the final combined and
alternative models reported in the study. Results from each
of the four separate runs (one for each of the imputed data
sets) were then re-combined in NORM. The results indi-
cated that the patterns of outcomes were similar for the
imputed and the reduced sample models. Given overall
similarities and our desire to maintain parsimony in
reporting, the findings reported in the study rely on the
reduced sample size (N = 1,451). All models were esti-
mated using LISREL 8.30 (Jo
¨
reskog and So
¨
rbom 2001).
Descriptive Analysis
Means, standard deviations, univariate skewness coeffi-
cients, and correlations among the demograp hic variables,
perceptions of each dimension of school climate and the
outcome variables in the model are presented in Table 1. The
effect sizes of the correlations are presented in Table 2. The
overall observed means show decl ines in each of the
dimensions of school climate and in the wellbeing variables
during middle school. The skewness coefficients ranged
from –.52 to 1.86, none indicating significant deviations
from symmetry. The four dimensions of school climate were
moderately and positively related to each other with the
highest coefficients for teacher support and clarity and con-
sistency in school rules. Significant correlations between the
dimensions of school climate and behavior problems,
depressive symptoms, and general self-esteem were in the
expected direction. Medium to large effect sizes were typi-
cally indicated for each dimension of school climate on
behavioral or psychological adjustment. Gender was con-
sistently, positively correlated with and indicated medium
effect sizes on each dimension of school climate except
student decision-making. Surprisingl y, no significa nt corre-
lation between gender and self-esteem or depressive
200 Am J Community Psychol (2007) 40:194–213
123
Table 1 Means, standard deviations, univariate skewness, and zero-order correlations for the perceptions of school climate, the outcome wellbeing measures, and the demographic variables
Variables 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
1. Time 1: Teacher support –
2. Time 2: Teacher support .42 –
3. Time 3: Teacher support .34 .43 –
4. Time 1: Peer support .44 .21 .20 –
5. Time 2: Peer support .23 .42 .24 .47 –
6. Time 3: Peer support .20 .38 .21 .37 .72 –
7. Time 1: Student decision-making .53 .27 .21 .28 .18 .15 –
8. Time 2: Student decision-making .29 .52 .29 .15 .29 .28 .39 –
9. Time 3: Student decision-making .24 .30 .57 .13 .16 .19 .29 .44 –
10. Time 1: Clarity and consistency in school rules .57 .32 .21 .46 .27 .21 .39 .24 .15 –
11. Time 2: Clarity and consistency in school rules .28 .63 .26 .22 .45 .38 .20 .40 .22 .43 –
12. Time 3: Clarity and consistency in school rules .21 .34 .56 .17 .24 .31 .15 .23 .39 .30 .45 –
13. Time 1: Behavior problems –.24 –.22 –.14 –.21 –.20 –.16 –.16 –.14 –.05 –.31 –.27 –.22 –
14. Time 2: Behavior problems –.17 –.33 –.21 –.18 –.29 –.25 –.09 –.18 –.13 –.29 –.42 –.29 .53 –
15. Time 3: Behavior problems –.12 –.24 –.27 –.15 –.22 –.29 –.12 –.16 –.19 –.24 –.34 –.37 .42 .54 –
16. Time 1: Self-esteem .17 .15 .11 .30 .24 .18 .12 .11 .11 .24 .14 .14 –.25 –.21 –.16 –
17. Time 2: Self-esteem .11 .21 .14 .16 .30 .25 .05 .14 .13 .15 .20 .14 –.15 –.23 –.15 .51 –
18. Time 3: Self-esteem .13 .13 .23 .17 .19 .20 .07 .06 .18 .17 .12 .23 –.17 –.23 –.27 .43 .48 –
19. Time 1: Depression –.24 –.19 –.14 –.32 –.31 –.24 –.15 –.12 –.12 –.29 –.20 –.17 .31 .24 .19 –.57 –.38 –.35 –
20. Time 2: Depression –.11 –.31 –.19 –.22 –.40 –.35 –.08 –.16 –.14 –.21 –.33 –.24 .26 .39 .27 –.38 –.56 –.39 .51 –
21. Time 3: Depression –.11 –.19 –.24 –.14 –.22 –.30 –.08 –.09 –.15 –.18 –.21 –.34 .22 .29 .38 –.34 –.35 –.56 .40 .48 –
22. Gender .15 .16 .13 .13 .14 .09 .04 .10 .02 .16 .19 .14 –.23 –.24 –.23 –.02 –.03 –.02 –.02 –.03 .01 –
23. Student SES .07 .06 .02 –.05 –.10 –.06 .05 .04 –.01 .08 –.02 –.03 .04 .08 .01 –.11 –.07 –.04 .14 .13 .05 –.02 –
24. Student minority status .08 –.02 .07 .01 –.05 –.02 .08 –.00 .02 –.08 –.08 –.02 .02 .09 –.01 .01 .03 .05 .08 .07 .00 –.02 .38 –
Mean 3.27 2.98 2.85 3.34 3.25 3.26 2.35 2.16 2.10 3.45 3.28 3.16 1.61 1.71 1.89 3.14 3.06 2.98 1.33 1.33 1.37 .54 .20 .09
Standard deviation .73 .70 .72 .53 .51 .49 .69 .69 .66 .47 .50 .48 .62 .70 .82 .58 .58 .61 .32 .32 .37 .50 .40 .29
Skewness –.31 –.26 –.22 –.41 –.37 –.41 .20 .29 .40 –.60 –.60 –.58 1.83 1.87 1.53 –.54 –.52 –.52 1.42 1.54 1.63 –.17 1.50 2.88
Am J Community Psychol (2007) 40:194–213 201
123
symptoms was obtained. Socioeconomic status was posi-
tively related to all 3 years of depressive symptoms. Apart
from this latter finding, SES did not show any consistent
pattern of association with the other variables in the model.
Initial Analytical Step: Unconditional Models
To examine how students’ perceptions of school climate as
well as indices of psychological and behavioral adjustment
change over the course of middle school, seven uncondi-
tional individual growth models were estimated, one each
for the repeated measures of the four dimensions of per-
ceived school climate (i.e., teacher support, peer support,
opportunities for student autonomy, and clarity and con-
sistency in school rules), behavioral problems, depressive
symptoms, and general self-esteem. This permitted the
examination of average growth trajectories as well as the
presence of individual variability about the average growth
parameters. Furthermore, we also examined the role of
gender and SES on each dimension of school climate by
regressing these predictors on the intercept and growth
latent factors. Since the focus of our paper is on the pre-
dictors and effects of four dimensions of perceived school
climate, for the unconditional models we did not examine
the influence of gender or SES on our three outcome
variables.
Change Over Time in Perceived School Climate
Results from the unconditional models can be seen in
Table 3. For the dimensions comprising teacher support
and peer support, the models fit the data reasonably well.
4
The average growth trajectory revealed significant (i.e.,
different from zero) levels of perceptions of teacher sup-
port and peer support in the sixth grade. Students had
positive perc eptions of both teacher and peer support at the
start of sixth grade. However, as students progressed
through middle school, their perceptions of teacher support
and peer support declined significantly every year (as can
be seen in the negative and significant slopes). Significant
heterogeneity in the initial status (at the sixth grade) and
slope of individual traject ories of teacher and peer support
were also found.
Table 2 Cohen’s d effect sizes for the perceptions of school climate, the outcome measures, and the demographic variables
Variables 1 2 3 4 5 6 7 8 9 10 11 12 13 14
1. Time 1: Teacher support –
2. Time 2: Teacher support 0.93 –
3. Time 3: Teacher support 0.72 0.95 –
4. Time 1: Peer support 0.98 0.43 0.41 –
5. Time 2: Peer support 0.47 0.93 0.49 1.06 –
6. Time 3: Peer support 0.41 0.82 0.43 0.80 2.08 –
7. Time 1: Student decision-making 1.25 0.56 0.43 0.58 0.37 0.30 –
8. Time 2: Student decision-making 0.61 1.22 0.61 0.30 0.61 0.58 0.85 –
9. Time 3: Student decision-making 0.49 0.63 1.39 0.26 0.32 0.39 0.61 0.98 –
10. Time 1: Clarity and consistency in school rules 1.39 0.68 0.43 1.04 0.56 0.43 0.85 0.49 0.30 –
11. Time 2: Clarity and consistency in school rules 0.58 1.62 0.54 0.45 1.01 0.82 0.41 0.87 0.45 0.95 –
12. Time 3: Clarity and consistency in school rules 0.43 0.72 1.35 0.35 0.49 0.65 0.30 0.47 0.85 0.63 1.01 –
13. Time 1: Behavior problems 0.49 0.45 0.28 0.43 0.41 0.32 0.32 0.28 0.10 0.65 0.56 0.45 –
14. Time 2: Behavior problems 0.35 0.70 0.43 0.37 0.61 0.52 0.18 0.37 0.26 0.61 0.93 0.61 1.25 –
15. Time 3: Behavior problems 0.24 0.49 0.56 0.30 0.45 0.61 0.24 0.32 0.39 0.49 0.72 0.80 0.93 1.28
16. Time 1: Self-esteem 0.35 0.30 0.22 0.63 0.49 0.37 0.24 0.22 0.22 0.49 0.28 0.28 0.52 0.43
17. Time 2: Self-esteem 0.22 0.43 0.28 0.32 0.63 0.52 0.10 0.28 0.26 0.30 0.41 0.28 0.30 0.47
18. Time 3: Self-esteem 0.26 0.26 0.47 0.35 0.39 0.41 0.14 0.12 0.37 0.35 0.24 0.47 0.35 0.47
19. Time 1: Depression 0.49 0.39 0.28 0.68 0.65 0.49 0.30 0.24 0.24 0.61 0.41 0.35 0.65 0.49
20. Time 2: Depression 0.22 0.65 0.39 0.45 0.87 0.75 0.16 0.32 0.28 0.43 0.70 0.49 0.54 0.85
21. Time 3: Depression 0.22 0.39 0.49 0.28 0.45 0.63 0.16 0.18 0.30 0.37 0.43 0.72 0.45 0.61
22. Gender 0.30 0.32 0.26 0.26 0.28 0.18 0.08 0.20 0.04 0.32 0.39 0.28 0.47 0.49
23. Student SES 0.14 0.12 0.04 0.10 0.20 0.12 0.10 0.08 0.02 0.16 0.04 0.06 0.08 0.16
Note: These effect sizes were calculated from the r values using the following formula: 2r/sqrt(1 – r
2
) (Rosenthal and Rosnow 1992). Absolute
values displayed
4
In none of the unconditional models were the errors allowed to
correlate over time, which may have resulted in better fitting models.
202 Am J Community Psychol (2007) 40:194–213
123
The models for stud ent autonomy and clarity and con-
sistency of school rules also fit the data reasonably well.
The average growth trajectory revealed significant (i.e.,
different from zero) levels of perceptions of student
autonomy and clarity and consistency of school rules in the
sixth grade. Moreover, the average trajectory for the slope
indicated that the perceptions of student autonomy as well
as those for clarity and consistency of school rules declined
over the course of middle school. Significant heterogeneity
in the initial status and slope of individual trajectories of
student autonomy and clarity and consistency of school
rules were also found.
Impact of SES and Gender
To further examine the nature of the significant individual
variation in initial status and growth over time in the tea-
cher and peer support, student autonomy, and clarity and
consistency in school rules, the intercept and the slope
factors of each of the models discussed above were
regressed on gender and SES (see Table 4). For each
model, non-significant pathways were specified to zero and
the models were re-fitted. In each case, the re-fitted models
did not show significant degradation from the originals.
The impact of gender and/or SES on each dimension of
school climate indicated a small to medium effect size as
indicated by the path coefficient.
For teacher support, girls at the beginning of the sixth
grade reported more teacher support than boys. Similarly,
adolescents from a lower socioeconomic background (i.e.,
receiving free or reduced price lunch) reported more
positive perceptions of teacher support at the sixth grade
than students from a higher socioeconomic status (i.e., not
eligible for free or reduced price lunch). With respect to
peer support, at the start of middle school, girls tended to
perceive more peer support than boys. However, as the
girls progressed through middle school, they evidenced
steeper declines in their perceptions of peer support than
the boys. In addition, students from a lower socioeconomic
background reported less positive peer support at the start
of middle school than the students from higher socioeco-
nomic backgrounds, but the change over time did not differ
by SES.
For student autonomy, gender predicted initial levels
such that girls at the start of middle school perceived
greater opportunities for student involvement in the deci-
sion-making in their classrooms than boys. There were no
SES differences indicated for the intercept or slope of
opportunities for student autonomy. For clarity and
Table 4 Model fit and estimates for the influence of gender and SES on school dimensions
Model v
2
(d.f.) RMSEA (CI) CFI v
diff
2
(d.f.) From Gender From SES
To status To slope To status To slope
Teacher support 48.81 (8) .06 (.04–.08) .97 5.56 (4) .21*** – .09** –
Peer support 38.42 (7) .06 (.04–.07) .98 1.31 (3) .29*** –.07* –.16** –
Student decision-making 47.82 (9) .05 (.04–.07) .95 10.33 (5) .07* – – –
Clarity and consistency in school rules 27.78 (8) .04 (.03–.05) .98 5.26 (4) .15*** – .10** –
Note: N = 1,451. The chi-square reported is for the final model. The RMSEA confidence interval is for the 90%CI. The difference in chi-square is
between the initial and the final re-fitted model
*p \ .05; **p \ .01; ***p \ .001
Table 3 Model fit and coefficients for the unconditional models
Model v
2
RMSEA (CI) CFI Coefficients Variance r
Status Slope Status Slope
Teacher support 19.51 .11 (.07–.16) .97 3.23*** –.20*** .25*** .04** –.37*
Peer support 26.24 .13 (.09–.18) .98 3.31*** –.02** .15*** .06*** –.29**
Student decision-making 19.63 .11 (.07–.16) .97 2.33*** –.12*** .23*** .05*** –.43***
Clarity and consistency in school rules 11.98 .09 (.05–.13) .98 3.45*** –.16*** .13*** .04*** –.46***
Behavior problems 9.06 .07 (.04–.12) .99 1.60*** .14*** .25*** .05*** –.14*
General self-esteem .003 \.01 (.00–.00) 1.0 3.13*** –.08*** .19*** .02* –.39*
Depressive symptoms 11.48 .08 (.05–.13) .99 1.32*** .02*** .05*** .01* –.18
Note: The v
2
is based on (1, N = 1,451). The RMSEA confidence interval is for the 90%CI. The correlation is between initial status and slope
*p \ .05; **p \ .01; ***p \ .001
Am J Community Psychol (2007) 40:194–213 203
123
consistency in school rules, girls perceived greater clarity,
consistency, and fairness at the start of middle school than
boys. Students from lower socioeconomic backgrounds
reported less clarity, consistency, and fairness at the start of
middle school than their more privileged counterparts.
Indices of Wellbeing
The unconditional models for behavior problems, general
self-esteem, and depression were also tested (see Table 3).
The average trajectory for behavior problems revealed the
following. Initial status of behavior problems was signifi-
cantly different from zero. The estimate of the slope was
positive and also significantly different from zero, indi-
cating population average increased in reports of behavior
problems over time. With respect to general self-esteem, a
negative significant slope factor indicated declining levels
of self-esteem over the course of middle school for the
average population growth traject ory. The average growth
trajectory for symptoms of depression revealed significant
levels of depression in the sixth grade. The estimate of the
slope was positive and also significantly different from
zero indicating an increase in reports of symptoms of
depression over time. Significant variance in both initial
status and slope factors was obtained for all three indices of
wellbeing, indicating the presence of heterogeneity in the
individual trajectories over time (see Table 3).
Summary
In summary, adolescents in the study tended to perceive
middle schools as increas ingly negative over time. Fur-
thermore, they report declining levels of wellbeing, as
indicated by increas ing levels of depressive symptoms,
behavior problems, and decreasing levels of general self-
esteem over the course of middle school. Gen der differ-
ences were detected for the initial levels of all four
dimensions of school climate, with girls perceiving more
positive school climates at the start of middle school. Girls
also reported sharper declines in peer support than boys.
Socioeconomic differences were also observed for the
initial levels of teacher support, peer support, and clarity
and consistency in school rules. There were no SES dif-
ferences in the rates of decline for any of the school climate
dimensions.
The Combined Models
The third aim of the current study was to examine the
nature of associations between each of the dimensions of
perceived school climate and psychological and behavioral
adjustment (see Fig. 1). To this end, we tested a combined
model that include d the unconditional models presented
above. Four different models were thus explored; one for
each dimension of school climate. The associations among
the growth parameters across the domains were estimated
via regression coefficients between the respective initial
status and slopes. Specifically, initial status and slopes of
the outcome variables (i.e., behavior problems, general
self-esteem, and depressive symptoms) were regressed on
the initial status of each of the dimensions of school cli-
mate to examine the influence of initial levels of the
specific dimension of school climate on behavior problems,
self-esteem, or depressive symptoms growth trajectories.
To explore how rates of changes each of the dimensions of
school climate predicted rates of changes in psychological
and behavioral outcomes over the course of middle school,
the slope of the outc ome variables were regressed on the
slope of school climate dimension. Additionally in each
model, the initial status and the slope of each of dimension
of school climate was regressed on gender, SES, and
minority status. Although we included minority status in
these models given research that suggests across group
variation in the association between school climate and
adjustment (see Way and Robinson 2003), we do not
interpret the nature of this relationship in any meaningful
way given the insufficient sam ple size for minority students
(only 9% of the sample). By including minority status,
however, any possible variation attributed to this variable
was controlled for in the models. Furthermore, in light of
previous research findings suggesting gender differences in
depressive symptoms, self-esteem, and behavioral prob-
lems (e.g., Block and Robins 1993; Cicchetti and Toth
1998; Nolen-Hoeksema and Girgus 1994) the growth
parameters for the outcome variables (i.e., both status and
slope factors) were regressed on gender.
Teacher Support
The combined model fit the data well (v
2
(43)
N = 1,451) = 192.34, RMSEA = .05, CI = .04–.06, CFI =
.98). Neither minority status nor SES significantly pre-
dicted the initial status or slope of teacher support.
Additionally, gender did not significantly predict depres-
sive symptoms or any of the slopes of the variables in the
model. We theref ore re-estimated this model setting these
paths to zero. The test of the nested chi-square revealed
that the model fit was not significantly degraded (v
d
2
(9) =
15.95, p [ .05). This latter model is thus reported (see
Fig. 2). Gender differences were found with respect to the
initial status of teacher support (as reported in the uncon-
ditional models), general self-esteem, and behavior
204 Am J Community Psychol (2007) 40:194–213
123
problems. At the beginning of sixth grade, girls reported
significantly lower levels of self-esteem, while boys
reported higher incidences of behavior problems at the start
of middle school. The intercept of teacher support (the
sixth grade assessment) was significantly associated with
the intercepts of self-esteem and depressive sym ptoms and
with the intercept and slope of problem behavior.
With respect to our primary question regarding the
association between changes over time in teacher support
on changes over time in depressive symptoms, self-esteem,
or behavior problems, the slope of teacher support was
significantly (i.e., different from zero) related to the slopes
of depressive symptoms and behavior problems. A decline
in perceptions of positive teacher support over time was
associated with corresponding increases in depressive
symptoms and behavior problems. The associations
between the intercept and/or slope of teacher support and
the intercept and/or slope of self-esteem, depressive
symptoms, or problem behavior sugges ted medium to large
effect sizes, with the slope of teacher support on the slopes
of problem behavior and depressive symptoms suggesting
particularly large effect sizes (see Fig. 2).
Peer Support
The model that assessed the degree of influence of peer
support in the school fit the data well (v
2
(43)
N = 1,451) = 155.23, RMSEA = .04, CI = .03–.05, CFI =
.98). Examination of this model revealed that gender
significantly predicted the initial status and slope of peer
support (as reported in the unconditional models), initial
status and slope of behavior problems, and initial sta tus of
both general self-esteem and depressive symptoms. SES
predicted the initial status of peer support (as reported in
the unconditional models). All non-significant paths in the
relationships between the demographic variables and the
other variables in the model were set to zero and the model
re-estimated. The chi-square difference test indicated that
the final model was not significantly degraded (v
d
2
(5) =
5.20, p [ .05) (see Fig. 3). As expected, girls reported
lower levels of behavior problems, lower self-esteem, and
higher levels of depressive symptoms at the beginning of
middle school. Boys were more likely to report sharper
increases in behavior problems over middle school when
compared to the girls. The intercept of peer support was
significantly associated with the intercept and slope of self-
esteem and depressive symptoms and the intercept of
problem behavior.
With respect to the association between change over
time in peer support and change over time in adjustment,
the findings indicated that perceptions of declining peer
support were associated with declines in psychological and
behavioral adjustment. As students reported declines in
peer support, there were corresponding increases in
depressive symptoms and behavior problems and declines
in general self-esteem. The association between the inter-
cept and/or slope of peer support and the intercept and/or
slope of self-esteem, depression, or problem behavior
indicated medium to large effect sizes (See Fig. 3).
Opportunities for Student Autonomy
The combined model for opportunities for student auton-
omy fit the data well (v
2
(43) N = 1,451) = 176.75,
62.-
50.-
02.
33
.
-
1
2.
-
92.
93.-
3
8.-
7
7
.-
Gender
Teacher Support
Intercept
Teacher Support
Slope
Bh. Problem
Intercept
Bh. Problem
Slope
Depression
Intercept
Depression
Slope
Self-Esteem
Slope
Self-Esteem
Intercept
Fig. 2 Final combined cross-
domain model for student
perceptions of teacher support.
For the sake of clarity, only
significant paths are
represented. All parameter
values are standardized. Model
v
2
(52, N = 1,451) = 208.29,
RMSEA = .05, CI = .04–.05,
CFI = .97
Am J Community Psychol (2007) 40:194–213 205
123
RMSEA = .05, CI = .04–.05, CFI = .97). With the excep-
tion of the path from gender to initial status of student
autonomy (as reported in the unconditional models) and
behavior probl ems, gender, SES, and minority status did
not significantly predict the variables in the model. We
therefore set these non-significant paths to zero and re-
estimated the model. There was no significant degradement
of the final model (v
d
2
(10) = 14.39, p [ .05) (see Fig. 4).
Boys reported higher levels of problem behaviors than the
girls at the sixth grade. The intercept of opportunities for
student autonomy was significantly associated with the
intercept of depressive symptoms and self-esteem and the
intercept and slope of behavi or problems.
The rate of change in perceptions of opportunities for
student autonomy was significantly related to the rate of
changes in depressive symptoms, behavior problems, and
self-esteem. As students perceived sharper declines in
opportunities for student decision-making in the classroom
over time, they also reported increases in depressive
symptoms and behavior problems over time, and decreases
in self-esteem. The association between the intercept and/
or slope of student autonomy and the intercept and/or slope
of self-esteem, depressive symptoms, or behavior problems
indicated medium effect sizes (see Fig. 4).
Clarity and Consistency in School Rules
The combined model for clarity and consistency in school
rules fit the data well (v
2
(43) N = 1,451) = 150.05,
03.-
70.
81.-
12.-
4
2.
82.-
43.-
53.-
72.
redneG
Student autonomy
Intercept
Student autonomy
Slope
Bh. Problem
Intercept
Bh. Problem
Slope
Depression
Intercept
Depression
Slope
Self-Esteem
Slope
Self-Esteem
Intercept
Fig. 4 Final combined cross-
domain model for student
perceptions of participation in
student decision-making. For
the sake of clarity, only
significant paths are
represented. All parameter
values are standardized. Model
v
2
(53, N = 1,451) = 191.14,
RMSEA = .04, CI = .04–.05,
CFI = .98
32.-
81.
43.-
7
4
.
04.-
0
4.
57.
-
43.
56.-
61.
-
60.-
7
0
.
90.-
01.-
01
.
-
redneG
Peer Support
tpecretnI
Peer Support
Slope
Bh. Problem
tpecretnI
Bh. Problem
Slope
noisserpeD
tpecretnI
noisserpeD
Slope
Self-Esteem
Slope
Self-Esteem
tpecretnI
SES tned
u
tS
Fig. 3 Final combined cross-
domain model for student
perceptions of peer support. For
the sake of clarity, only
significant paths are
represented. All parameter
values are standardized. Model
v
2
(48, N = 1,451) = 160.43,
RMSEA = .04, CI = .03–.05,
CFI = .98
206 Am J Community Psychol (2007) 40:194–213
123
RMSEA = .04, CI = .03–.05, CFI = .98). SES status did
not significantly predict either the initial status or slope of
clarity and consistency in school rules. In addition,
minority status and gender were significantly associated
with the intercept but not with the slope of clarity and
consistency in school rules. Gender also predicted each of
the outcome variables (i.e., behavior problems, depressive
symptoms, and self-esteem). The non-significant paths
were next set to zero and the model was re-estimated.
There was no significant degradement of the final model
(v
d
2
(7) = 8.58, p [ .05). This final model is graphically
displayed in Fig. 5. Girls in the sixth grade reported
higher levels of depressive symptoms and lower levels of
self-esteem than boys, while boys reported higher levels
of problem behavior than girls. The intercept of clarity
and consistency in school rules was significant associa ted
with the interc ept of self-esteem and depressive symp-
toms and with the intercept and slope of problem
behavior.
Our findings with regards to the slope of clarity and
consistency in school rules indicated that for students who
reported an increase in perceptions of clarity and consis-
tency in school rules, there was a corresponding decrease in
depressive symptoms and behavior problems over time. The
association between the intercept and/or slope of clarity and
consistency in school rules and the intercept and/or slop e of
self-esteem, depressive symptoms, or behavior problems
indicated medium to large effect sizes (See Fig. 5).
In summary, students’ perceptions of teacher and peer
support, student autonom y, and clarity and consistency in
school rules in sixth grade (initial status) was significantly
associated with students’ reported levels of depressive
symptoms, general self-esteem, and behavior problems in
the sixth grade. In addition, students who perceived their
school as having increasingly more teacher and peer sup-
port, a greater degree of student autonomy, and more
clarity and consistency in school over time also reported
declines over time in depressive symptoms and behavior
problems. Moreover, reports of increasingly positive peer
support and a greater degree of student autonomy were
significantly related to increases over time in student
reports of general self-esteem. The effect sizes in each
model suggested medium to large effect sizes, with the
slope of peer support, teacher support, and clarity and
consistency in school rules on the slope of depre ssive
symptoms suggesting a larger effect size than the other
paths in the model (with the exception of the slope of
teacher support and the slope of problem behavior).
Testing for Direction of Effe cts
To assess whe ther the effects of each dimension of school
climate on psychological and behavioral adjustment were
bi-directional rather than simply unidirectional, a series of
alternative models for each of the four dimensions of
school climate were tested . In these models, each dimen-
sion of school climate was viewed as the outcome variable
and the three indicators of wellbeing (i.e., behavior prob-
lems, general self-esteem, and depressive symptoms) were
entered as predictors. Student gender, socioeconomic, and
minority status were included in the model as defined
earlier. We could thus make qualitative comparisons of
model effects and gain some degree of insight as to the
direction of effects from the different dimensions of school
climate to wellbeing.
81.-
22.
25.-
43
.
04.-
57.
-
94.-
9
0
.
01.-
80.-
72.-
redneG
Clarity and consistency in
school rules
Intercept
Clarity and
consistency in school
rules
Slope
Bh. Problem
Intercept
Bh. Problem
Slope
noisserpeD
Intercept
Depression
Slope
Self-Esteem
Slope
Self-Esteem
Intercept
Minority
suta
t
S
Fig. 5 Final combined cross-
domain model for student
perceptions of school clarity and
consistency in school rules. For
the sake of clarity, only
significant paths are
represented. All parameter
values are standardized. Model
v
2
(50, N = 1,451) = 158.64,
RMSEA = .04, CI = .03–.05,
CFI = .98
Am J Community Psychol (2007) 40:194–213 207
123
Teacher Support
The overall model fit the data as well as the original (v
2
(43)
N = 1,451) = 172.68, RMSEA = .05, CI = .04–.05, CFI =
.98). Initial status of behavior problems were significantly
related to the initial status of perceptions of teacher support
such that students who report higher incidences of behavior
problems in the sixth grade also tended to report less
positive perceptions of teacher support (b = –.29,
p \ .001). A significant negative relationship was also
found between the slope of behavior problems and the
slope of teacher support such that students who report
increasing incidences of behavior problems over middle
school also tended to show steeper declines in perceptions
of teacher support over time (b = –.40, p \ .001). How-
ever, neither the slopes of general self-esteem nor
depressive symptoms significantly predicted the slope of
perceptions of teacher support (self-esteem: b = .12,
p [ .10; depressive symptoms: b = –.24, p [ .10).
Peer Support
The overall alternative model for the nature of peer support
in the school fit the data as well as the original (v
2
(43)
N = 1,451) = 190.39, RMSEA = .05, CI = .04–.05, CFI =
.98). In this alternative model, apart from the significant
path from the initial status of behavior problems and the
initial status of peer support (b = –.11, p \ .01), the only
other significant paths were from the initial status and slope
of depressive symptoms to the status (b = –.59, p \ .001)
and slope (b = –.85, p \ .02) of peer support. Students who
reported increasing levels of depressive symptoms over the
course of middle school also tended to report declining
levels of positive peer support in the school. However,
neither the slopes of general self-esteem nor problem
behavior significantly predicted the slope of peer suppor t.
Opportunities for Student Autonomy
The overall model fit the data as well as the original (v
2
(43)
N = 1,451) = 158.26, RMSEA = .04, CI = .04–.05, CFI =
.98). As in the model for teacher support, the only sig-
nificant paths were found for behavior problems. The initial
status and slope of behavior problems were significantly
related to the intercept (b = –.22, p \ .001) and slope
(b = –.49, p \ .05) of student autonomy, respectively.
Students who report increases in behavior problems over
the course of middle school also tended to show steeper
declines in perceptions of student autonomy. Yet the initial
status or slopes of general self-esteem or depressive
symptoms did not significa ntly predict the initial status or
slope of student decision-making.
Clarity and Consistency in School Rules
The overall model for the alternative hypothesis for clarity
and consistency in school rules fit the data as well as the
original model (v
2
(43) N = 1,451) = 153.27, RMSEA =
.04, CI = .03–.05, CFI = .98). Once again, the only sig-
nificant path was between the slope of behavior problems
and the slope of clarity and consistency in school rules
(b = –.40, p \ .01). Students who reported an increasing
incidence of behavior problems over middle school also
tended to show steeper declines in perceptions of clarity
and consistency in school rules over time. Additionally, the
initial status of depressive symptoms (b = –.26, p \ .05)
and behavior problems (b = –.33, p \ .001) were signifi-
cantly related to the initial status of clarity and consistency
in school rules. However, neither the slopes of depressive
symptoms nor general self-esteem predicted the slope of
clarity and consistency in school rules.
In summary, the pattern of findings suggests that the
influence of teacher support, student autonomy, and clarity
and consistency in school rules on depressive symptoms is
indeed from the dimensions of perceived school climate to
depressive symptoms rather than the reverse. Furthermore,
the pathway of influence between student decision-making
and self-esteem appear to be from changes in student
autonomy to changes in self-esteem rather than the reverse.
Finally, the pathways of influence for peer support appear
to be from changes in perceptions of peer support to
changes in behavior problems and general self-esteem
rather than the reverse.
5
Discussion
The major goals of this study were to examine changes in
students’ perceptions of teacher and peer support, oppor-
tunities for student autonomy in the classroom, and clarity
and consistency of school rules during the middle school
years, the effects of gender and SES on those changes, and
the psychological and behavioral effects of various patterns
of change over time in the different dimensions of per-
ceived school climate. Our unconditional models revealed
that, across gender and SES, perceptions of all four
dimensions of school climate declined over the 3 years of
middle school. The decline in perceptions of school climate
may be based on a growing mismatch between what stu-
dents need (i.e., relatedness, opportunities for autonomy,
and consistency) and what the school actually provides
5
Results from the multiple imputations more strongly indicate that
the pathways of influence do emerge from the dimensions of school
climate to student adjustment implying the need to further study these
processes in a larger sample.
208 Am J Community Psychol (2007) 40:194–213
123
(Eccles et al. 1993; Roeser et al. 1998; Connell and
Wellborn 1991). It is also possible, however, that middle
school students grow increasingly dissatisfied with their
schools, irrespective of the actual climate of their schools.
Numerous researchers have documented the biological,
cognitive, and psychological changes that early adolesce nts
endure (Eccles et al. 1993). Such difficult changes may
lead early adolescents to perceive their school environ-
ments as increasingly more negative over time regardless
of the actual schools they attend. Future research should
examine both the objective (e.g., how much oppo rtunity do
the students have in their classes to participate in decision-
making) and subjective aspects of school climate to assess
the association between what the school actually provides
for its students and how the students perceive the school.
Although students reported declines in each of the four
dimensions of school climate, girls reported sharper
declines in peer support than boys over time (although the
effect size for gender was small). This finding is consistent
with our hypothesis and supports the research that suggests
that peer relations during middle school are particularly
difficult for girls (Crick and Rose 2000 ; Sadker and Sadker
1994). Our findings did not, however, consistently suppor t
our hypotheses that girls would experience their school
more negatively than boys. In the sixth grade, girls reported
better teacher and peer support and better clarity and
consistency in school rules and student autonomy than
boys. Previous research with secondary school teachers
suggest that teachers often believe that girls work harder in
school than boys, and report having better relationships
with their female students than with their male students
(Suarez-Orozco and Qin-Hilliard 2004). Th ese biases and
beliefs favoring girls may lead girls in the sixth grade to
perceive their school envi ronment as more supportive than
boys. Yet these biases do not seem to protect the girls from
perceiving their school to be decreasingly supportive,
consistent, and fair over time.
Similar to our findings regarding gender differences, our
findings with respect to SES differences in perceptions of
school climate both supported and failed to support our
hypotheses. In support of our hypotheses, we found that
lower SES students compared to higher SES students
reported lower peer support and poorer perceptions of
clarity and consistency in school rules at the sixth grade
assessment. However contrary to our hypothesis, lower
SES students reported higher leve ls of teacher support at
the sixth grade assessment and both the lower and higher
SES students reported similar rates of decline in all four
dimensions of school climate. Furthermore, the effect sizes
for SES on perceptions of school climate were typically
small. These findings suggest that while SES appears to
differentiate, at least to some degree, students’ perceptions
of school climate in the sixth grade (but not necessarily in a
consistent manner), SES does not make a difference in the
trajectory of students’ perceptions of school climat e over
time. Understanding when and how SES makes a differ-
ence in students’ perceptions of their school is an important
direction for future research.
Another primary goal of our analysis was to explore the
association between rates of change in perceptions of each of
the four dimensions of school climate and rates of change in
psychological and behavioral adjustment. As hypothesized,
the rates of change for teacher and peer support, student
autonomy, and clarity and consist ency in school rules were
significantly associated with the rates of change for self-
esteem, depressive symptoms, and problem behavior .
Declines in teacher support, peer support, student autonomy,
and clarity and consistency in school rules were associated
with declines in psychological and behavioral adjustment.
Furthermore, these models of change indicated medium to
large effect sizes. These findings underscore the importance
of not only examining the association between students
perceptions of their schools and their adjustment, but also of
examining the ways in which change in perceptions of school
is associated with change in adjustment over time. Our data
clearly indicate that neither of these types of perceptions or
experiences are static or unchanging over time. Thus, any
analysis assessing their mutual influence should account for
the dynamic nature of these processes.
Strikingly, our test for direction of effect indicated that the
influence of teacher support, peer support, or clarity and
consistency in school rules on depressive symptoms was
unidirectional; the influence of peer support and student
autonomy on self-esteem was unidirectional; and the influ-
ence of peer support on problem behavior was unidirectional.
These findings indicate that it is often the students’ percep-
tions of school climate, particularly their perceptions of peer
support that predicts adjustment and not the students’
adjustment predicting perceptions of school climate.
Taken together, these analyses underscore the role of
teacher and peer support, opportunities for student auton-
omy, and clarity and consistency in school rules and
regulations in the psychological and behavioral health of
students during the middle school years. We must take
students’ perceptions of school seriously as they consis-
tently appear to be significantly associated with students’
health and wellbeing. Students’ perceptions of the school
climate appear to be important not just during the transition
from elementary to middle school (see Eccles et al. 1993),
but also during middle school itself. Drops in self-esteem
and increases in depressive symptoms have become almost
expected markers of adolescence, presumed to be as de-
contextualized and inevitable as the loss of a child’s first
set of teeth. Yet our results suggest that students’ emotional
struggles, especially with depressive symptoms, are closely
linked with their perceptions of the school culture. Indeed,
Am J Community Psychol (2007) 40:194–213 209
123
the covariation of students’ adjustment with their percep-
tions of different dimensions of school climate over
3 years, as well as the unidirectional influence of percep-
tions of school climate on adjustment, particula rly
depressive symptoms, provides us with unusually strong
evidence for the influence of perceptions of school on
student wellbeing.
Limitations
Although the current study provides important data
regarding the experience of middle school for students as
well as the correlates of such experience, there were sev-
eral limitations to the study. A key limitation was the small
number of ethnic minority students in the sample. Although
we included ethnic minority adolescents in the analysis, our
small numbers did not allow us to compare across ethnic
groups. Our assessment of SES was also limited. Free or
reduced lunch can be an impre cise measure of SES given
that not all students who qualify sign up for this program
and standards for qualifying vary across school districts.
However, it is a common ly used technique of assessing
SES and is more reliable as a marker of SES than student
self-report of family income (Conchas and Noguera 2004;
Way 1998). The study sample, which had relatively high
rates of both parental college educa tion and free lunch
eligibility, was not necessarily representative of the broader
US population. These particular demographics may have
been a function of the broad array of communities—urban,
suburban, sma ll town, and rural—that were sampled as
well as an oversampling of schools with high levels of
economically and socially disadvantaged students.
Future research should also examine additional predic-
tors of perceived school climate aside from gender or SES.
For example, the size of the student body, as well as the
location (urban versus rural) may have an impact on per-
ceptions of school climate. While a preliminary data
analysis indicated that inclusion of grade size (e.g., sixth
grade) did not change the pattern of results observed in the
current study,
7
future research should investigate school-
level factors that may shape adolescents’ perceptions of
their school environment.
Finally, although the current data structure is very well
suited for hierarchi cal linear modeling, cross-domain
growth modeling was sel ected as the aim was to study the
nature of the relationship of change across different
domains. We acknowledge that the nested structure of the
current data is not incorporated in the planned analyses and
this may compromise, to some unknown degree, the
resulting standard errors. We have thus been conservative
in interpreting the significance of our results relying on a
more stringent alpha leve l.
Conclusion
Our findings sugges t that stud ents’ experiences of the
school environment during the middle school years may be
as important in shaping psychological and behavioral
adjustment as the transition from elementary school to
middle school. Indeed, students’ perceptions appear
mutable during middle school and, in fact, initially positive
perceptions can grow increasingly less positive and these
perceptions appear to affect adaptation. These patterns of
decline in student perceptions of the school climate may
even be more dramatic if one were to examine student
perceptions in the last 7 years as schools have become
more focused on testing in response to the ‘‘no child left
behind’’ policies and harsher and more regimented due to
catastrophes like Columbine and Jonesboro. Our findings
further suggest that there is considerable variation among
students in perceptions of the teacher and peer support,
clarity and consistency in school rules, and student
autonomy and that the influence of such perceptions extend
beyond academic outcomes to include students’ emotional
and behavioral wellbeing. Taken together, these findings
highlight the need for addi tional research on this critical
period of development. Why do most students perceive
their school environment more negatively over time? What
accounts for the variation in these perceptions? What types
of students maint ain positive perceptions over time? What
can be done to improve students’ perceptions of middle
schools? These questions seem particularly critical when
the overwhelming focus in many schools in the United
States at this time is on academic standards and testing
rather than on creating an environment that is perceived by
the students to be positive, supportive, and safe.
Acknowledgment We gratefully acknowledge the support of Peter
Mulhall at the Illinois Center for Prevention Research. In addition, we
are grateful for the support of the Spencer Foundation who provided
the first and third author with a grant to analyze the data presented in
this paper.
Appendix A
Instructions: How often are the following things true about
THIS SCHOOL? Mark the best answer for each statement.
Never Hardly ever Sometimes Most of the time Always
(1) (5)
7
Preliminary models that included school size were examined.
Inclusion of grade size did not change the pattern of results observed
in the current study.
210 Am J Community Psychol (2007) 40:194–213
123
Teacher Support
1. Teachers take a personal interest in students.
2. Teachers go out of their way to help students.
3. If students want to talk about something, teachers will
find time to do it.
4. Students really enjoy their classes.
5. Teachers help student s to organize their work.
6. Teachers help students catch up when they return from
an absence.
Peer Support
Negative peer interactions
1. Students in this school are mean to each other.
2. There are kids in this school who pick on other kids.
3. Students in this school have trouble getting along with
each other.
4. In classes, students find it hard to get along with each
other.
5. Students in this school fell students are mean to them.
Affiliation
1. Students in this school get to know each other really
well.
2. Students in this school are very interested in getting to
know other students.
3. Students enjoy working together on projects.
4. Students get to know each other well in classes .
5. Students enjoy doing things with each other in school
activities.
Student Autonomy Climate
1. Students in this school have a say in how things work.
2. Students help decide how class time is spent.
3. In our school, students are given the chance to help
make decisions.
4. Students get to help decide some of the rules in this
school.
5. Teachers ask students what they want to learn about.
Clarity and Consistency in School Rules
School Structure
1. Teachers make a point to sticking to the rules in
classes.
2. When teachers make a rule, they mean it.
3. Students are given clear instructions about how to do
their work in classes.
4. Students understand what will happen to them if they
break a rule.
5. If some students are acting up in class, the teachers
will do something about it.
School Harshness
1. Teachers are very strict here.
2. Students get in trouble for breaking small rules.
3. Students get in trouble for talking.
4. It is easy for a student to get kicked out of class in this
school.
5. The rules in this school are too strict.
References
Achenbach, T. M., & Edelbrock, C. (1987). Manual for the youth and
self-report and profile. Burlington, VT: University of Vermont.
Alvidrez, J., & Weinstein, R. (1999). Early teacher perceptions and
later student academic achievement. Journal of educational
psychology, 91, 731–746.
American Association of University Women (1992). How schools
shortchange girls. New York: Marlowe & Company.
Anyon, J., & Wilson, W. J. (1997). Ghetto schooling: A political
economy of urban reform . New York: Teachers College Press.
Bachman, J. G., & O’Malley, P. M. (1986). Self-concepts, self-
esteem, and educational experiences: The frog pond revisited
(again). Journal of Personality and Social Psychology, 50,
35–46.
Block, J., & Robins, R. W. (1993). A longitudinal study of
consistency and change in self-esteem from early adolescence
to early adulthood. Child Development, 64, 909–923.
Blyth, D., Simmons, R. G., & Carlton-Ford, S. (1983). The
adjustment of early adolescents to school transitions. Journal
of Early Adolescence, 3, 105–120.
Brand, S. Felner, R. D., DuBois, D. L., Mulhall, P. M., & Flowers, N.
(1994). School climat e in middle schools; Development and
prelimiary validation of the PSCS. Unpublished Manuscript.
Brand, S., Felner, R., Shim, M., Seitsinger, A., & Dumas, T. (2003).
Middle school improvement and reform: Development and
validation of a school-level assessment of climate, cultural
pluralism, and school safety. Journal of Educational Psychology,
95, 570–588.
Brown, L. M., & Gilligan, C. (1992). Meeting at the Crossroads.
Cambridge, MA: Harvard University Press.
Bryk, A. S., & Driscoll, M. E. (1988). The high school as community:
Contextual influences and consequences for students and
teachers. Madison: National Center on Effective Secondary
Schools, University of Wisconsin.
Cicchetti, D., & Toth, S. L. (1998). The development of depression in
children and adolescents. American Psychologist, 53 , 221–241.
Cohen, J. (1988). Statistical power analysis for the behavioral
sciences (2nd ed.). Hillsdale, NJ: Erlbaum Associates.
Conchas, G., & Noguera, P. (2004). Explaining the academic success
of African American males. In N. Way & J. Chu (Eds.),
Am J Community Psychol (2007) 40:194–213 211
123
Adolescent development in context (pp. 134–176). New York:
New York University Press.
Connell, J. P., & Ryan, R. M. (1984). A developmental theory of
motivation in the classroom. Teacher Education Quarterly, 11,
64–77.
Connell, J. P., & Wellborn, J. G. (1991). Competence, autonomy, and
relatedness: A motivational analysis of self-system processes. In
M. R. Gunnar & L. A. Sroufe (Eds.), Self processes and
development. The Minnesota symposium on child psychology.
(pp. 43–77). Hillsdale, NJ : Lawrence Erlbaum Associates, Inc.
Crick, N., & Rose, A. (2000). Toward a gender-balanced approach to
the study of social-emotional development: A look at relational
aggression. In P. Miller & E. Kofsky Scholnick (Eds.), Toward a
feminist developmental psychology (pp. 153–168). Florence, KY:
Routledge Press.
DuBois, D. L., & Felner, R. D. (1991). Reliability and validity data
for the self-esteem questionnaire. Unpublished raw data.
DuBois, D. L., Felner, R. D., Brand, S., Phillips, R. S. C., & Lease, A.
M. (1996). Early adolescent self-esteem: A developmental-
ecological framework and assessment strategy. Journal of
Research on Adolescence, 6, 543–579.
Eccles, J., Barber, B., Jozefowicz, D., Malenchuk, O., Vida, M.
(1999). Self-evaluations of competence, task values, and self-
esteem. In N. Johnson & M. Roberts (Eds.), Beyond appearance:
A new look at adolescent girl (pp. 53–83). Washington, DC:
American Psychological Association.
Eccles, J. S., & Midgley, C. (1989). Stage/environment fit: Devel-
opmentally appropriate classrooms for early adolescents. In R. E.
Amers & C. Amers (Eds.), Research on motivation in education
(pp. 139–186). San Diego, CA: Academic Press.
Eccles, J. S., Midgley, C., Buchanan, C. M., Wigfield, A., Reuman,
D., & Mac Iver, D. (1993). Development during adolescence:
The impact of stage/environment fit. American Psychologist, 48,
90–101.
Eccles, J., & Roeser, R. (1999). School and community influences on
human development. In M. Bornstein & M. Lamb (Eds.),
Developmental psychology: An advanced textbook (4th ed., pp.
503–554). Mahwah, NJ: Erlbaum Press.
Epstein J., & Karweit N. (Eds.). (1983). Friends in school. New York:
Academic Press.
Felner, R., Jackson, A., Kasak, D., Mulhall, P, Brand, S., & Flowers,
N. (1997). The impact of school reform for the middle grades: A
longitudinal study of a network engaged in Turning Points-based
comprehensive school transformation. In R. Takanishi & D.
Hamburg (Eds.), Preparing adolescents for the twenty-first
century: Challenges facing Europe and the United States (pp.
38–69). New York: Sage Press.
Fine, M. (1991). Framing dropouts: Notes on the politics of an urban
public high school. New York: SUNY Press.
Griffith, J. (1995). An empirical examination of a model of social
climate in elementary schools. Basic and Applied Social
Psychology, 17, 97–117.
Haynes, N., & Emmons, C. (1994). The role of school climate in
students’ psycho-social development. New Haven, Ct: Lecture
given at invited talk at Yale University.
Hoge, D. R., Smit, E. K., & Hanson, S. L. (1990). School experiences
predicting changes in self-esteem of sixth- and seventh-grade
students. Journal of Educational Psychology, 82, 117–127.
Jo
¨
reskog, K., & So
¨
rbom, D. (2001). LISREL 8.30 [Computer
Software]. Chicago, IL: Scientific Software International, Inc.
Kasen, S., Johnson, J., & Cohen, P. (1990). The impact of school
emotional climate on student psychopathology. Journal of
Abnormal Child Psychology, 18, 165–177.
Kline, R. (1998).
Principles and practice of structural equation
modeling: Methodology in the social sciences. New York, NY:
Guilford Press.
Kovacs, M. (1980/1981). Affective disorders in children and adoles-
cents. American Psychologist, 44, 209–215.
Kozol, J. (1992). Savage inequalities: Children in America’s schools.
New York: Harper Perennial Press.
Kuperminc, G., Leadbeater, B. J., Emmons, C., & Blatt, S. J. (1997).
Perceived school climate and problem behaviors in middle-
school students: The protective function of a positive educational
environment. Journal of Applied Developmental Science, 1,
76–88.
Nolen-Hoeksema, S., & Girgus, J. S. (1994). The emergence of
gender differences in depression during adolescence. Psycho-
logical Bulletin, 115(3), 424–443.
Orenstein, P. (1994). Schoolgirls: young women, self esteem, and the
confidence gap. Wilmington, NC: Anchor Publishers.
Reid, K. (1983). Retrospection and persistent school absenteeism.
Educational Research, 25, 110–115.
Roeser, R. W., & Eccles, J. S. (1998). Adolescents’ perceptions of
middle school: Relation to longitudinal changes in academic and
psychological adjustment. Journal of Research on Adolescence,
8, 123–158.
Roeser, R., Eccles, J., & Sameroff, A. (1998). Academic and
emotional functioning in early adolescence: Longitudinal rela-
tions, patterns, and prediction by experience in middle school.
Developmental and Psychopathology, 10, 321–352.
Roeser, R., Eccles, J., & Sameroff, A. (2000). School as a context of
early adolescents’ academic and social-emotional development:
A summary of research findings. Elementary School Journal,
100, 443–471.
Rosenthal, R., & Rosnow, R. L. (1992). Essentials of behavioral
research: Methods and data analysis (2nd ed.). New York:
McGraw Hill.
Rutter, M., Maughan, B., Mortimore, P., Ouston, J., & Smith, A.
(1979). Fifteen thousand hours: Second schools and their effects
on children. Cambridge, MA: Harvard University Press.
Rys G., & Bear, G. (1997). Relational aggression and peer relations:
Gender and developmental issues. Merrill-Palmer Quarterly, 43,
87–106.
Sadker, M., & Sadker, D. (1994). Failing at fairness: How America’s
schools cheat girls. New York: Charles Scribner’s Sons.
Sayer, A. G., & Willett, J. B. (1998). A cross-domain model for
growth in adolescent alcohol expectancies. Multivariate Behav-
ioral Research, 33, 509–543.
Schafer, J. L. (2000). NORM: Multiple imputation of incomplete
multivariate data under a normal model, version 2.03 [Computer
Software]. Available: http://www.stat.psu.edu/* jls/misoftwa.
html.
Seidman, E., Allen, L., Aber, L., & Mitchell, C. (1994). The impact of
school transitions in early adolescence on the self-esteem and
perceived social context of poor urban youth. Child Develop-
ment, 65, 507–522.
Skinner, E. A., Wellborn., J. G., & Connell, J. P. (1990). What it takes
to do well in school and whether I’ve got it: A process model of
perceived control and children’s engagement and achievement in
school. Journal of Educational Psychology, 82, 22–32.
Skinner, E. A., & Wellborn, J. G. (1997). Children’s coping in the
academic domain. In S. A. Wolchik & I. N. Sandler (Eds.),
Handbook of children’s coping: Linking theory and intervention.
Issues in clinical child psychology (pp. 387–422). New York:
Plenum Press.
Sommer, B. (1985). What’s different about truants? A comparision
study of eight graders.
Journal of Youth and Adolescence, 14 ,
411–422.
Suarez-Orozco, C., & Qin-Hilliard, D. (2004). Immigrant Boys’
Experiences in U.S. Schools. In N. Way & J. Chu (Eds.),
Adolescent boys: Exploring diverse cultures of boyhood (pp.
295–316). New York: New York University Press.
212 Am J Community Psychol (2007) 40:194–213
123
Trickett, E., & Moos, R. H. (1973). The social environment of junior
high and high school classrooms. Journal of Educational
Psychology, 65, 93–102.
Way, N. (1998). Everyday courage: The lives and stories of urban
teenagers. New York: New York University Press.
Way, N. & Robinson, M. G. (2003). Effects of perceived family,
friends, and school experiences on change in self-esteem among
urban, low-SES adolescents. Journal of Adolescent Research,
18, 324–346.
Willett, J. B., & Keiley, M. K. (2000). Using covariance structure
analysis to model change over time. In H. E. A. Tinsley & S.
Brown (Eds.), Handbook of applied multivariate statistics and
mathematical modeling: A comprehensive guide for applied
researchers in the biological sciences, social sciences, and
humanities. San Diego, CA: Academic Press.
Willet, J. B., & Sayer, A. G. (1996). Cross-domain analyses of change
over time: Combining growth modeling, covariance structure
analysis. In G. A. Marcoulides, & R. E. Schumacker (Eds.),
Advanced structural equation modeling (pp. 125–157). Hillsdale,
NJ: Erlbaum.
Am J Community Psychol (2007) 40:194–213 213
123
