See discussions, stats, and author profiles for this publication at: [Link]

net/publication/5876976

Students’ Perceptions of School Climate During the Middle School Years:

Associations with Trajectories of Psychological and Behavioral Adjustment

Article in American Journal of Community Psychology · January 2008

DOI: 10.1007/s10464-007-9143-y · Source: PubMed

CITATIONS READS

546 2,406

3 authors, including:

Niobe Way
New York University
93 PUBLICATIONS 8,139 CITATIONS

SEE PROFILE

All content following this page was uploaded by Niobe Way on 15 May 2014.

The user has requested enhancement of the downloaded file.

Am J Community Psychol (2007) 40:194–213
DOI 10.1007/s10464-007-9143-y

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 Introduction

used to examine the trajectories of change in student per-
ceptions of four critical dimensions of school climate (i.e., Middle schools play an important role in the development
teacher support, peer support, student autonomy in the of adolescents. They provide a context in which students
classroom, and clarity and consistency in school rules and learn, implicitly and explicitly, about themselves and
regulations) among 1,451 early adolescents from the relationships, and about how to navigate the turbulent
beginning of sixth through the end of eighth grade; and the waters of early adolescence. There is considerable evi-
effects of such trajectories on the rate of change in psy- dence that the interpersonal, organizational, and
chological and behavioral adjustment. Findings indicated instructional ‘‘climate’’ of middle schools strongly influ-
that all of the dimensions of perceived school climate ences students’ adjustment across multiple domains
declined over the 3 years of middle school. Furthermore, (Eccles and Roeser 1999; Roeser et al. 1998, 2000; Ku-
declines in each of the dimensions of perceived school perminc et al. 1997; Skinner and Wellborn 1997). Middle
climate were associated with declines over time in psy- school climates often challenge early adolescents’ adap-
chological and behavioral adjustment. Moreover, the tive capacities and account for at least some of the
direction of effects between each dimension of perceived difficulties that arise during this developmental stage
school climate and psychological or behavioral adjustment (Roeser et al. 1998). Despite growing interest in students’
were often unidirectional rather than bi-directional, perceptions of their middle schools, and evidence for the
underscoring the role of perceived school climate in the influence of such perceptions on psychological and
psychological and behavioral health of early adolescents. behavioral adjustment, our understanding of this construct
Gender and socioeconomic class differences in these pat- and its association with adjustment rests on a relatively
terns are noted. small base of empirical findings (Bachman and O’Malley
1986; Brand et al. 2003; Hoge et al. 1990; Kuperminc
Keywords School climate ! Adolescent adjustment ! et al. 1997; Roeser and Eccles 1998; Roeser et al. 1998;
Middle school ! Latent growth curve modeling 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-
N. Way (&)
Department of Applied Psychology, New York University, ered self-esteem, and behavioral difficulties often
Steinhardt School of Education, East Building, Suite 400, 239 coincides with the middle-school years (Blyth et al. 1983;
Greene Street, New York, NY 10003, USA Eccles et al. 1999; Seidman et al. 1994).
e-mail: nw4@[Link]
Moreover, studies of school climate have predominantly
R. Reddy ! J. Rhodes focused on how students’ perceptions change during the
University of Massachusetts, Boston, MA, USA transition from elementary to middle school, as opposed to

123
Am J Community Psychol (2007) 40:194–213 195

the changing patterns of students’ perceptions during Wellborn 1991; Eccles et al. 1993). An important compo-
middle school itself. Such research has also typically nent of these models linking school climate with
treated students’ perceptions of school climate as static, psychological and behavioral adjustment is that it is the
baseline predictors of other outcomes variables (see Brand adolescents’ perceptions or experiences of the school
et al. 2003; Roeser and Eccles 1998; Kuperminc et al. environment rather than a more objective account of the
1997; Way and Robinson 2003). Yet subtle factors, such as school that is most relevant for understanding adolescents’
shifts in student–teacher relationships or peer relationships, adjustment and wellbeing (Connell and Ryan 1987; Con-
can easily influence adolescents’ perceptions of their nell and Wellborn 1991; Eccles et al. 1993).
schools. These changes in perceptions of school climate, in The empirical research examining the association
turn, are likely to have various psychological and behav- between students’ perceptions of the teacher support, peer
ioral implications. support, student autonomy, and clarify and consistency in
In this study, we used cross-domain latent growth curve school rules on psychological or behavioral adjustment has
models to examine trajectories of change in students’ suggested that they are strongly associated with one
perceptions of four critical components of school climate another (Brand et al. 2003; Kuperminc et al. 1997; Roeser
(i.e., teacher support, peer support, student autonomy in the et al. 1998; Hoge et al. 1990; Way and Robinson 2003).
classroom, and clarity and consistency in school rules) For example, Roeser et al. (1998) found that students’
among early adolescents from the beginning of sixth perceptions of their schools’ learning environment (i.e.,
through the end of eighth grade, and the effects of such teacher support, and support for student autonomy) during
trajectories on the rate of change in psychological and seventh grade significantly predicted change over time in
behavioral adjustment. This investigation is a secondary emotional functioning from seventh to eighth grade, even
analysis of a data set drawn from the Illinois Center for after accounting for demographic characteristics. In other
Prevention Research (see Felner et al. 1997). 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-
Theoretical and Empirical Background omy during the seventh grade. Similarly, Hoge et al.
(1990) found that more positive perceptions of school cli-
Theorists and researchers have underscored the importance mate, including perceptions of teacher support, predicted
for middle school students’ wellbeing of four aspects of increases in global self-esteem from fall to spring during
school climate: (1) the nature of relationships between the sixth and seventh grade, after accounting for IQ, sex,
teachers and students; (2) the nature of relationships race, single parent family, and socioeconomic status.
between students; (3) the extent to which student autonomy Others have also found significant associations between
is allowed in the decision-making process; and (4) the teacher and/or peer support, opportunities for autonomy,
extent to which the school provides clear, consistent, and and clarity and consistency in school rules and students’
fair rules and regulations (Bachman and O’Malley 1986; emotional adjustment (see Brand et al. 2003; Bachman and
Connell and Wellborn 1991; Epstein and Karweit 1983; O’Malley 1986; Roeser et al. 1998; Kuperminc et al. 1997;
Kuperminc et al. 1997; Haynes and Emmons 1994; Roeser Way and Robinson 2003).
and Eccles 1998; Roeser et al. 1998; Skinner and Wellborn In addition, linkages between teacher and/or peer sup-
1997). These elements of the school climate (i.e., related- port, opportunities for autonomy, and clarity and
ness, opportunities for autonomy, and clarity/consistency) consistency in school rules and behavior problems have
are considered particularly important for students during been reported (Brand et al. 2003; Bryk and Driscoll 1988;
the middle school years because they match with the Kuperminc et al. 1997; Rutter et al. 1979; Sommer 1985).
developmental needs of early adolescents (Eccles and For example, Kasen et al. (1990) found that an increase in
Midgley 1989; Eccles et al. 1993; Roeser et al. 1998; students’ perceptions of autonomy in the classroom in
Connell and Wellborn 1991). Researchers have noted that middle school over one-year period predicted decreases in
if there is not a ‘‘person-environment fit’’ or the school is alcohol use, conduct problems, and oppositional behavior.
not perceived by the students as supporting their need for In their study of 188 middle schools, Brand et al. (2003)
relatedness, autonomy, and consistency,1 their psycholog- found that students’ perceptions of different dimensions of
ical and behavioral health will be at risk (Connell and school climate, including teacher and peer support (and
negative peer interaction), was significantly associated with
1
Connell and Wellborn (1991) refer to a psychological need for self-esteem, depressive symptoms, delinquency, and sub-
sense of relatedness, autonomy, and competency among students
stance use.
rather than relatedness, autonomy, and consistency. However, they
note that a sense of competency will often emerge in an environment Taken together, these studies shed light on the range of
that is fair and consistent. psychological and behavioral outcomes associated with

123
196 Am J Community Psychol (2007) 40:194–213

students’ perceptions of middle schools environments and 1997) and may lead girls to grow increasingly dissatisfied
underscore the importance of schools addressing the psy- with their peers—a key component of the interpersonal
chological needs of early adolescence (Connell and dimension of school climate.
Wellborn 1991). To a certain extent, however, the statis- Along similar lines, lower SES students may hold more
tical methods that have been most commonly employed negative views of their schools than their higher SES
have masked patterns of change and adjustment over time. counterparts, as the former groups are more likely to attend
These techniques, which have spanned from correlational under-funded and understaffed schools that have few
and mean level analyses of groups of students (e.g., resources to create positive learning environments (Alvi-
ANOVA’s, MANOVA’s), to more complex grouping and drez and Weinstein 1999; Conchas and Noguera 2004;
structural models, rest on a more static view of a period Kuperminc et al. 1997; Way 1998). Researchers over the
that is typically characterized by heterogeneity and flux. past three decades have documented the negative percep-
Snapshot views of this developmental phase do not ade- tions of school evident among many low SES students
quately describe the different patterns over time of (Anyon and Wilson 1997; Fine 1991; Kozol 1992; Way
perceived school climate nor do they explain the changing 1998). Furthermore, the perceptions of school climate
nature of psychological and behavioral adjustment. For among lower SES middle school students may become
some students, the middle schools years may be marked by more negative over time, in comparison with higher SES
ongoing patterns of success and emotional wellbeing. Ini- students, as their ability to recognize the limits of their
tial positive reactions to the school environment might school grows stronger (Conchas and Noguera 2004).
portend ongoing opportunities for positive growth and
development. For others, however, a honeymoon period
might gradually give way to disappointment with their Current Study
middle school’s limitations. Still others might experience
transitional struggles that forecast either continued nega- In this study, we sought to examine four questions: (1)
tive perceptions of the school or gradual adaptation to the How do students’ perceptions of teacher support, peer
setting. Each of these scenarios is likely to predict different support, opportunities for autonomy in the classroom, and
levels of emotional and behavioral adjustment. The goal in clarity and consistency of school rules and regulations
this study is to extend the research on school climate by change during the middle school years? (2) Does gender or
tracking the effects of various patterns of perceptions of SES explain variations in perceptions of these four
school climate over time. dimensions of school climate over time? (3) Are changes in
Our work is not only guided by developmental theories perceptions of these four dimensions associated with
such as those by Eccles and colleagues (e.g., Eccles et al. changes in psychological and behavioral adjustment? (4)
1993; Roeser et al. 1998) and motivational theories such as What is the direction of effect between each dimension of
those by Connell and colleagues (Connell and Wellborn perceived school climate and psychological and behavior
1991), but also by an integrated systems perspective that adjustment? Although researchers have tested the hypoth-
recognizes that perceptions of school climate are closely esis that students’ perceptions of school climate influence
tied to various demographic variables such as gender and their psychological and behavioral adjustment, it is also
socioeconomic status (SES). Research, for example, has plausible that students’ adjustment may influence their
suggested that adolescent girls may experience school as a perceptions of their school environments. Researchers have
less supportive environment than boys, stemming from not, however, tested both possibilities in one study.
institutional gender biases, and a greater likelihood of Based on previous research (e.g., Eccles et al. 1993), we
sexual harassment from male peers (AAUW 1992; Brown hypothesized that students’ perceptions of the teacher and
and Gilligan 1992; Orenstein 1994; Sadker and Sadker student support, opportunities for autonomy, and consis-
1994). These difficult experiences may translate into more tency and clarity of school rules would become more
negative perceptions of school. In addition, since the negative during the middle school years. We also expected
harassment from peers often increases during middle that boys and higher SES students would report more
school as girls reach puberty, girls’ perceptions of school positive perceptions of the school environment and less
climate may be more likely than boys to decline over time sharp declines over time than girls and lower SES students,
(Orenstein 1994). Furthermore, increasing levels of rela- respectively. Finally, we hypothesized that perceptions of
tional aggression among girls in middle schools has been school climate would influence adjustment, rather than the
noted by researchers (Crick and Rose 2000; Rys and Bear reverse (see Fig. 1).

123
Am J Community Psychol (2007) 40:194–213 197

Fig. 1 Perceived middle school Clarity of Consistency

Teacher Peer Opportunities for and fairness
environment: Theoretical model Support Support Student Autonomy School Rules

School School
School Instructional Organizational
Interpersonal Climate Climate
Climate

Behavior Problems General Self-Esteem Depression

Method At the beginning of middle school, 20% of the sample

reported receiving free or reduced-price lunches. A little
Source of Data over one-third of the students reported that their mothers
had completed college (37.4%) and over half reported that
Data were drawn from a larger ongoing longitudinal study their fathers had two or more years of college (17%) or had
of the role of educational environments and state-wide completed college (41.6%). The majority of the adoles-
change efforts on early adolescents in middle schools.2 The cents (85%) also reported living in two-parent families.
schools that participated in this study were all members of The students were drawn from 22 Midwestern schools all
a statewide middle school association, which serves stu- housing grades six through eight. The schools represented
dents from geographically and socio-economically diverse diverse geographical locales with half of the schools in
backgrounds. Administrators in the schools agreed to par- small town or rural areas and 40.8% situated in urban areas.
ticipate in the study as part of statewide school assessment Six grade enrollment in each school ranged from 68 to 414
plan. The assessment was not tied to a particular students.
intervention.
The original study included 2,860 students from 30
schools. The response rate ranged from 70 to 90%. We Procedures
eliminated those students who had missing information and
those who did not have complete data for all 3 years of the Data collection occurred in participating schools during the
study. Since we sought to understand changes during the Spring of 1995, 1996, and 1997. The IRB at the partici-
3 years of middle school, we also eliminated schools that pating University reviewed and approved Human Subject
did not have grades six, seven, and eight. The current study procedures.3 The informed consent of students and their
focused on the 1,451 middle school students (from 22 parents was obtained and confidentiality was assured.
middle schools) who had complete data from the three Teachers distributed the surveys in classrooms to students,
consecutive years of the study, 1995–1997, when the stu- and read them aloud to reduce disparities in reading level.
dents were in sixth, seventh, or eighth grade. Teachers were always present during survey administra-
tion. The surveys were typically completed in two class
3
Participants 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
At Time 1, the sample consisted of 1,451 early adolescents each school to create a Parent Advisory Team (PAT) that was
who had begun sixth grade in 1995. Slightly over half the reflective of the children and families attending that school. The PAT
sample was female (54.2%) and the majority was European was asked to review the surveys and consent form that would
American (91%). Eligibility to receive free or reduced- subsequently be sent home to parents for their approval. The PAT was
to consider community norms and values when reviewing the topics
price lunches was taken as a marker of low-income status. 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
2
We gratefully acknowledge the contribution of the researchers at parents who might have questions supporting the surveys, and copies
the Illinois Center for Prevention Research and of Robert D. Felner, were kept in the school from office for parental review. Once the PAT
who served as Principal Investigator of the Alliance of Illinois Middle signed the advisory form, schools were required to send home letters
Schools network evaluation. The evaluation was funded by a to parents and were asked to return the form if they did not want their
Carnegie Foundation grant to the University of Illinois. child to participate.

123
198 Am J Community Psychol (2007) 40:194–213

periods over a 2-day period. The schools determined the the composite score of the two subscales was used. Higher
class to administer the survey. Some schools administered scores on this dimension indicate a positive peer climate.
the survey during homeroom while others used a class Internal consistency for the combined scale for the 3 years
within the school day. Once surveys were completed, of the present study was .77, .72, and .75, respectively.
surveys were collected and returned to the university for
processing.
Opportunities for Student Autonomy

Measures This construct was assessed through the 5-item student

decision-making subscale of the Perceived School Climate
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 order to examine school climate, we examined teacher in how things worked in school, helped decide how class
support, peer support, student autonomy and clarity and time was spent, given a chance to help make decisions, and
consistency in school rules (see Appendix A). These four asked what they would like to learn about. Higher scores
dimensions were assessed through subscales of the Per- imply a greater degree of student decision-making. Internal
ceived School Climate Scale (1993) developed by the consistency for the 3 years of the present study was .70,
Center for Prevention Research and Development (CPRD) .72, and .75, respectively.
(see Felner et al. 1997). The questionnaire is based on the
Classroom Environment Scale (Trickett and Moos 1973), a
widely used and well-validated measure. Changes to the Clarity and Consistency in School Rules
Classroom Environment Scale involved rewording items to
eliminate double negatives and providing more familiar Two subscales of the Perceived School Climate Scale,
language for colloquial terms. The Perceived School Cli- school structure and school harshness, were combined to
mate Scale has been found to be a valid and reliable assess clarity and consistency in school rules and regula-
measure of school climate (see Brand et al. 1994, 2003). 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
Teacher Support scale (1 = never to 5 = always), indexed the degree to
which students perceived clarity in the school rule structure
The six items on this subscale of the Perceived School and their perceptions of arbitrary or punitive discipline
Climate Scale indexed students’ perceptions of teacher practices in the school. The harshness subscale was reco-
support in the school. Students rated on a 5-point scale ded and then a composite subscale score was obtained.
(1 = never to 5 = always) how true statements were for Higher scores on this dimension implied perceptions of
their school such as teachers go out of their way to help clarity, consistency, and fairness in the school. Internal
students and teachers take a personal interest in students. consistency for the combined scale for the 3 years of the
Higher scores indicate higher levels of teacher support. present study was .62, .65, and .59, respectively.
Internal consistency of the scale for the 3 years of the
present study was .74, .74, and .79, respectively.
Indices of Wellbeing

Peer Support Behavior Problems

Two subscales of the Perceived School Climate Scale The Behavior Problems Scale was based on the Delin-
assessed adolescent’s perceptions of peer support: negative quency Scale of the Youth Self-Report (Achenbach and
peer interactions and affiliation. The negative peer inter- Edelbrock 1987). The scale is a shortened version of the
actions subscale (Time 1 a = .77, Time 2 a = .77, Time 3 original scale indexing frequency rather than severity of
a = .82) and the affiliation subscale (Time 1 a = .69, Time problem behavior. The scale was shortened by The CPRD
2 a = .70, Time 3 a = .78) assessed on a 5-point scale (Felner et al. 1997). The 10-item measure indexed on a 5-
(1 = never to 5 = always) how well students in the school point scale (1 = never to 5 = more than 12 times) how
got along with each other, enjoyed working together, often students, over the past 6 months, reported having
helped each other, or were mean to each other. The items committed a problem behavior, including telling lies,
on the negative peer interaction subscale were recoded and breaking rules at school, cutting classes or skipping school,

123
Am J Community Psychol (2007) 40:194–213 199

hitting other people, and acting mean toward others. Cross-domain modeling capitalizes on the richness of two
Internal consistency over 3 years for the current study was methodologies, individual growth modeling, and covari-
a = .85, a = .88, and a = .90 respectively. 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
Self-esteem of testing inter-relationships among these changes. For
instance, one can investigate not only whether individual
Self-esteem was indexed by the six-item general self- differences in change in a particular domain are related to a
esteem subscale of the Self-Evaluation Questionnaire specific set of predictors but also if change in one domain is
(SEQ, DuBois and Felner 1991; DuBois et al. 1996). related to change in another.
DuBois et al. (1996) have reported convergent and dis- Individual growth models specify ‘‘intercept’’ and
criminant validity of the scale. Students rated statements ‘‘slope’’ growth parameters and define within-person or
such as ‘‘I like being the way I am,’’ ‘‘I am happy with level-1 models. The intercept parameter represents initial
myself as a person,’’ ‘‘I am as good a person as I want to status as defined by the investigator whereas the slope
be,’’ on a 4-point scale (strongly disagree to strongly agree) parameter represents the rate of change over time.
with higher ratings indicating higher levels of self-esteem. Covariance structure analyses then allows the specification
Internal consistency of the subscale for the 3 years was and test of relationships among different individual growth
a = .83, a = .84, and a = .86, respectively. 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
Depressive Symptoms equation modeling terms, each level-1 model can be treated
as an exogenous or endogenous measurement models. The
Depressive symptoms were indexed with a shortened ver- repeated observations of each construct form indicators
sion of the Children’s Depressive Inventory (Kovacs 1980/ that delineate two latent growth parameters, the intercept
1981). The sixteen-item self-report scale measures cogni- and slope. An important step in defining both growth
tive, affective, and behavioral symptoms of depressive parameters is choosing a suitable mathematical model to
symptoms. For each item, students selected one of three represent change over time. Since we were constrained by
statements (increasing in symptom severity) that best only three waves of data, a linear trajectory in growth was
described their feelings and ideas during the previous selected for the current model. As the observations were
2 weeks. The scale yields a single aggregate measure of taken during each academic year, this defined a time
depressive symptomology reflecting an underlying unidi- interval of t = 1, 2, 3. In order for the intercept to represent
mensional construct. Lower scores indexed lower levels of status in the sixth grade, we rescaled the time metric by
reported depressive symptoms. Reliability for the 3 years subtracting one (so that t = 0, 1, 2). Apart from ensuring
was a = .85, a = .80, and a = .89, respectively. 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
Demographic Characteristics defined by the first latent factor and represents the initial
levels or status of the measures when the adolescent is in
Single items were used to assess students’ grade level, the sixth grade. The second growth parameter, slope, is
gender (female coded 1), whether they received free lunch represented by the equal spaced loadings (i.e., 0, 1, and 2)
or reduced price lunch at school (free or reduced-price of the observed variables and defines the linear rate of
lunch coded 1), race or ethnicity (minority status coded 1), change of growth over time. Note that all individuals of the
father and mother’s educational level, and with whom they sample are assumed to have the same linear form of growth
have lived in the past year. 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
Analytic Strategy the estimate of the average true intercept and slope in the
population. Heterogeneity in growth is captured by the
Cross-domain growth modeling was used to test the variances of the intercept and slope latent factors. Predic-
hypotheses of the current study. Technical discussions tors of this variability can thus be easily introduced using
(e.g., Sayer and Willett 1998; Willet and Keiley 2000) as the structural equation modeling (SEM) framework by
well as longitudinal applications (e.g., Willett and Sayer regressing the predictors on the intercept and slope latent
1996) of this methodology are available in the literature. factors.

123
200 Am J Community Psychol (2007) 40:194–213

The power of the cross-domain analyses is obtained list-wise deletion of the student data also resulted in the
when such individual latent growth models are combined elimination of one school from the sample. The eliminated
together. In SEM terminology, these combined models school had a greater percentage of African American stu-
represent structural models that specify relationships dents than the retained schools in the sample. Ethnic
between the individual growth models. These combined variability in school composition was controlled for by
models therefore define between-person or level-2 models. controlling for minority status in the combined models.
These models can further be expanded by adding specific To test the nature of any probable bias in the list-wise
person-level characteristics that are used as predictors of elimination process, multiple imputation procedures were
initial status and growth. conducted using NORM software (Schafer 2000). Pre-
In the case of the current study, we were interested in (a) liminary data screening revealed percent of missing data
unconditional models that would highlight the average for each variable in the analyses, which ranged from .80 to
trajectories of change over time for each dimension of 7.87%. Four imputed data sets were created using a two-
school climate as well as each outcome variable, (b) the step process under the multivariate normal model. These
effects of student gender and socioeconomic status on both data sets were then utilized to obtain parameter estimates
the initial levels and slopes of each dimension of perceived for the unconditional models, the conditional models for
school climate, (c) the associations among changes in the the school climate dimensions, and the final combined and
different dimensions of school climate perceptions and the alternative models reported in the study. Results from each
concomitant changes in self-esteem, depressive symptoms, of the four separate runs (one for each of the imputed data
and behavior problems, and (d) models that examined the sets) were then re-combined in NORM. The results indi-
alternative hypothesis, namely, that the direction of effects cated that the patterns of outcomes were similar for the
between each dimension of perceived school climate and imputed and the reduced sample models. Given overall
adjustment is bi-directional rather than unidirectional. similarities and our desire to maintain parsimony in
Secondary to these analyses, we also report the nature of reporting, the findings reported in the study rely on the
the relation between initial levels of the school climate reduced sample size (N = 1,451). All models were esti-
dimensions and initial levels of depressive symptoms, mated using LISREL 8.30 (Jöreskog and Sörbom 2001).
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, Descriptive Analysis
and behavior problems. Cross-domain growth modeling
was selected for the current study because it allows us to Means, standard deviations, univariate skewness coeffi-
fully examine each of our research questions. All models cients, and correlations among the demographic variables,
were estimated using LISREL 8.50 (Jöreskog and Sörbom perceptions of each dimension of school climate and the
2001). In the LISREL model, the standardized regression outcome variables in the model are presented in Table 1. The
coefficient is the effect size (Kline 1998). As suggested by effect sizes of the correlations are presented in Table 2. The
Kline, the interpretation of the absolute magnitudes of path overall observed means show declines in each of the
coefficients can follow the general recommendations by dimensions of school climate and in the wellbeing variables
Cohen (1988). Standardized path coefficients with absolute during middle school. The skewness coefficients ranged
values less than .10 indicate a ‘‘small’’ effect; values from –.52 to 1.86, none indicating significant deviations
around .30 a ‘‘medium’’ effect; and values of .50 and above from symmetry. The four dimensions of school climate were
reflect a ‘‘large’’ effect. 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
Results dimensions of school climate and behavior problems,
depressive symptoms, and general self-esteem were in the
Treatment of Missing Data expected direction. Medium to large effect sizes were typi-
cally indicated for each dimension of school climate on
Comparison of the retained and eliminated sample revealed behavioral or psychological adjustment. Gender was con-
that students who were retained for analysis were more sistently, positively correlated with and indicated medium
likely to be female, white, and have mothers who were effect sizes on each dimension of school climate except
college educated. Furthermore these students were also student decision-making. Surprisingly, no significant corre-
more likely to come from two-parent families. The lation between gender and self-esteem or depressive

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 –
Am J Community Psychol (2007) 40:194–213

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
201

123
202 Am J Community Psychol (2007) 40:194–213

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 – r2) (Rosenthal and Rosnow 1992). Absolute
values displayed

symptoms was obtained. Socioeconomic status was posi- climate, for the unconditional models we did not examine
tively related to all 3 years of depressive symptoms. Apart the influence of gender or SES on our three outcome
from this latter finding, SES did not show any consistent variables.
pattern of association with the other variables in the model.

Change Over Time in Perceived School Climate

Initial Analytical Step: Unconditional Models
Results from the unconditional models can be seen in
To examine how students’ perceptions of school climate as Table 3. For the dimensions comprising teacher support
well as indices of psychological and behavioral adjustment and peer support, the models fit the data reasonably well.4
change over the course of middle school, seven uncondi- The average growth trajectory revealed significant (i.e.,
tional individual growth models were estimated, one each different from zero) levels of perceptions of teacher sup-
for the repeated measures of the four dimensions of per- port and peer support in the sixth grade. Students had
ceived school climate (i.e., teacher support, peer support, positive perceptions of both teacher and peer support at the
opportunities for student autonomy, and clarity and con- start of sixth grade. However, as students progressed
sistency in school rules), behavioral problems, depressive through middle school, their perceptions of teacher support
symptoms, and general self-esteem. This permitted the and peer support declined significantly every year (as can
examination of average growth trajectories as well as the be seen in the negative and significant slopes). Significant
presence of individual variability about the average growth heterogeneity in the initial status (at the sixth grade) and
parameters. Furthermore, we also examined the role of slope of individual trajectories of teacher and peer support
gender and SES on each dimension of school climate by were also found.
regressing these predictors on the intercept and growth
latent factors. Since the focus of our paper is on the pre- 4
In none of the unconditional models were the errors allowed to
dictors and effects of four dimensions of perceived school correlate over time, which may have resulted in better fitting models.

123
Am J Community Psychol (2007) 40:194–213 203

Table 3 Model fit and coefficients for the unconditional models

Model v2 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
2
Note: The v 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

Table 4 Model fit and estimates for the influence of gender and SES on school dimensions
Model v2(d.f.) RMSEA (CI) CFI v2diff (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

The models for student autonomy and clarity and con- The impact of gender and/or SES on each dimension of
sistency of school rules also fit the data reasonably well. school climate indicated a small to medium effect size as
The average growth trajectory revealed significant (i.e., indicated by the path coefficient.
different from zero) levels of perceptions of student For teacher support, girls at the beginning of the sixth
autonomy and clarity and consistency of school rules in the grade reported more teacher support than boys. Similarly,
sixth grade. Moreover, the average trajectory for the slope adolescents from a lower socioeconomic background (i.e.,
indicated that the perceptions of student autonomy as well receiving free or reduced price lunch) reported more
as those for clarity and consistency of school rules declined positive perceptions of teacher support at the sixth grade
over the course of middle school. Significant heterogeneity than students from a higher socioeconomic status (i.e., not
in the initial status and slope of individual trajectories of eligible for free or reduced price lunch). With respect to
student autonomy and clarity and consistency of school peer support, at the start of middle school, girls tended to
rules were also found. 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
Impact of SES and Gender the boys. In addition, students from a lower socioeconomic
background reported less positive peer support at the start
To further examine the nature of the significant individual of middle school than the students from higher socioeco-
variation in initial status and growth over time in the tea- nomic backgrounds, but the change over time did not differ
cher and peer support, student autonomy, and clarity and by SES.
consistency in school rules, the intercept and the slope For student autonomy, gender predicted initial levels
factors of each of the models discussed above were such that girls at the start of middle school perceived
regressed on gender and SES (see Table 4). For each greater opportunities for student involvement in the deci-
model, non-significant pathways were specified to zero and sion-making in their classrooms than boys. There were no
the models were re-fitted. In each case, the re-fitted models SES differences indicated for the intercept or slope of
did not show significant degradation from the originals. opportunities for student autonomy. For clarity and

123
204 Am J Community Psychol (2007) 40:194–213

consistency in school rules, girls perceived greater clarity, perceived school climate and psychological and behavioral
consistency, and fairness at the start of middle school than adjustment (see Fig. 1). To this end, we tested a combined
boys. Students from lower socioeconomic backgrounds model that included the unconditional models presented
reported less clarity, consistency, and fairness at the start of above. Four different models were thus explored; one for
middle school than their more privileged counterparts. each dimension of school climate. The associations among
the growth parameters across the domains were estimated
via regression coefficients between the respective initial
Indices of Wellbeing status and slopes. Specifically, initial status and slopes of
the outcome variables (i.e., behavior problems, general
The unconditional models for behavior problems, general self-esteem, and depressive symptoms) were regressed on
self-esteem, and depression were also tested (see Table 3). the initial status of each of the dimensions of school cli-
The average trajectory for behavior problems revealed the mate to examine the influence of initial levels of the
following. Initial status of behavior problems was signifi- specific dimension of school climate on behavior problems,
cantly different from zero. The estimate of the slope was self-esteem, or depressive symptoms growth trajectories.
positive and also significantly different from zero, indi- To explore how rates of changes each of the dimensions of
cating population average increased in reports of behavior school climate predicted rates of changes in psychological
problems over time. With respect to general self-esteem, a and behavioral outcomes over the course of middle school,
negative significant slope factor indicated declining levels the slope of the outcome variables were regressed on the
of self-esteem over the course of middle school for the slope of school climate dimension. Additionally in each
average population growth trajectory. The average growth model, the initial status and the slope of each of dimension
trajectory for symptoms of depression revealed significant of school climate was regressed on gender, SES, and
levels of depression in the sixth grade. The estimate of the minority status. Although we included minority status in
slope was positive and also significantly different from these models given research that suggests across group
zero indicating an increase in reports of symptoms of variation in the association between school climate and
depression over time. Significant variance in both initial adjustment (see Way and Robinson 2003), we do not
status and slope factors was obtained for all three indices of interpret the nature of this relationship in any meaningful
wellbeing, indicating the presence of heterogeneity in the way given the insufficient sample size for minority students
individual trajectories over time (see Table 3). (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
Summary previous research findings suggesting gender differences in
depressive symptoms, self-esteem, and behavioral prob-
In summary, adolescents in the study tended to perceive lems (e.g., Block and Robins 1993; Cicchetti and Toth
middle schools as increasingly negative over time. Fur- 1998; Nolen-Hoeksema and Girgus 1994) the growth
thermore, they report declining levels of wellbeing, as parameters for the outcome variables (i.e., both status and
indicated by increasing levels of depressive symptoms, slope factors) were regressed on gender.
behavior problems, and decreasing levels of general self-
esteem over the course of middle school. Gender differ-
ences were detected for the initial levels of all four Teacher Support
dimensions of school climate, with girls perceiving more
positive school climates at the start of middle school. Girls The combined model fit the data well (v2(43)
also reported sharper declines in peer support than boys. N = 1,451) = 192.34, RMSEA = .05, CI = .04–.06, CFI =
Socioeconomic differences were also observed for the .98). Neither minority status nor SES significantly pre-
initial levels of teacher support, peer support, and clarity dicted the initial status or slope of teacher support.
and consistency in school rules. There were no SES dif- Additionally, gender did not significantly predict depres-
ferences in the rates of decline for any of the school climate sive symptoms or any of the slopes of the variables in the
dimensions. model. We therefore 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 (v2d(9) =
The Combined Models 15.95, p [ .05). This latter model is thus reported (see
Fig. 2). Gender differences were found with respect to the
The third aim of the current study was to examine the initial status of teacher support (as reported in the uncon-
nature of associations between each of the dimensions of ditional models), general self-esteem, and behavior

123
Am J Community Psychol (2007) 40:194–213 205

Fig. 2 Final combined cross- Gender

domain model for student
perceptions of teacher support.
-.2 6
For the sake of clarity, only -.05 Bh. Problem
significant paths are .20 Intercept
represented. All parameter
values are standardized. Model
v2(52, N = 1,451) = 208.29, -.33 Bh. Problem
Slope
RMSEA = .05, CI = .04–.05, Teacher Support -.21
CFI = .97 Intercept .29
Self-Esteem
Intercept
-.39

Self-Esteem
-.83 Slope

Teacher Support
Slope
Depression
-.77 Intercept

Depression
Slope

problems. At the beginning of sixth grade, girls reported both general self-esteem and depressive symptoms. SES
significantly lower levels of self-esteem, while boys predicted the initial status of peer support (as reported in
reported higher incidences of behavior problems at the start the unconditional models). All non-significant paths in the
of middle school. The intercept of teacher support (the relationships between the demographic variables and the
sixth grade assessment) was significantly associated with other variables in the model were set to zero and the model
the intercepts of self-esteem and depressive symptoms and re-estimated. The chi-square difference test indicated that
with the intercept and slope of problem behavior. the final model was not significantly degraded (v2d(5) =
With respect to our primary question regarding the 5.20, p [ .05) (see Fig. 3). As expected, girls reported
association between changes over time in teacher support lower levels of behavior problems, lower self-esteem, and
on changes over time in depressive symptoms, self-esteem, higher levels of depressive symptoms at the beginning of
or behavior problems, the slope of teacher support was middle school. Boys were more likely to report sharper
significantly (i.e., different from zero) related to the slopes increases in behavior problems over middle school when
of depressive symptoms and behavior problems. A decline compared to the girls. The intercept of peer support was
in perceptions of positive teacher support over time was significantly associated with the intercept and slope of self-
associated with corresponding increases in depressive esteem and depressive symptoms and the intercept of
symptoms and behavior problems. The associations problem behavior.
between the intercept and/or slope of teacher support and With respect to the association between change over
the intercept and/or slope of self-esteem, depressive time in peer support and change over time in adjustment,
symptoms, or problem behavior suggested medium to large the findings indicated that perceptions of declining peer
effect sizes, with the slope of teacher support on the slopes support were associated with declines in psychological and
of problem behavior and depressive symptoms suggesting behavioral adjustment. As students reported declines in
particularly large effect sizes (see Fig. 2). peer support, there were corresponding increases in
depressive symptoms and behavior problems and declines
in general self-esteem. The association between the inter-
Peer Support cept and/or slope of peer support and the intercept and/or
slope of self-esteem, depression, or problem behavior
The model that assessed the degree of influence of peer indicated medium to large effect sizes (See Fig. 3).
support in the school fit the data well (v2(43)
N = 1,451) = 155.23, RMSEA = .04, CI = .03–.05, CFI =
.98). Examination of this model revealed that gender Opportunities for Student Autonomy
significantly predicted the initial status and slope of peer
support (as reported in the unconditional models), initial The combined model for opportunities for student auton-
status and slope of behavior problems, and initial status of omy fit the data well (v2(43) N = 1,451) = 176.75,

123
206 Am J Community Psychol (2007) 40:194–213

Fig. 3 Final combined cross- G en d e r

-.23
domain model for student
perceptions of peer support. For
- .09
the sake of clarity, only .18 Bh. Problem
-.06 I nt ercept
significant paths are .07
-.10
represented. All parameter
values are standardized. Model
-.34 Bh. Problem
v2(48, N = 1,451) = 160.43, Slope

RMSEA = .04, CI = .03–.05, Peer Support

I nt ercept .47
CFI = .98
-.16 Self-Esteem
.34 -.65 I nt ercept

Self-Esteem
Slope
-.40
Peer Support .40
Slope
Depression
-.10 -.75 I nt ercept

Stu dent SES Depression

Slope

RMSEA = .05, CI = .04–.05, CFI = .97). With the excep- changes in depressive symptoms, behavior problems, and
tion of the path from gender to initial status of student self-esteem. As students perceived sharper declines in
autonomy (as reported in the unconditional models) and opportunities for student decision-making in the classroom
behavior problems, gender, SES, and minority status did over time, they also reported increases in depressive
not significantly predict the variables in the model. We symptoms and behavior problems over time, and decreases
therefore set these non-significant paths to zero and re- in self-esteem. The association between the intercept and/
estimated the model. There was no significant degradement or slope of student autonomy and the intercept and/or slope
of the final model (v2d(10) = 14.39, p [ .05) (see Fig. 4). of self-esteem, depressive symptoms, or behavior problems
Boys reported higher levels of problem behaviors than the indicated medium effect sizes (see Fig. 4).
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 Clarity and Consistency in School Rules
intercept and slope of behavior problems.
The rate of change in perceptions of opportunities for The combined model for clarity and consistency in school
student autonomy was significantly related to the rate of rules fit the data well (v2(43) N = 1,451) = 150.05,

Fig. 4 Final combined cross- G e n d er

domain model for student
perceptions of participation in
-.3 0
student decision-making. For Bh. Problem
the sake of clarity, only .07 Intercept

significant paths are

represented. All parameter -.18
values are standardized. Model Bh. Problem
Slope
v2(53, N = 1,451) = 191.14, Student autonomy -.21
RMSEA = .04, CI = .04–.05, Intercept
.24
CFI = .98 Self-Esteem
Intercept
-.28

Self-Esteem
-.34 Slope

Student autonomy .27

Slope
Depression
-.35 Intercept

Depression
Slope

123
Am J Community Psychol (2007) 40:194–213 207

Fig. 5 Final combined cross- G en d e r

-.18
domain model for student
perceptions of school clarity and
consistency in school rules. For .22 Bh. Problem
Intercept
the sake of clarity, only .09
-.10
significant paths are
represented. All parameter
Bh. Problem
values are standardized. Model -.52 Slope

v2(50, N = 1,451) = 158.64, Clarity and consistency in -.27

school rules
RMSEA = .04, CI = .03–.05, Intercept .34
Self-Esteem
CFI = .98 Intercept
-.49

Self-Esteem
-.40 Slope

Clarity and
consistency in school
rules
-.08 Slope Depression
-.75 Intercept

Minority Depression
Status Slope

RMSEA = .04, CI = .03–.05, CFI = .98). SES status did the sixth grade. In addition, students who perceived their
not significantly predict either the initial status or slope of school as having increasingly more teacher and peer sup-
clarity and consistency in school rules. In addition, port, a greater degree of student autonomy, and more
minority status and gender were significantly associated clarity and consistency in school over time also reported
with the intercept but not with the slope of clarity and declines over time in depressive symptoms and behavior
consistency in school rules. Gender also predicted each of problems. Moreover, reports of increasingly positive peer
the outcome variables (i.e., behavior problems, depressive support and a greater degree of student autonomy were
symptoms, and self-esteem). The non-significant paths significantly related to increases over time in student
were next set to zero and the model was re-estimated. reports of general self-esteem. The effect sizes in each
There was no significant degradement of the final model model suggested medium to large effect sizes, with the
(v2d(7) = 8.58, p [ .05). This final model is graphically slope of peer support, teacher support, and clarity and
displayed in Fig. 5. Girls in the sixth grade reported consistency in school rules on the slope of depressive
higher levels of depressive symptoms and lower levels of symptoms suggesting a larger effect size than the other
self-esteem than boys, while boys reported higher levels paths in the model (with the exception of the slope of
of problem behavior than girls. The intercept of clarity teacher support and the slope of problem behavior).
and consistency in school rules was significant associated
with the intercept of self-esteem and depressive symp-
toms and with the intercept and slope of problem Testing for Direction of Effects
behavior.
Our findings with regards to the slope of clarity and To assess whether the effects of each dimension of school
consistency in school rules indicated that for students who climate on psychological and behavioral adjustment were
reported an increase in perceptions of clarity and consis- bi-directional rather than simply unidirectional, a series of
tency in school rules, there was a corresponding decrease in alternative models for each of the four dimensions of
depressive symptoms and behavior problems over time. The school climate were tested. In these models, each dimen-
association between the intercept and/or slope of clarity and sion of school climate was viewed as the outcome variable
consistency in school rules and the intercept and/or slope of and the three indicators of wellbeing (i.e., behavior prob-
self-esteem, depressive symptoms, or behavior problems lems, general self-esteem, and depressive symptoms) were
indicated medium to large effect sizes (See Fig. 5). entered as predictors. Student gender, socioeconomic, and
In summary, students’ perceptions of teacher and peer minority status were included in the model as defined
support, student autonomy, and clarity and consistency in earlier. We could thus make qualitative comparisons of
school rules in sixth grade (initial status) was significantly model effects and gain some degree of insight as to the
associated with students’ reported levels of depressive direction of effects from the different dimensions of school
symptoms, general self-esteem, and behavior problems in climate to wellbeing.

123
208 Am J Community Psychol (2007) 40:194–213

Teacher Support Clarity and Consistency in School Rules

The overall model fit the data as well as the original (v2(43) The overall model for the alternative hypothesis for clarity
N = 1,451) = 172.68, RMSEA = .05, CI = .04–.05, CFI = and consistency in school rules fit the data as well as the
.98). Initial status of behavior problems were significantly original model (v2(43) N = 1,451) = 153.27, RMSEA =
related to the initial status of perceptions of teacher support .04, CI = .03–.05, CFI = .98). Once again, the only sig-
such that students who report higher incidences of behavior nificant path was between the slope of behavior problems
problems in the sixth grade also tended to report less and the slope of clarity and consistency in school rules
positive perceptions of teacher support (b = –.29, (b = –.40, p \ .01). Students who reported an increasing
p \ .001). A significant negative relationship was also incidence of behavior problems over middle school also
found between the slope of behavior problems and the tended to show steeper declines in perceptions of clarity
slope of teacher support such that students who report and consistency in school rules over time. Additionally, the
increasing incidences of behavior problems over middle initial status of depressive symptoms (b = –.26, p \ .05)
school also tended to show steeper declines in perceptions and behavior problems (b = –.33, p \ .001) were signifi-
of teacher support over time (b = –.40, p \ .001). How- cantly related to the initial status of clarity and consistency
ever, neither the slopes of general self-esteem nor in school rules. However, neither the slopes of depressive
depressive symptoms significantly predicted the slope of symptoms nor general self-esteem predicted the slope of
perceptions of teacher support (self-esteem: b = .12, clarity and consistency in school rules.
p [ .10; depressive symptoms: b = –.24, p [ .10). 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
Peer Support
indeed from the dimensions of perceived school climate to
depressive symptoms rather than the reverse. Furthermore,
The overall alternative model for the nature of peer support
the pathway of influence between student decision-making
in the school fit the data as well as the original (v2(43)
and self-esteem appear to be from changes in student
N = 1,451) = 190.39, RMSEA = .05, CI = .04–.05, CFI =
autonomy to changes in self-esteem rather than the reverse.
.98). In this alternative model, apart from the significant
Finally, the pathways of influence for peer support appear
path from the initial status of behavior problems and the
to be from changes in perceptions of peer support to
initial status of peer support (b = –.11, p \ .01), the only
changes in behavior problems and general self-esteem
other significant paths were from the initial status and slope
rather than the reverse.5
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
Discussion
course of middle school also tended to report declining
levels of positive peer support in the school. However,
The major goals of this study were to examine changes in
neither the slopes of general self-esteem nor problem
students’ perceptions of teacher and peer support, oppor-
behavior significantly predicted the slope of peer support.
tunities for student autonomy in the classroom, and clarity
and consistency of school rules during the middle school
Opportunities for Student Autonomy years, the effects of gender and SES on those changes, and
the psychological and behavioral effects of various patterns
The overall model fit the data as well as the original (v2(43) of change over time in the different dimensions of per-
N = 1,451) = 158.26, RMSEA = .04, CI = .04–.05, CFI = ceived school climate. Our unconditional models revealed
.98). As in the model for teacher support, the only sig- that, across gender and SES, perceptions of all four
nificant paths were found for behavior problems. The initial dimensions of school climate declined over the 3 years of
status and slope of behavior problems were significantly middle school. The decline in perceptions of school climate
related to the intercept (b = –.22, p \ .001) and slope may be based on a growing mismatch between what stu-
(b = –.49, p \ .05) of student autonomy, respectively. dents need (i.e., relatedness, opportunities for autonomy,
Students who report increases in behavior problems over and consistency) and what the school actually provides
the course of middle school also tended to show steeper
declines in perceptions of student autonomy. Yet the initial 5
Results from the multiple imputations more strongly indicate that
status or slopes of general self-esteem or depressive
the pathways of influence do emerge from the dimensions of school
symptoms did not significantly predict the initial status or climate to student adjustment implying the need to further study these
slope of student decision-making. processes in a larger sample.

123
Am J Community Psychol (2007) 40:194–213 209

(Eccles et al. 1993; Roeser et al. 1998; Connell and consistent manner), SES does not make a difference in the
Wellborn 1991). It is also possible, however, that middle trajectory of students’ perceptions of school climate over
school students grow increasingly dissatisfied with their time. Understanding when and how SES makes a differ-
schools, irrespective of the actual climate of their schools. ence in students’ perceptions of their school is an important
Numerous researchers have documented the biological, direction for future research.
cognitive, and psychological changes that early adolescents Another primary goal of our analysis was to explore the
endure (Eccles et al. 1993). Such difficult changes may association between rates of change in perceptions of each of
lead early adolescents to perceive their school environ- the four dimensions of school climate and rates of change in
ments as increasingly more negative over time regardless psychological and behavioral adjustment. As hypothesized,
of the actual schools they attend. Future research should the rates of change for teacher and peer support, student
examine both the objective (e.g., how much opportunity do autonomy, and clarity and consistency in school rules were
the students have in their classes to participate in decision- significantly associated with the rates of change for self-
making) and subjective aspects of school climate to assess esteem, depressive symptoms, and problem behavior.
the association between what the school actually provides Declines in teacher support, peer support, student autonomy,
for its students and how the students perceive the school. and clarity and consistency in school rules were associated
Although students reported declines in each of the four with declines in psychological and behavioral adjustment.
dimensions of school climate, girls reported sharper Furthermore, these models of change indicated medium to
declines in peer support than boys over time (although the large effect sizes. These findings underscore the importance
effect size for gender was small). This finding is consistent of not only examining the association between students
with our hypothesis and supports the research that suggests perceptions of their schools and their adjustment, but also of
that peer relations during middle school are particularly examining the ways in which change in perceptions of school
difficult for girls (Crick and Rose 2000; Sadker and Sadker is associated with change in adjustment over time. Our data
1994). Our findings did not, however, consistently support clearly indicate that neither of these types of perceptions or
our hypotheses that girls would experience their school experiences are static or unchanging over time. Thus, any
more negatively than boys. In the sixth grade, girls reported analysis assessing their mutual influence should account for
better teacher and peer support and better clarity and the dynamic nature of these processes.
consistency in school rules and student autonomy than Strikingly, our test for direction of effect indicated that the
boys. Previous research with secondary school teachers influence of teacher support, peer support, or clarity and
suggest that teachers often believe that girls work harder in consistency in school rules on depressive symptoms was
school than boys, and report having better relationships unidirectional; the influence of peer support and student
with their female students than with their male students autonomy on self-esteem was unidirectional; and the influ-
(Suarez-Orozco and Qin-Hilliard 2004). These biases and ence of peer support on problem behavior was unidirectional.
beliefs favoring girls may lead girls in the sixth grade to These findings indicate that it is often the students’ percep-
perceive their school environment as more supportive than tions of school climate, particularly their perceptions of peer
boys. Yet these biases do not seem to protect the girls from support that predicts adjustment and not the students’
perceiving their school to be decreasingly supportive, adjustment predicting perceptions of school climate.
consistent, and fair over time. Taken together, these analyses underscore the role of
Similar to our findings regarding gender differences, our teacher and peer support, opportunities for student auton-
findings with respect to SES differences in perceptions of omy, and clarity and consistency in school rules and
school climate both supported and failed to support our regulations in the psychological and behavioral health of
hypotheses. In support of our hypotheses, we found that students during the middle school years. We must take
lower SES students compared to higher SES students students’ perceptions of school seriously as they consis-
reported lower peer support and poorer perceptions of tently appear to be significantly associated with students’
clarity and consistency in school rules at the sixth grade health and wellbeing. Students’ perceptions of the school
assessment. However contrary to our hypothesis, lower climate appear to be important not just during the transition
SES students reported higher levels of teacher support at from elementary to middle school (see Eccles et al. 1993),
the sixth grade assessment and both the lower and higher but also during middle school itself. Drops in self-esteem
SES students reported similar rates of decline in all four and increases in depressive symptoms have become almost
dimensions of school climate. Furthermore, the effect sizes expected markers of adolescence, presumed to be as de-
for SES on perceptions of school climate were typically contextualized and inevitable as the loss of a child’s first
small. These findings suggest that while SES appears to set of teeth. Yet our results suggest that students’ emotional
differentiate, at least to some degree, students’ perceptions struggles, especially with depressive symptoms, are closely
of school climate in the sixth grade (but not necessarily in a linked with their perceptions of the school culture. Indeed,

123
210 Am J Community Psychol (2007) 40:194–213

the covariation of students’ adjustment with their percep- resulting standard errors. We have thus been conservative
tions of different dimensions of school climate over in interpreting the significance of our results relying on a
3 years, as well as the unidirectional influence of percep- more stringent alpha level.
tions of school climate on adjustment, particularly
depressive symptoms, provides us with unusually strong
evidence for the influence of perceptions of school on Conclusion
student wellbeing.
Our findings suggest that students’ experiences of the
school environment during the middle school years may be
Limitations
as important in shaping psychological and behavioral
adjustment as the transition from elementary school to
Although the current study provides important data
middle school. Indeed, students’ perceptions appear
regarding the experience of middle school for students as
mutable during middle school and, in fact, initially positive
well as the correlates of such experience, there were sev-
perceptions can grow increasingly less positive and these
eral limitations to the study. A key limitation was the small
perceptions appear to affect adaptation. These patterns of
number of ethnic minority students in the sample. Although
decline in student perceptions of the school climate may
we included ethnic minority adolescents in the analysis, our
even be more dramatic if one were to examine student
small numbers did not allow us to compare across ethnic
perceptions in the last 7 years as schools have become
groups. Our assessment of SES was also limited. Free or
more focused on testing in response to the ‘‘no child left
reduced lunch can be an imprecise measure of SES given
behind’’ policies and harsher and more regimented due to
that not all students who qualify sign up for this program
catastrophes like Columbine and Jonesboro. Our findings
and standards for qualifying vary across school districts.
further suggest that there is considerable variation among
However, it is a commonly used technique of assessing
students in perceptions of the teacher and peer support,
SES and is more reliable as a marker of SES than student
clarity and consistency in school rules, and student
self-report of family income (Conchas and Noguera 2004;
autonomy and that the influence of such perceptions extend
Way 1998). The study sample, which had relatively high
beyond academic outcomes to include students’ emotional
rates of both parental college education and free lunch
and behavioral wellbeing. Taken together, these findings
eligibility, was not necessarily representative of the broader
highlight the need for additional research on this critical
US population. These particular demographics may have
period of development. Why do most students perceive
been a function of the broad array of communities—urban,
their school environment more negatively over time? What
suburban, small town, and rural—that were sampled as
accounts for the variation in these perceptions? What types
well as an oversampling of schools with high levels of
of students maintain positive perceptions over time? What
economically and socially disadvantaged students.
can be done to improve students’ perceptions of middle
Future research should also examine additional predic-
schools? These questions seem particularly critical when
tors of perceived school climate aside from gender or SES.
the overwhelming focus in many schools in the United
For example, the size of the student body, as well as the
States at this time is on academic standards and testing
location (urban versus rural) may have an impact on per-
rather than on creating an environment that is perceived by
ceptions of school climate. While a preliminary data
the students to be positive, supportive, and safe.
analysis indicated that inclusion of grade size (e.g., sixth
grade) did not change the pattern of results observed in the Acknowledgment We gratefully acknowledge the support of Peter
current study,7 future research should investigate school- Mulhall at the Illinois Center for Prevention Research. In addition, we
level factors that may shape adolescents’ perceptions of 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
their school environment. this paper.
Finally, although the current data structure is very well
suited for hierarchical linear modeling, cross-domain
growth modeling was selected as the aim was to study the Appendix A
nature of the relationship of change across different
domains. We acknowledge that the nested structure of the Instructions: How often are the following things true about
current data is not incorporated in the planned analyses and THIS SCHOOL? Mark the best answer for each statement.
this may compromise, to some unknown degree, the
7
Preliminary models that included school size were examined. Never Hardly ever Sometimes Most of the time Always
Inclusion of grade size did not change the pattern of results observed
(1) (5)
in the current study.

123
Am J Community Psychol (2007) 40:194–213 211

Teacher Support 2. When teachers make a rule, they mean it.

3. Students are given clear instructions about how to do
1. Teachers take a personal interest in students. their work in classes.
2. Teachers go out of their way to help students. 4. Students understand what will happen to them if they
3. If students want to talk about something, teachers will break a rule.
find time to do it. 5. If some students are acting up in class, the teachers
4. Students really enjoy their classes. will do something about it.
5. Teachers help students to organize their work.
6. Teachers help students catch up when they return from
an absence.
School Harshness

1. Teachers are very strict here.

Peer Support 2. Students get in trouble for breaking small rules.
3. Students get in trouble for talking.
Negative peer interactions 4. It is easy for a student to get kicked out of class in this
school.
1. Students in this school are mean to each other. 5. The rules in this school are too strict.
2. There are kids in this school who pick on other kids.
3. Students in this school have trouble getting along with
each other.
References
4. In classes, students find it hard to get along with each
other. Achenbach, T. M., & Edelbrock, C. (1987). Manual for the youth and
5. Students in this school fell students are mean to them. 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
Affiliation
psychology, 91, 731–746.
American Association of University Women (1992). How schools
1. Students in this school get to know each other really shortchange girls. New York: Marlowe & Company.
well. Anyon, J., & Wilson, W. J. (1997). Ghetto schooling: A political
economy of urban reform. New York: Teachers College Press.
2. Students in this school are very interested in getting to
Bachman, J. G., & O’Malley, P. M. (1986). Self-concepts, self-
know other students. esteem, and educational experiences: The frog pond revisited
3. Students enjoy working together on projects. (again). Journal of Personality and Social Psychology, 50,
4. Students get to know each other well in classes. 35–46.
5. Students enjoy doing things with each other in school Block, J., & Robins, R. W. (1993). A longitudinal study of
consistency and change in self-esteem from early adolescence
activities. to early adulthood. Child Development, 64, 909–923.
Blyth, D., Simmons, R. G., & Carlton-Ford, S. (1983). The
Student Autonomy Climate adjustment of early adolescents to school transitions. Journal
of Early Adolescence, 3, 105–120.
1. Students in this school have a say in how things work. Brand, S. Felner, R. D., DuBois, D. L., Mulhall, P. M., & Flowers, N.
(1994). School climate in middle schools; Development and
2. Students help decide how class time is spent. prelimiary validation of the PSCS. Unpublished Manuscript.
3. In our school, students are given the chance to help Brand, S., Felner, R., Shim, M., Seitsinger, A., & Dumas, T. (2003).
make decisions. Middle school improvement and reform: Development and
4. Students get to help decide some of the rules in this validation of a school-level assessment of climate, cultural
pluralism, and school safety. Journal of Educational Psychology,
school. 95, 570–588.
5. Teachers ask students what they want to learn about. 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
Clarity and Consistency in School Rules Schools, University of Wisconsin.
Cicchetti, D., & Toth, S. L. (1998). The development of depression in
School Structure children and adolescents. American Psychologist, 53, 221–241.
Cohen, J. (1988). Statistical power analysis for the behavioral
sciences (2nd ed.). Hillsdale, NJ: Erlbaum Associates.
1. Teachers make a point to sticking to the rules in Conchas, G., & Noguera, P. (2004). Explaining the academic success
classes. of African American males. In N. Way & J. Chu (Eds.),

123
212 Am J Community Psychol (2007) 40:194–213

Adolescent development in context (pp. 134–176). New York: Kovacs, M. (1980/1981). Affective disorders in children and adoles-
New York University Press. cents. American Psychologist, 44, 209–215.
Connell, J. P., & Ryan, R. M. (1984). A developmental theory of Kozol, J. (1992). Savage inequalities: Children in America’s schools.
motivation in the classroom. Teacher Education Quarterly, 11, New York: Harper Perennial Press.
64–77. Kuperminc, G., Leadbeater, B. J., Emmons, C., & Blatt, S. J. (1997).
Connell, J. P., & Wellborn, J. G. (1991). Competence, autonomy, and Perceived school climate and problem behaviors in middle-
relatedness: A motivational analysis of self-system processes. In school students: The protective function of a positive educational
M. R. Gunnar & L. A. Sroufe (Eds.), Self processes and environment. Journal of Applied Developmental Science, 1,
development. The Minnesota symposium on child psychology. 76–88.
(pp. 43–77). Hillsdale, NJ : Lawrence Erlbaum Associates, Inc. Nolen-Hoeksema, S., & Girgus, J. S. (1994). The emergence of
Crick, N., & Rose, A. (2000). Toward a gender-balanced approach to gender differences in depression during adolescence. Psycho-
the study of social-emotional development: A look at relational logical Bulletin, 115(3), 424–443.
aggression. In P. Miller & E. Kofsky Scholnick (Eds.), Toward a Orenstein, P. (1994). Schoolgirls: young women, self esteem, and the
feminist developmental psychology (pp. 153–168). Florence, KY: confidence gap. Wilmington, NC: Anchor Publishers.
Routledge Press. Reid, K. (1983). Retrospection and persistent school absenteeism.
DuBois, D. L., & Felner, R. D. (1991). Reliability and validity data Educational Research, 25, 110–115.
for the self-esteem questionnaire. Unpublished raw data. Roeser, R. W., & Eccles, J. S. (1998). Adolescents’ perceptions of
DuBois, D. L., Felner, R. D., Brand, S., Phillips, R. S. C., & Lease, A. middle school: Relation to longitudinal changes in academic and
M. (1996). Early adolescent self-esteem: A developmental- psychological adjustment. Journal of Research on Adolescence,
ecological framework and assessment strategy. Journal of 8, 123–158.
Research on Adolescence, 6, 543–579. Roeser, R., Eccles, J., & Sameroff, A. (1998). Academic and
Eccles, J., Barber, B., Jozefowicz, D., Malenchuk, O., Vida, M. emotional functioning in early adolescence: Longitudinal rela-
(1999). Self-evaluations of competence, task values, and self- tions, patterns, and prediction by experience in middle school.
esteem. In N. Johnson & M. Roberts (Eds.), Beyond appearance: Developmental and Psychopathology, 10, 321–352.
A new look at adolescent girl (pp. 53–83). Washington, DC: Roeser, R., Eccles, J., & Sameroff, A. (2000). School as a context of
American Psychological Association. early adolescents’ academic and social-emotional development:
Eccles, J. S., & Midgley, C. (1989). Stage/environment fit: Devel- A summary of research findings. Elementary School Journal,
opmentally appropriate classrooms for early adolescents. In R. E. 100, 443–471.
Amers & C. Amers (Eds.), Research on motivation in education Rosenthal, R., & Rosnow, R. L. (1992). Essentials of behavioral
(pp. 139–186). San Diego, CA: Academic Press. research: Methods and data analysis (2nd ed.). New York:
Eccles, J. S., Midgley, C., Buchanan, C. M., Wigfield, A., Reuman, McGraw Hill.
D., & Mac Iver, D. (1993). Development during adolescence: Rutter, M., Maughan, B., Mortimore, P., Ouston, J., & Smith, A.
The impact of stage/environment fit. American Psychologist, 48, (1979). Fifteen thousand hours: Second schools and their effects
90–101. on children. Cambridge, MA: Harvard University Press.
Eccles, J., & Roeser, R. (1999). School and community influences on Rys G., & Bear, G. (1997). Relational aggression and peer relations:
human development. In M. Bornstein & M. Lamb (Eds.), Gender and developmental issues. Merrill-Palmer Quarterly, 43,
Developmental psychology: An advanced textbook (4th ed., pp. 87–106.
503–554). Mahwah, NJ: Erlbaum Press. Sadker, M., & Sadker, D. (1994). Failing at fairness: How America’s
Epstein J., & Karweit N. (Eds.). (1983). Friends in school. New York: schools cheat girls. New York: Charles Scribner’s Sons.
Academic Press. Sayer, A. G., & Willett, J. B. (1998). A cross-domain model for
Felner, R., Jackson, A., Kasak, D., Mulhall, P, Brand, S., & Flowers, growth in adolescent alcohol expectancies. Multivariate Behav-
N. (1997). The impact of school reform for the middle grades: A ioral Research, 33, 509–543.
longitudinal study of a network engaged in Turning Points-based Schafer, J. L. (2000). NORM: Multiple imputation of incomplete
comprehensive school transformation. In R. Takanishi & D. multivariate data under a normal model, version 2.03 [Computer
Hamburg (Eds.), Preparing adolescents for the twenty-first Software]. Available: [Link]
century: Challenges facing Europe and the United States (pp. html.
38–69). New York: Sage Press. Seidman, E., Allen, L., Aber, L., & Mitchell, C. (1994). The impact of
Fine, M. (1991). Framing dropouts: Notes on the politics of an urban school transitions in early adolescence on the self-esteem and
public high school. New York: SUNY Press. perceived social context of poor urban youth. Child Develop-
Griffith, J. (1995). An empirical examination of a model of social ment, 65, 507–522.
climate in elementary schools. Basic and Applied Social Skinner, E. A., Wellborn., J. G., & Connell, J. P. (1990). What it takes
Psychology, 17, 97–117. to do well in school and whether I’ve got it: A process model of
Haynes, N., & Emmons, C. (1994). The role of school climate in perceived control and children’s engagement and achievement in
students’ psycho-social development. New Haven, Ct: Lecture school. Journal of Educational Psychology, 82, 22–32.
given at invited talk at Yale University. Skinner, E. A., & Wellborn, J. G. (1997). Children’s coping in the
Hoge, D. R., Smit, E. K., & Hanson, S. L. (1990). School experiences academic domain. In S. A. Wolchik & I. N. Sandler (Eds.),
predicting changes in self-esteem of sixth- and seventh-grade Handbook of children’s coping: Linking theory and intervention.
students. Journal of Educational Psychology, 82, 117–127. Issues in clinical child psychology (pp. 387–422). New York:
Jöreskog, K., & Sörbom, D. (2001). LISREL 8.30 [Computer Plenum Press.
Software]. Chicago, IL: Scientific Software International, Inc. Sommer, B. (1985). What’s different about truants? A comparision
Kasen, S., Johnson, J., & Cohen, P. (1990). The impact of school study of eight graders. Journal of Youth and Adolescence, 14,
emotional climate on student psychopathology. Journal of 411–422.
Abnormal Child Psychology, 18, 165–177. Suarez-Orozco, C., & Qin-Hilliard, D. (2004). Immigrant Boys’
Kline, R. (1998). Principles and practice of structural equation Experiences in U.S. Schools. In N. Way & J. Chu (Eds.),
modeling: Methodology in the social sciences. New York, NY: Adolescent boys: Exploring diverse cultures of boyhood (pp.
Guilford Press. 295–316). New York: New York University Press.

123
 Am J Community Psychol (2007) 40:194–213 213

Trickett, E., & Moos, R. H. (1973). The social environment of junior Brown (Eds.), Handbook of applied multivariate statistics and
high and high school classrooms. Journal of Educational mathematical modeling: A comprehensive guide for applied
Psychology, 65, 93–102. researchers in the biological sciences, social sciences, and
Way, N. (1998). Everyday courage: The lives and stories of urban humanities. San Diego, CA: Academic Press.
teenagers. New York: New York University Press. Willet, J. B., & Sayer, A. G. (1996). Cross-domain analyses of change
Way, N. & Robinson, M. G. (2003). Effects of perceived family, over time: Combining growth modeling, covariance structure
friends, and school experiences on change in self-esteem among analysis. In G. A. Marcoulides, & R. E. Schumacker (Eds.),
urban, low-SES adolescents. Journal of Adolescent Research, Advanced structural equation modeling (pp. 125–157). Hillsdale,
18, 324–346. NJ: Erlbaum.
Willett, J. B., & Keiley, M. K. (2000). Using covariance structure
analysis to model change over time. In H. E. A. Tinsley & S.

123

View publication stats