Journal of School Psychology 51 (2013) 599–610

Contents lists available at ScienceDirect

Journal of School Psychology

journal homepage: www.elsevier.com/locate/jschpsyc

Examining classroom influences on student perceptions

of school climate: The role of classroom management and
exclusionary discipline strategies☆
Mary M. Mitchell, Catherine P. Bradshaw ⁎
Johns Hopkins Center for the Prevention of Youth Violence, Johns Hopkins Bloomberg School of Public Health, USA

a r t i c l e i n f o a b s t r a c t

Article history: There is growing emphasis on the use of positive behavior supports rather than exclusionary
Received 7 June 2012 discipline strategies to promote a positive classroom environment. Yet, there has been limited
Received in revised form 9 May 2013 research examining the association between these two different approaches to classroom
Accepted 13 May 2013
management and students' perceptions of school climate. Data from 1902 students within 93
classrooms that were nested within 37 elementary schools were examined using multilevel
Keywords: structural equation modeling procedures to investigate the association between two different
School climate classroom management strategies (i.e., exclusionary discipline strategies and the use of positive
Classroom management
behavior supports) and student ratings of school climate (i.e., fairness, order and discipline,
Office discipline referrals
student–teacher relationship, and academic motivation). The analyses indicated that greater use
Positive behavior support
Multilevel modeling of exclusionary discipline strategies was associated with lower order and discipline scores,
whereas greater use of classroom-based positive behavior supports was associated with higher
scores on order and discipline, fairness, and student–teacher relationship. These findings suggest
that pre-service training and professional development activities should promote teachers' use of
positive behavior support strategies and encourage reduced reliance on exclusionary discipline
strategies in order to enhance the school climate and conditions for learning.
© 2013 Society for the Study of School Psychology. Published by Elsevier Ltd. All rights reserved.

1. Introduction

Schools are charged with providing a safe and supportive environment for learning (Epstein et al., 2008). Consequently, there is
increasing interest in factors associated with positive school climate and conditions for learning. One factor that has been shown to
influence students' perceptions of school climate is the classroom environment (Koth, Bradshaw, & Leaf, 2008). While there has been
considerable attention paid to structural aspects of the classroom environment, such as student–teacher ratio (Finn, Pannozzo, &
Achilles, 2003), other aspects of the classroom context may also influence the students' perception of the environment. For example,
the way in which the teacher manages the classroom and uses different types of disciplinary strategies, such as office discipline
referrals, likely influences students' school climate perceptions. Having an enhanced understanding of classroom factors that
influence students' perceptions of climate may inform school improvement initiatives and programs aiming to promote positive
outcomes for students and staff.

1.1. Variation in perceptions of school climate

School climate is recognized as a critical element of a successful and effective educational environment (Brand, Felner, Shim,
Seitsinger, & Dumas, 2003). School climate has been defined as shared beliefs, values, and attitudes that shape interactions

☆ This research was supported by the Institute of Education Sciences (R324A07118, R305A090307, R324A110107) and the Centers for Disease Control and Prevention
(1U49CE 000728-011 and K01CE001333-01).
⁎ Corresponding author at: Department of Mental Health, Bloomberg School of Public Health, Johns Hopkins University, 624 N. Broadway, Baltimore, MD 21205, USA.
E-mail address: [email protected] (C.P. Bradshaw).
ACTION EDITORS: Robin Codding and Kent McIntosh.

0022-4405/$ – see front matter © 2013 Society for the Study of School Psychology. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jsp.2013.05.005
600 M.M. Mitchell, C.P. Bradshaw / Journal of School Psychology 51 (2013) 599–610

between students, teachers, and administrators and set the parameters of acceptable behavior and norms for the school
(Emmons, Comer, & Haynes, 1996; Kuperminc, Leadbeater, Emmons, & Blatt, 1997). It is a product of social interactions among
students and teachers, is influenced by educational and social values, and has been shown to relate to social situations within
classrooms and to the school as a whole. School climate has been linked with academic achievement and performance (Battistich,
Solomon, Kim, Watson, & Schaps, 1995; Griffith, 1999), student misconduct, aggression, and behavioral problems (Battistich et al.,
1995; Kuperminc et al., 1997; Shochet, Dadds, Ham, & Montague, 2006; Wilson, 2004), adjustment problems (Kuperminc et al.,
1997), and social and personal attitudes (Battistich et al., 1995). Poor perceptions of school climate also have been linked with
low student engagement, truancy, school dropout, delinquency, and bullying behaviors (see Bradshaw, O'Brennan, & McNeely,
2008; Bradshaw, Sawyer, & O'Brennan, 2009; Wilson, 2004).
There is a growing body of research documenting variation in perceptions of school climate by individual-level factors, such as
gender, ethnicity, and age. Among students, girls tend to perceive greater consistency and fairness in school rules, whereas
minority students tend to report less favorable attitudes toward academics (Battistich et al., 1995; Way, Reddy, & Rhodes, 2007).
Also, it is likely that teachers' attitudes and perceptions regarding the school environment would influence their students'
attitudes. Extant research indicates that among elementary school teachers, being male or an ethnic minority has been linked
with less favorable perceptions of the school environment (Bevans, Bradshaw, Miech, & Leaf, 2007). Similarly, younger teachers,
who tend to have less teaching experience, may feel less supported or effective at their job, and therefore may perceive the
environment less favorably than their more experienced colleagues (Hoy & Woolfolk, 1993; Koth et al., 2008). Thus far, it is
unknown exactly how teachers' perceptions of the environment shape students' attitudes about school climate, but it is important
to consider both student and teacher characteristics when assessing school climate.

1.2. Classroom contextual influences

Bronfenbrenner's (1979) social ecological theory suggests that factors at multiple levels may influence student perceptions of
the school environment. For example, indicators of social disorganization at the school and classroom levels have been inversely
related to perceptions of school climate (Bevans et al., 2007; Birnbaum et al., 2003; Bradshaw, Sawyer, et al., 2009; Braham, 2004;
Mijanovich & Weitzman, 2003; Mitchell, Bradshaw, & Leaf, 2010; Plank, Bradshaw, & Young, 2009). In contrast, several studies
have shown that effectively managed classrooms are associated with positive academic performance and fewer behavioral
problems (Arbuckle & Little, 2004). Classroom management includes both maintaining control over students through the use of
discipline and promoting positive environments that foster academic learning and appropriate behavior (Little & Akin-Little,
2008). Classroom management is a common target for classroom-based programs aiming to increase positive student behavior
and academic performance (Epstein et al., 2008), but it may also improve students' perceptions of the climate.
Even teachers in well-managed classrooms occasionally have to address student discipline problems. A common disciplinary
strategy employed by teachers is the use of exclusionary discipline strategies, such as office discipline referrals (ODRs) or suspensions.
ODRs are events in which a staff member observes a student violating a school rule and submits documentation to the administrative
leadership, who then delivers a consequence (Irvin, Tobin, Sprague, Sugai, & Vincent, 2004; Pas, Bradshaw, & Mitchell, 2011).
Exclusionary discipline strategies typically include the immediate removal of the student from the classroom, as the student is
typically sent to the principal's office in an effort to try to stem the problem behavior. As such, the use of this exclusionary discipline
strategy is likely perceived as punitive by students; however, some students may be rewarded by such practices if they are motivated
to avoid or escape from the classroom (Sugai, Sprague, Horner, & Walker, 2000). In fact, some research suggests that students
attribute their peers' misbehavior to their teachers' usage of punitive strategies (Miller, Ferguson, & Byrne, 2000). In addition,
students in classrooms with more misbehavior reported more punitive and aggressive discipline strategies (Lewis, 2001), which
suggests an escalating feedback loop between student misbehavior and their teachers' punitive or exclusionary discipline strategies.
The use of exclusionary disciplinary strategies is often associated with confrontational student–teacher interactions, which
could promote more negative views of school climate for all students. For example, Nelson and Roberts (2000) found that an
average of four negative confrontations occurred between the student and teacher prior to the student receiving an ODR and that
these confrontations could continue as a consequence of making an ODR. Similarly, a cross-cultural study by Lewis, Romi, Qui, and
Katz (2005) documented a significant association between the teachers' use of “coercive” discipline strategies and students'
misbehavior. Students who engaged in higher levels of disruption and problem behavior perceived their teachers' disciplinary
behavior to be more aggressive. It appears that students' provocative behavior evoked higher levels of anger in their teachers,
which resulted in teachers' use of reactive and aggressive disciplinary strategies. This finding suggests that teachers' use of
reactive and exclusionary discipline strategies could negatively impact students' perceptions of their relationship with the teacher
and the classroom context. Research also suggests that exclusionary discipline strategies may only temporarily reduce problem
behaviors, but they do not fully alleviate the problematic behavior or prevent the onset of other behavior problems (Sugai &
Horner, 2006; Sugai, Todd, & Horner, 2000).
Consequently, there has been a shift from the use of exclusionary discipline strategies to the use of positive, proactive
classroom management strategies, such as establishing consistent behavioral expectations and reinforcing those expectations
(Epstein et al., 2008). Classroom-based strategies that actively teach and reward positive behavioral expectations have been
shown to be effective at reducing behavior problems, and in turn, may improve the classroom climate (Sugai et al., 1999).
However, additional research is needed to examine how positive classroom management strategies and reduced reliance on
exclusionary discipline strategies relate to students' perception of the school climate. Such research may identify a potential
mechanism by which positive behavior support and the use of exclusionary discipline strategies influence school climate.
 M.M. Mitchell, C.P. Bradshaw / Journal of School Psychology 51 (2013) 599–610 601

1.3. Overview of the current study

The current study aimed to examine the association between classroom management strategies and student perceptions of school
climate, while adjusting for student and teacher demographic characteristics. Despite the importance of school climate and its
relationship with social, behavioral, and academic outcomes (Wilson, 2004), relatively few studies have investigated classroom
contextual predictors of student-perceived school climate. For example, research by Mitchell et al. (2010) reported that there are often
discrepant perceptions of school climate between teachers and students, whereas Koth et al. (2008) identified student and teacher
demographic characteristics that were associated with school climate (e.g., class size and level of disruptive behavior). The current
study extends the previous research on classroom contextual factors by examining more process-oriented classroom-level factors, such
as teachers' use of exclusionary disciplinary strategies and positive behavior supports on students' perceptions of school climate. We
also employed a novel analytic approach; we used a multilevel latent variable framework to simultaneously examine a broad set of
inter-related school climate variables. This innovative analytic approach enabled us to both account for measurement error and jointly
consider several aspects of school climate, including student perceptions of fairness, order and discipline, student–teacher relationship,
and academic motivation (Wilson, 2004). This approach in turn provides the opportunity for a more comprehensive exploration of
macro-level classroom influences on students' perception of school climate.
The data for this study came from a randomized controlled trial of school improvement and reform initiatives in 37 elementary
schools, with a particular focus on School-wide Positive Behavioral Interventions and Supports (SWPBIS; Sugai & Horner, 2006).
Although intervention status was included as a covariate in the analyses, it was not a central focus of the current study; furthermore,
our preliminary analyses indicated that SWPBIS intervention status did not have a significant impact on the outcomes of interest in
this study, although significant effects have been reported on other outcomes (e.g., Bradshaw, Koth, Thornton, & Leaf, 2009;
Bradshaw, Mitchell, & Leaf, 2010; Bradshaw, Waasdorp, & Leaf, 2012; Waasdorp, Bradshaw, & Leaf, 2012).
We used a multilevel latent variable modeling framework to examine the association between teacher-level factors and student
perceptions of climate, while adjusting for student demographic characteristics. First, we hypothesized that the percentage of students
within a classroom who received exclusionary discipline practices would be inversely associated with aspects of student-rated school
climate such as fairness, order and discipline, student–teacher relationship, and achievement motivation. Second, we hypothesized that
the use of effective classroom management practices (i.e., positive behavior support strategies) would be associated with higher scores
on students' ratings of school climate. Additional child and teacher demographic characteristics were included in the model as control
variables.

2. Method

2.1. Participants

Data for the current study were collected from 1902 fifth-grade students and 93 fifth-grade homeroom teachers at 37
Maryland public elementary schools participating in a large-scale study of school improvement and reform initiatives, with a
particular focus on SWPBIS (Sugai & Horner, 2006); 21 of the schools had received training in SWPBIS, whereas the remaining 16

Table 1
Student-level and teacher/classroom-level demographic characteristics.

Demographic characteristics Mean (SD) N (%)

Student-level (N = 1902)
Student's gender
Boys 973 (51.2)
Girls 929 (48.8)
Student's ethnicity
White 765 (40.2)
Black 637 (33.5)
Other 500 (26.3)
Teacher/classroom-level (N = 93)
Teacher's age
30 or younger 42 (46.7)
Older than 30 51 (53.3)
Teacher's gender
Women 82 (88.2)
Men 11 (11.8)
Teacher's ethnicity
White 81 (87.1)
Non-White 12 (12.9)
Class size 21.88 (3.72)
Classroom management (% in-place)a 76.29 (21.26)
% of students receiving exclusionary disciplinarily practices 21.48 (15.32)

Note.
a
Assessed via the Effective Behavior Survey.
602 M.M. Mitchell, C.P. Bradshaw / Journal of School Psychology 51 (2013) 599–610

schools had not received training in the school-wide model. All general education fifth-grade classrooms were eligible for inclusion
and participated in the data collection; however, special education classrooms were excluded. The schools were scheduled such that
when the students changed classrooms (e.g., for science and math), they traveled to other classrooms as an intact cohort. In addition,
homeroom teachers spent at least half of the school day with the students whom they reported on the receipt of exclusionary
discipline practices. The school climate data were collected at a single time point from an anonymous sample of fifth-grade students
whose data were linked by their homeroom teacher and school. Most students were White (40.2%) or Black (33.5%); 26.3% were other
ethnicities, and just 2.7% were English Language Learners. Nearly half of the teachers were 30 years of age or younger (46.7%), and the
majority were White (87.1%; see Table 1 for additional sample demographic characteristics). The schools ranged in size from 223 to
791 students (M = 472, SD = 135). Forty-eight percent of the schools were located in suburban communities, 41% were located in
urban fringe communities, and 11% were located in rural communities.

2.2. Measures

2.2.1. School climate

Students completed the elementary school version of School Climate Survey (SCS; Haynes, Emmons, & Ben-Avie, 1997), which
measured the following different facets of climate: fairness, order and discipline, student–teacher relations, and achievement
motivation. Students responded to each item on the scale by indicating “agree” or “disagree;” some items were reverse scored so
that a higher score indicated a more positive school environment (see Table 2 for item-level latent variable loadings and
variances). The 4-item Fairness subscale (Cronbach's alpha [α] = .80 in this study) included items such as, “At my school, boys
and girls are treated equally well” and “At my school, children of all races are treated the same.” The Order and Discipline subscale
(α = .74) consisted of 11 items, such as “My school is a safe place” and “It is easy for teachers at my school to control the
children.” The Student–Teacher Relationship subscale (α = .85) was composed of 10 items including “My teachers care about
me” and “Teachers at my school help us children with our problems.” The Achievement Motivation subscale (α = .65, which is
minimally acceptable, DeVellis, 1991), was composed of 6 items such as, “I feel I can do well in this school” and “My teachers
believe I can do well in my school work.” All of the alpha values for the SCS subscales found in this study are comparable to those
reported in a previous study using similar samples (Koth et al., 2008). These alphas are also comparable to previous psychometric
properties reported by Haynes et al. (1997), which published reliabilities of .83 for the Fairness subscale, .75 for the Order and

Table 2
Factor loadings and error variance estimates for the indicator variables and error variance estimates for the latent variables.

Latent variable Loading Error variance

Fairness .98
Everyone is treated equally well at my school. .92 .15
At my school, boys and girls are treated equally well. .83 .32
At my school, children of all races are treated the same. .79 .38
At my school, all children are treated the same, even if their parents are rich or poor. .68 .54
Order and Discipline .98
My school is a safe place .67 .55
Children at my school listen to the teachers. .67 .55
At my school, the children disobey the rules. .67 .55
Children at my school fight a lot. .65 .48
Some children carry guns or knives in my school. .59 .65
Children at my school often get hurt in school. .59 .65
My school is usually very noisy. .53 .71
This school is quiet enough. .52 .73
Many children at my school are put on suspension. .50 .75
Children at my school are neat and tidy. .47 .77
It is easy for teachers at my school to control the children. .42 .84
Student–Teacher Relationship .97
At my school, teachers are fair to everyone. .85 .28
My teachers make me feel good about myself. .84 .30
Teachers at my school help us children with our school problem. .79 .38
My teachers care about me. .78 .39
Teachers at my school help us children with our problems. .77 .41
My teachers work hard to get me to do well on tests. .73 .47
I can talk to my teachers about my problems. .73 .46
The principal at my school cares about us children. .67 .56
At my school, the teachers do not respect the children. .66 .56
Students at my school trust the teachers. .57 .68
Academic Motivation .98
I enjoy learning at this school. .86 .27
My teachers believe I can do well in my school work. .79 .47
I like coming to school. .72 .48
I feel that I can do well in this school. .71 .49
I do all my school work. .34 .88
I do extra school work at my school. .17 .97
 M.M. Mitchell, C.P. Bradshaw / Journal of School Psychology 51 (2013) 599–610 603

Discipline subscale, and .87 for the Student–Teacher Relationship subscale; however, the alpha value for the Achievement
Motivation subscale was not previously published.

2.2.2. Student and teacher demographic information

Demographic data on the students (i.e., gender and race/ethnicity) were obtained from the students, whereas demographic
data on the teachers (i.e., age, gender, race/ethnicity) were obtained through a brief self-report measure completed by all
participating staff. Students were coded 1 for girls, 0 for boys, and 1 for White and 0 for non-White. Teachers were coded 1 for
women, 0 for men, 1 for White, 0 for non-White, and 1 for 30 years or younger and 0 for older than 30 years.

2.2.3. Classroom-level variables

Three classroom-level variables were analyzed in the current study: teachers' use of positive behavior supports, teachers' use of
exclusionary disciplinary strategies, and class size. Although the Effective Behavior Support Survey (EBS Version 1; Sugai, Todd, et al.,
2000), which is a teacher-reported measure of the quality of classroom management, has four subscales including School-Wide,
Non-Classroom, Classroom, and Individual Student systems (Safran, 2006), we only used the Classroom subscale because we were
specifically interested in the classroom context as a predictor of student perceptions of school climate. Participants reported the
extent to which each item was “in place” within their classroom. Sample items include “Expected student behavior & routines in
classrooms are stated positively & defined clearly” (94.6% reported this feature as being in place), “Problem behaviors are defined
clearly” (89.2%), “Classroom-based options exist to allow classroom instruction to continue when problem behavior occurs” (71.0%),
“Transitions between instructional & noninstructional activities are efficient & orderly” (74.2%), “Procedures for expected & problem
behaviors are consistent with school-wide procedures” (79.6%), “Expected student behaviors are acknowledged regularly (positively
reinforced)” (>4 positives to 1 negative) (71.0%), and “Problem behaviors receive consistent consequences” (71.0%). The reliability
and validity evidence supporting the EBS and its subscales has been generated through previous studies (see Safran, 2006; Walker,
Cheney, & Stage, 2009). The EBS measure was scored by indicating the proportion of items marked as in place at the classroom level,
such that a higher score indicated more positive behavior supports. The 12-item Cronbach's alpha was .83 for the Classroom subscale
in this sample. A previous study of the EBS documented the alpha for the Classroom subscale to be .74 (Safran, 2006).
Teachers' use of exclusionary disciplinary strategies was operationalized by the percent of students that the teachers reported
that they had referred to the principal's office for a disciplinary infraction. Specifically, during the last month of the school year,
the teachers completed a brief checklist in reference to each student in their classroom, on which they indicated whether the
child had had been “sent to the principal's office” for disciplinary reasons during the current school year. Previous research from
this project on this teacher-report variable in conjunction with administrative data from the School-wide Information System
(SWIS; Irvin et al., 2004) regarding student-specific receipt of ODRs indicated that teacher reports are a valid and reliable source
of information on ODRs (Pas et al., 2011). Specifically, Pas et al. (2011) used a variety of methods, including ROC curves, to
demonstrate an adequate level of sensitivity and specificity of the teacher-reported exclusionary discipline strategy; they also
provided evidence of a moderate level of convergence between teacher-report and administrative (i.e., SWIS) data on the use of
these practices. In the current study, the number of students receiving at least one exclusionary discipline practice was summed
and divided by the total number of students in the class to yield a percent score for each teacher or classroom. Given that larger
classrooms are typically more difficult for teachers to manage (Birnbaum et al., 2003; Finn et al., 2003), we also included a class
size variable in the models as a classroom-level covariate. This variable was computed by summing the total number of students
per class for each teacher.

2.3. Procedure

The teachers' data were collected in May of a single school year via an individually-addressed survey packet. The survey
packets were mailed in bulk to the school and distributed to the school staff. Teacher participation was voluntary, and participants
provided written consent. To ensure confidentiality, teachers completed the study materials on their own time and returned the
materials directly to the researchers through the United States postal service. Each teacher questionnaire packet included a small
incentive (e.g., a disposable ballpoint pen or bookmark, with an approximate value less than one dollar). The teacher response
rate was 89.9%. Teachers also completed a brief checklist regarding each student, on which they indicated whether or not the
student had been referred to the principal's office for disciplinary problems during the current school year (see Pas et al., 2011 for
additional information regarding procedures and validity of this approach).
All fifth graders completed the School Climate Surveys, which were administered anonymously but were linked to the
student's homeroom teacher. The surveys were group administered by trained project staff members who provided a brief
overview of the purpose of the survey and read each question aloud as the students completed the survey. The student
response rate was 94.1% based on fall enrollment data. Reasons for the less than 6% missing student school climate data
were absences (because all data were collected on one day), student transfers to other schools or grade levels over the
course of the school year, and parental non-consent for participation (which was less than 1%). Student participation
occurred through a passive parental consent process. The Institutional Review Board at the researchers' institution provided
approval for this study.
604 M.M. Mitchell, C.P. Bradshaw / Journal of School Psychology 51 (2013) 599–610

2.4. Analysis

We used Mplus (Muthén & Muthén, 1998–2010) to fit a series of multilevel structural equation models (ML-SEMs) based on
student- and classroom-level data. We used the individual manifest or measured items from each of the four subscales from the
School Climate Survey to create four latent variables at the student level. Using latent variables rather than manifest variables to
represent the constructs targeted by these subscales enabled us to control measurement error (Marsh et al., 2009). We then
computed intercorrelations among the four latent variables (i.e., Fairness, Order and Discipline, Student–Teacher Relationship,
and Achievement Motivation) and the teacher and classroom-level continuous manifest variables (see Table 3).
We then created the ML-SEM model, in which we allowed the latent school climate variables to covary. We also included manifest
variables at the teacher level, which made the model multilevel. Specifically, the latent variables in the model, which included the
student-level school climate latent variables, necessitated the use of a structural equation model, while the nested nature of the data
of students within classrooms required a multilevel approach (Raudenbush & Bryk, 2002). These two methods can be synthesized to
create a multilevel structural equation modeling analysis design (Rabe-Hesketh, Skrondal, & Zheng, 2012). A ML-SEM is considered
doubly latent because it controls for error associated with manifest variables as well as error due to sampling clusters of individuals
(Marsh et al., 2009). This type of model has been used previously to investigate the relation between school-level latent variables and
student-level latent variables (Rabe-Hesketh et al., 2012). Although ML-SEM has been used to investigate multilevel mediation
models (Preacher, Zyphur, & Zhang, 2010) and latent variables at both the within (i.e., student) and between (i.e., classroom) levels
(Mehta & Neale, 2005), our study uses a simpler application of the ML-SEM framework. Moreover, this study combines the strengths
of multilevel modeling, such as using a model intended to handle dependencies inherent in clustered data with the advantages of
structural equation modeling, which accounts for the measurement error with latent constructs (Mehta & Neale, 2005). After
reviewing different software packages that could model ML-SEM, Bovaird (2007) recommended Mplus due to its versatility in
handling all types of latent-variable modeling. Mplus also allowed us to model all four latent school climate outcomes simultaneously,
which is consistent with our overall interest in examining the comprehensive construct of school climate. For further explanation of
ML-SEM, see Muthén and Muthén (1998–2010).
We also adjusted the standard errors of the estimates due to clustering of classrooms within schools. Although we adjusted
estimates at the school-level, we did not have an adequate sample size to test school-level effects. Also, it is necessary to use the
multilevel approach with nested data because correlation between participants is expected (Zucker, 1990) and even low levels of
correlation, as measured by the intraclass correlation coefficient, can increase Type I error (Baldwin et al., 2011).
To increase parsimony, we simultaneously modeled all four within-classroom latent variables that represented the following four
factors of student-rated school climate: Fairness, Order and Discipline, Student–Teacher Relationship, and Achievement Motivation.
Each latent variable was specified in the measurement model as having categorical indicators, because the individual survey items
were binary. In a single multilevel structural equation model, these four latent variables were regressed on six manifest variables at
the classroom level: classroom management (i.e., EBS scores), percent of students receiving an ODR, class size, and teacher
demographic characteristics such as age, ethnicity, and gender. Two student demographic characteristics (i.e., ethnicity and gender)
were entered as covariates at the within-classroom level (Muthén & Muthén, 1998–2010). An alpha-level of .05 was used for testing
the statistical significance of each path in the model.
In the Mplus code, we used the “Type = twolevel complex” command because this command takes clustering into account
and adjusts the parameter and standard error estimates, and chi-square fit statistic (Muthén, 2002a, 2002b). The “cluster =
school teacher” statement tells Mplus which clusters individuals belong to so that multilevel modeling can be done (Muthén,
2002a, 2002b). We also used the “Between =” statement to specify which variables were entered at the teacher/classroom level
(see Fig. 1). Although SWPBIS intervention effects on student reports of school climate were not the focus of the current analyses,

Table 3
Correlations among latent and manifest study variables.

Variables 1 2 3 4 5 6 7 8 9

1. Fairness a –
2. Order and Disciplinea .72⁎⁎⁎ –
3. Student–Teacher Relationshipa .79⁎⁎⁎ .67⁎⁎⁎ –
4. Achievement Motivationa .66⁎⁎⁎ .61⁎⁎⁎ .89⁎⁎⁎ –
5. Percent of students receiving exclusionary discipline practices −.21⁎ −.57⁎⁎⁎ −.22 −.22 –
6. Classroom managementb .20 .27⁎ .22 .29⁎⁎ −.31⁎⁎ –
7. Class size −.05 .11 −.05 −.01 −.17 −.01 –
8. Teacher ethnicity .06 .13 .26⁎⁎⁎ .20 .10 .00 .02 –
9. Teacher age −.05 −.15 −.08 −.07 .23⁎ −.10 −.08 −.09 –
10. Teacher gender .11 .16 .13 .07 −.15 .00 .04 −.04 −.12

Note. Teacher Ethnicity coded 1 = White, Minority = 0; Teacher Age: 30 or younger = 1, Over 30 = 0; Teacher Gender: Woman = 1, Man = 0.
a
Indicates a latent variable.
b
Assessed via the Effective Behavior Support Survey.
⁎ p b .05.
⁎⁎ p b .01.
⁎⁎⁎ p b .001.
 M.M. Mitchell, C.P. Bradshaw / Journal of School Psychology 51 (2013) 599–610 605

we conducted sensitivity analyses to explore for such an effect; therefore, we entered the school-level intervention variable to
test for intervention effects of SWPBIS on students' ratings of school climate.
The measurement model fit indices that were used included the comparative fit index (CFI; ≥.95 indicates good model fit),
Tucker–Lewis index (TLI; ≥ .95 indicates good model fit), and root mean square error of approximation (RMSEA; ≤ .05 indicates
good model fit). Model fit indices for the ML-SEM included the Akaike information criterion (AIC) and sample-size adjusted
Bayesian information criterion (aBIC), such that lower scores indicated better fit (Hancock & Mueller, 2001; Luke, 2004). Because
the indicators of the latent variables were binary, we used the default estimator for this type of model, which is maximum
likelihood with robust standard errors using a numerical integration algorithm (Muthén & Muthén, 1998–2010). Because the
chi-square, AIC, and aBIC are tests of relative fit, there are no established cut-off values for these fit indices. For further explanation
of ML-SEM, estimators, and fit indices, see Muthén and Muthén (1998–2010).

2.4.1. Missing data

Of the 1902 children included in the analysis, none of the children were missing values identifying their gender or ethnicity.
On the School Climate Survey, between 2 (0.1%) and 13 (0.7%) children were missing on each item. By using maximum likelihood
estimation in the structural equation model, the latent variables are estimated using all available data, which means that all
children continue to be included in the analysis; this method is a distinct advantage of using maximum likelihood compared to
other estimators that employ listwise deletion.

3. Results

3.1. Preliminary analysis

We fit a measurement model using confirmatory factor analysis with the four within-group factors that represented student-rated
school climate. These latent variables, which included Fairness, Order and Discipline, Student–Teacher Relationship, and Achievement
Motivation, exhibited adequate fit according to established cutoff values for the comparative fit index and root mean square error of
approximation, χ2 = 490.12, df = 84; CFI = .95; TLI = .98; RMSEA = .04 (Hancock & Mueller, 2001).
We then computed intercorrelations among the four latent variables (i.e., Fairness, Order and Discipline, Student–Teacher
Relationship, and Achievement Motivation) and the teacher and classroom-level continuous manifest variables (see Table 3). We

Student Student
Ethnicity Gender

.12 .04
.10
Within Group
.15 .03 .18
.02
.12
Student-
Order & Achievement
Fairness Teacher
Discipline Motivation
Relationship

Between Group

Exclusionary Poor Teacher Teacher Teacher

Discipline Classroom Class Size
Age Ethnicity Gender
Strategies Management

.48 .16 -.11 .02

-.04
-.01 .06
.17 .08 .27 .28
-.16 -.10 .10
.28 -.24 .27
-.30 -.04
.07 .00
.12 -.08
-.08 Student- Achievement
Fairness Order &
Teacher Motivation
Discipline
Relationship

Fig. 1. Multi-level structural equation model of student-rated climate regressed on teacher and classroom characteristics. Student Ethnicity coded White = 1,
Minority = 0; Student Gender coded girls = 1, boys = 0; Teacher Ethnicity coded 1 = White, Minority = 0; Teacher Age: 30 or younger = 1, Over 30 = 0; Teacher
Gender: Woman = 1, Man = 0; Solid lines indicate statistically significant standardized path coefficient at p b .05 and dotted lines indicate nonsignificant
standardized path coefficients at p ≥ .05. Fit indices: Akaike Information Criterion (AIC) = 53944.69, Bayesian Information Criterion (BIC) = 54705.13, Sample-Size
Adjusted BIC (aBIC) = 54269.88, −2 Log likelihood (−2LL) = −26835.34, Number of free parameters = 137.
606 M.M. Mitchell, C.P. Bradshaw / Journal of School Psychology 51 (2013) 599–610

calculated point biserial correlations to measure associations between dichotomous and continuous variables and phi coefficients
for correlations between two dichotomous variables. The bivariate analyses suggested that the four school climate latent variables
were highly correlated. Also, the teachers' greater use of exclusionary disciplinary strategies (i.e., percentage of students receiving
a referral to the principal's office) was associated with lower ratings of student reports of Fairness (r = − .21, p = .041) and
Order and Discipline (r = − .57, p = .001). In addition, younger teachers assigned more ODRs (r = .23, p = 001). The percent of
students who received a referral to the principal's office by teachers was inversely associated with scores on the EBS (r = − .31,
p = 001). Finally, EBS scores were associated with greater student-reported Order and Discipline (r = .27, p = .030) and
Achievement Motivation (r = .29, p = .003).

3.1.1. ML-SEM depicting the association between classroom context and climate
We fit a multilevel structural equation model to assess the main effects of classroom and teacher-level effects on students'
reports of school climate, while also adjusting for clustering at the school level (see Fig. 1). The model that we fit had no
convergence issues and had a − 2 Log Likelihood of − 26835.34, AIC = 53944.69, BIC = 54705.13, and aBIC = 54269.88. Due to
the estimation method (maximum likelihood robust with Monte Carlo integration), the CFI and RMSEA, which have established
cut-off values, were not computed (Muthén & Muthén, 1998–2010). Therefore, the fit indices reported are used for model
comparisons, rather than individual model fit.
As evident in the top panel of Fig. 1, we found that compared to minority students, White students rated each of the four
school climate latent variables more favorably (standardized path coefficients ranging from .12 to .18, ps b .001). It is important
to note that these models adjusted for covariates, which may have a potential influence affecting school climate. With regard to
gender differences, only achievement motivation varied significantly by gender, such that girls rated this latent variable
significantly higher than did boys. At the classroom or between-level (as evident in the bottom panel of Fig. 1), a statistically
significant negative effect of the teacher's use of exclusionary disciplinary strategies was evident on the students' ratings of order
and discipline; a greater rate of exclusionary disciplinary strategies was associated with lower order and discipline scores.
Classroom management was positively associated with three of the four latent variables of school climate with standardized path
coefficients ranging from .17 for order and discipline to .48 for fairness; these coefficients can be interpreted as effect sizes.
Teacher age demonstrated a statistically significant effect on student–teacher relationships and achievement motivation;
students with younger teachers rated student–teacher relationship and achievement motivation more positively than did
students with teachers over the age of 30.
There were, however, some nonsignificant associations, such as student gender was not associated with Fairness or Order and
Discipline (see Fig. 1). At the classroom-level, teacher gender and class size were unrelated to any of the latent school climate
variables. Similarly, teacher age was not associated with Fairness or Order and Discipline, and teacher ethnicity was only
associated with Order and Discipline. The use of exclusionary discipline strategies was only associated with Order and Discipline,
but not any of the other three latent school climate variables.
Although the SWPBIS intervention effects were not the focus of the current study, we conducted sensitivity analyses to explore
any such effects on student reports of school climate in the final models. Therefore, we entered intervention status (i.e., SWPBIS
vs. Comparison school) as a covariate at the between-classrooms level of the model and found that this variable had no significant
association with any of the four student climate latent variables at the within-classroom level (ps > .05).

4. Discussion

The current paper used ML-SEM to examine the association between classroom variables (i.e., use of exclusionary discipline
practices and use of classroom management strategies) and student perceptions of school climate. We found support for our
hypothesis that the use of exclusionary discipline strategies would be associated with less favorable perceptions of school climate
—specifically the order and discipline latent variable. The association with this particular facet of climate is logical, because this
latent variable relates to the use of behavior management strategies. Although some may have expected the opposite association
because exclusionary discipline practices were intended to maintain order and discipline (Irvin et al., 2004), our results indicated
that they were associated with lower order and discipline. This finding is consistent with previous studies, which found that use
of exclusionary disciplinary strategies was associated with greater student misbehavior (Lewis et al., 2005). These results suggest
that teachers' use of exclusionary discipline practices may signal to the students that the classroom is disruptive, whereas
classrooms with teachers who use fewer exclusionary discipline practices appear to be perceived as more orderly. However, it is
also possible that teachers who have disruptive children in their classrooms are more likely to use exclusionary discipline
practices to manage the behavior. Prior research and logic suggest that using positive behavioral strategies would reduce the level
of classroom disruption and need for exclusionary discipline practices (Sugai, Sprague, et al., 2000; Sugai, Todd, et al., 2000b).
We were, however, surprised by the lack of support for the proposed negative association between exclusionary disciplinary
strategies and the other three school climate latent variables, which showed no significant associations. Although the four climate
latent variables were highly correlated, it appears that students differentiated between perceptions of discipline compared to
perceptions of their personal relationships with their teachers, perceptions of fairness, and beliefs about their academic
motivation. Nevertheless, the findings regarding the association between ODRs and perceptions of order and discipline support
the notion that exclusionary discipline strategies may actually have iatrogenic effects on the students' perceptions of the
classroom context.
 M.M. Mitchell, C.P. Bradshaw / Journal of School Psychology 51 (2013) 599–610 607

With regard to our secondary hypothesis that aspects of classroom management would be positively correlated with the
school climate factors, we found that using positive behavioral strategies was positively associated with three of the four latent
school climate variables (i.e., Fairness, Order & Discipline, and Student–Teacher Relationship). These results suggest that teachers'
use of proactive and positive rather than exclusionary and negative strategies may help foster a more constructive and supportive
learning environment for students. This finding is consistent with the movement toward less punitive and more positive
approaches to school discipline and classroom management (Epstein et al., 2008; Lewis et al., 2005; Sugai & Horner, 2006).
Taken together, the results of the current study suggest that students are sensitive to classroom contextual factors, such as the
use of different types of disciplinary and classroom management strategies. Although the teachers' use of exclusionary and
castigatory discipline strategies may have a significant influence on the way in which students perceive the classroom and the
school more generally, we are cautious in interpreting these findings as causal. These results do suggest that students' perceptions
may be shaped not only by individual-level factors, such as gender and ethnicity, but also by classroom-level factors such as
teacher age and classroom management strategies that include exclusionary discipline and positive behavior supports. The
influence of the use of exclusionary discipline strategies was localized to students' perceptions of order and discipline, whereas
classroom management was associated with three of the four school climate variables. Therefore, future studies should consider
contextual factors at both the intrapersonal and interpersonal levels on students' perceptions of climate and should more closely
consider these contextual factors in relation to different aspects of school climate.

4.1. Limitations

It is important to consider some limitations when interpreting these findings, including those related to the study design,
measurement, and modeling. First, we used cross-sectional data, which limits our ability to make definitive causal statements
regarding the direction of effects between classroom management and student ratings of school climate. Despite this limitation,
we conceptualized the theory of change process such that teachers' use of disciplinary strategies predicted the students'
perception of school climate (Epstein et al., 2008); however, it is possible that the school climate predicts teachers' use of different
discipline strategies.
Due to students' reading and comprehension levels, we only obtained data from fifth-grade students and their teachers. A
different pattern of findings may have emerged if younger elementary school students and teachers from other grade levels were
assessed. Also, our sample of fifth-grade students may not generalize to other fifth-grade classrooms that have varying class sizes
or include students with special needs or for whom English is not their first language.
The school climate response options were dichotomous, which generally produces a weaker measure than using continuous
items, but perhaps easier for younger students to complete than traditional Likert-scaled items. Also, because only the overall use
of exclusionary disciplinary strategies was assessed, rather than the number of students with multiple referrals to the principal's
office, caution should be taken in interpreting these conclusions. Although our student response rate of over 94% was high for a
study of this scale, we were missing data on a subset of students who were absent, changed schools or grade levels, and less than
1% of whom had parents who did not consent to the study.
Because the EBS measure classroom management was self-report, it may be influenced by social desirability. While classroom
management data were not available from multiple sources in the current study, future research should seek to incorporate a wider
array of data sources. As noted above, one item on the EBS scale is “Problem behaviors receive consistent consequences;” therefore,
the EBS does pertain to consequences, which may include ODRs, but the scale also includes an item “Classroom-based options exist to
allow classroom instruction to continue when problem behavior occurs,” which reflects alternatives to exclusionary approaches. We
operationalized the exclusionary discipline variable as the percent of students per classroom who had referred to the principal's office.
We used teacher reports of these referrals because ODRs were not standard data elements collected by all schools in the project. The
potential of inaccurate recall is also a possible limitation. However, a previous study by Pas et al. (2011) validated teacher-reports of
exclusionary discipline practices with administrative records among the schools for whom ODR data were maintained. Yet, there is
some evidence that the validity of exclusionary discipline practices may vary depending on whether there is a standardized
(McIntosh, Campbell, Carter, & Zumbo, 2009) or unstandardized (Nelson, Benner, Reid, Epstein, & Currin, 2002) procedure for
documentation. Perhaps this issue of standardization is more relevant for archival records than for teacher-report, which may be
based largely on recall, or when the inferences are made about specific students, whereas our approach was to aggregate up to the
classroom level. Nevertheless, we acknowledge that assessing exclusionary discipline practices using teacher reports is a major
limitation of the study. It is also important to note that exclusionary discipline practices can reinforce the very behaviors they were
meant to punish by giving students the opportunity to escape schoolwork in a classroom environment or to gain attention from
school administrators, as suggested in the escape/reward behavioral model (Lewis, 2001; Lewis et al., 2005). Due to the anonymity of
the school climate data collection, we could not link specific students' reports of climate with their personal experience of a referral to
the principal's office.
The academic achievement subscale of the Student Climate Survey had a minimally acceptable alpha of .65, which could
attenuate associations between this latent variable and the other covariates. However, this situation was improved by using a
latent variable to represent these six items, as the latent variable by definition accounts for the measurement error (Little,
Lindenberger, & Nesselroade, 1999; Muthén, 2002a, 2002b). In addition, it should be noted that some researchers (e.g., Boyle &
Pickles, 1998) have found that while latent variable methods improve the reliability of measured variables, this technique can also
produce larger standard errors, which could explain the lack of significance in the association between the EBS and achievement
motivation.
608 M.M. Mitchell, C.P. Bradshaw / Journal of School Psychology 51 (2013) 599–610

The social ecological model also suggests that there are likely bi-directional influences, whereby students' behavior influences
the teachers' behavior, which in turn influences the students' perception of the climate. These types of transactional processes are
difficult to assess and are not captured in the current data analysis. For example, it is possible that low order and discipline could
lead to increased use of exclusionary discipline practices. Similarly, children who receive high levels of exclusionary discipline
practices may perceive the school climate differently than their peers (e.g., reduced student–teacher relationship scores);
however, these student-level associations could not be examined because exclusionary discipline practices were measured at the
classroom level. There may also be factors at the school level that influence students' perceptions of climate. Although we did use
the Huber–White correction to adjust for the clustering at the school-level, we are not able to model covariates at the school-level
within the Mplus framework. We conducted additional sensitivity analyses with the intervention status variable (i.e., SWPBIS vs.
Comparison) at the classroom level; however, we found no significant differences in student climate ratings between the two
intervention conditions. We do not have treatment integrity data on SWPBIS implementation at the classroom-level, only at the
school level (see Bradshaw, Reinke, Brown, Bevans, & Leaf, 2008). We also acknowledge that it is quite possible that there are
other interventions in place within the schools beyond SWPBIS that may have influenced the pattern of findings. However, we
believe this to be true of the larger population of schools, as most schools are implementing some type of intervention
(Gottfredson & Gottfredson, 2001). The study was not designed to examine the functional relationship among all possible
school-based interventions and the outcomes.
Although other multilevel modeling software programs (e.g., hierarchal linear modeling [HLM 7.01]; Raudenbush et al., 2011)
would allow us to model a third level, they preclude us from modeling latent variables or the four outcomes simultaneously, as we
have done in the current study. We believe this joint analysis of the four different aspects of school climate simultaneously to be a
strength of the study, as it allows for a more comprehensive understanding of the broader construct of school climate (Wilson,
2004). Due to the nature of SEM, there may be competing models that fit as well or better. However, we chose this model due to a
strong theoretical rationale.

4.2. Implications for future research and school psychology practice

The current findings have several important implications for research and practice. These results extend previous research
indicating that students in well-managed classrooms are more likely to behave better (Lewis, 2001) to suggest that they may also
perceive the climate more favorably. Future studies should consider additional measures of exclusionary discipline practices, such
as the total number of referrals to the principal's office for each student, rather than number of students receiving a referral.
Similarly, the timing of the referral may also be important, whereby receiving a referral early in the school year may have less
impact on a student's perception of school climate at the end of the year (as was assessed in the current study) than a student
who received a referral in closer proximity to the assessment of school climate. This alternative operationalization of exclusionary
disciplinary practices would allow researchers to examine the magnitude of these practices in terms of multiple offenses per
student. Other indicators of exclusionary or punitive discipline could be considered for use in future studies, because referrals can
occur in response to minor offenses such as disrespect or to major problems such as fighting. It is possible that these associations
could vary by the type of behavioral infractions. For example, more serious behavior problems that lead to a referral to the
principal's office may be perceived as justified by students, whereas minor infractions that lead to exclusionary disciplinary
practices may be perceived as harsh. In contrast, high levels of classroom disruption that are not dealt with effectively may also
signal that the classroom is out of the teacher's control (Koth et al., 2008). These factors could in turn contribute to a less favorable
perception of the climate.
Future analyses could incorporate additional school-level covariates, such as student mobility and faculty turnover, which
have been associated with student-rated climate in previous studies (Mitchell et al., 2010). In addition, further research is needed
with students at diverse grade levels to determine the extent to which our findings generalize across a variety of children. Our
findings also highlight the importance of targeting teachers who use high levels of exclusionary discipline strategies to manage
their classrooms. These results support previous research, which has indicated that using exclusionary discipline strategies can
promote negative feelings between teachers and students in the classroom (Nelson & Roberts, 2000).
With regard to implications for school psychology practice, professional development should be provided to educate teachers
on the potential iatrogenic effects of exclusionary discipline practices on student perceptions of school climate. Helping teachers
to replace the use of exclusionary discipline practices with proactive management strategies, such as using precorrections,
establishing classroom rules and expectations, the use of global and specific praise, and involving students in setting classroom
rules for expected behaviors, may assist teachers in managing the classroom environment and enhancing students' perceptions of
the schools' climate (for review, see Lewis, 2001; Lewis et al., 2005; Sugai et al., 1999). The adoption of these alternative strategies
should in turn reduce the need for office referrals by preventing the occurrence of disciplinary problems (Lewis, 2001). Moreover,
the integration of a classroom system of support within the broader school-wide positive behavior support system also holds
great promise for improving the overall climate of schools and student discipline (Horner et al., 2009; Sugai & Horner, 2006).
Although these findings are correlational, they do suggest that school psychologists interested in promoting a more
supportive school climate could focus on helping teachers to implement positively oriented classroom management
strategies; such approaches may translate into more favorable perceptions of school climate, as well as reduce rates of
behavior problems and optimize academic outcomes (e.g., Bradshaw, Zmuda, Kellam, & Ialongo, 2009; Epstein et al., 2008;
Lewis et al., 2005).
 M.M. Mitchell, C.P. Bradshaw / Journal of School Psychology 51 (2013) 599–610 609

References

Arbuckle, C., & Little, E. (2004). Teachers' perceptions and management of disruptive classroom behaviour during the middle years (years five to nine). Australian
Journal of Educational and Developmental Psychology, 4, 59–70.
Baldwin, S. A., Murray, D. M., Shadish, W. R., Pals, S. L., Holland, J. M., Abramowitz, J. S., et al. (2011). Intraclass correlation associated with therapists: Estimates
and applications in planning psychotherapy research. Cognitive Behaviour Therapy, 40, 15–33. http://dx.doi.org/10.1080/16506073.2010.520731.
Battistich, V., Solomon, D., Kim, D., Watson, M., & Schaps, E. (1995). Schools as communities, poverty levels of student populations, and students' attitudes,
motives, and performance: A multilevel analysis. American Educational Research Journal, 32, 627–658. http://dx.doi.org/10.3102/00028312032003627.
Bevans, K. B., Bradshaw, C. P., Miech, R., & Leaf, P. J. (2007). Staff- and school-level predictors of school organizational health: A multilevel analysis. Journal of
School Health, 77, 294–302. http://dx.doi.org/10.1111/j.1746-1561.2007.00210.x.
Birnbaum, A. S., Lytle, L. A., Hannan, P. J., Murray, D. M., Perry, C. L., & Forster, J. L. (2003). School functioning and violent behavior among young adolescents: A
contextual analysis. Health Education Research, 18, 389–403. http://dx.doi.org/10.1093/her/cyf036.
Bovaird, J. A. (2007). Multilevel structural equation models for contextual factors. In T. D. Little, J. A. Bovaird, & N. A. Card (Eds.), Modeling contextual effects in
longitudinal studies (pp. 149–182). Mahwah, NJ: Erlbaum.
Boyle, M. H., & Pickles, A. R. (1998). Strategies to manipulate reliability: Impact on statistical associations. Journal of the American Academy of Child and Adolescent
Psychiatry, 37(10), 1077–1084. http://dx.doi.org/10.1097/00004583-199810000-00018.
Bradshaw, C. P., Koth, C. W., Thornton, L. A., & Leaf, P. J. (2009a). Altering school climate through School-wide Positive Behavioral Interventions and Supports:
Findings from a group-randomized effectiveness trial. Prevention Science, 10, 100–115. http://dx.doi.org/10.1007/s11121-008-0114-9.
Bradshaw, C. P., Mitchell, M. M., & Leaf, P. J. (2010). Examining the effects of School-wide Positive Behavioral Interventions and Supports on student outcomes: Results from a
randomized controlled effectiveness trial in elementary schools. Journal of Positive Behavior Interventions, 12, 133–148. http://dx.doi.org/10.1177/1098300709334798.
Bradshaw, C. P., O'Brennan, L. M., & McNeely, C. A. (2008a). Core competencies and the prevention of school failure and early school leaving. New Directions for
Child and Adolescent Development, 122, 19–32. http://dx.doi.org/10.1002/cd.226.
Bradshaw, C. P., Reinke, W. M., Brown, L. D., Bevans, K. B., & Leaf, P. J. (2008b). Implementation of School-wide Positive Behavioral Interventions and Supports
(PBIS) in elementary schools: Observations from a randomized trial. Education and Treatment of Children, 31, 1–26. http://dx.doi.org/10.1353/etc.0.0025.
Bradshaw, C. P., Sawyer, A. L., & O'Brennan, L. M. (2009b). A social disorganization perspective on bullying-related attitudes and behaviors: The influence of school
context. American Journal of Community Psychology, 43, 204–220. http://dx.doi.org/10.1007/s10464-009-9240-1.
Bradshaw, C. P., Waasdorp, T. E., & Leaf, P. J. (2012). Effects of School-wide Positive Behavioral Interventions and Supports on child behavior problems. Pediatrics,
130, e1136–e1145. http://dx.doi.org/10.1542/peds.2012-0243.
Bradshaw, C. P., Zmuda, J. H., Kellam, S. G., & Ialongo, N. S. (2009c). Longitudinal impact of two universal preventive interventions in first grade on educational
outcomes in high school. Journal of Educational Psychology, 101(4), 926–937. http://dx.doi.org/10.1037/a0016586.
Braham, D. (2004). The wise man builds his house upon the rock: The effects of inadequate school building infrastructure on student attendance. Social Science
Quarterly, 85, 1112–1128. http://dx.doi.org/10.1111/j.0038-4941.2004.00266.x.
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. http://dx.doi.org/10.1037/0022-0663.95.3.570.
Bronfenbrenner, U. (1979). The ecology of human development: Experiments by nature and design. Cambridge, MA: Harvard University Press.
DeVellis, R. F. (1991). Scale development. Newbury Park, NJ: Sages.
Emmons, C. L., Comer, J. P., & Haynes, N. M. (1996). Translating theory into practice: Comer's theory of school reform. In J. P. Comer, N. M. Haynes, E. Joyner, & M.
Ben-Avie (Eds.), Rallying the whole village (pp. 27–41). New York, NY: Teachers College Press.
Epstein, M., Atkins, M., Cullinan, D., Kutash, K., Weaver, R., & Woodbridge, M. (2008). Reducing behavior problems in the elementary school classroom. Institute of
Education Sciences (Retrieved from http://ies.ed.gov/ncee/wwc/pdf/practice_guides/behavior_pg_092308.pdf).
Finn, J. D., Pannozzo, G. M., & Achilles, C. M. (2003). The “why's” of class size: Student behavior in small classes. Review of Educational Research, 73, 321–368.
http://dx.doi.org/10.3102/00346543073003321.
Gottfredson, G. D., & Gottfredson, D. C. (2001). What schools do to prevent problem behavior and promote safe environments. Journal of Educational and
Psychological Consultation, 12, 313–344. http://dx.doi.org/10.1207/S1532768XJEPC1204_02.
Griffith, J. (1999). School climate as “social order” and “social action” A multi-level analysis of public elementary school student perceptions. Social Psychology of
Education, 2, 339–369. http://dx.doi.org/10.1023/A:1009657422344.
Hancock, G. R., & Mueller, R. O. (2001). Rethinking construct reliability within latent variable systems. In R. Cudeck, S. du Toit, & D. Sorbom (Eds.), Structural
equation modeling: Present and future. Festschrift in honor of Karl Joreskog (pp. 195–216). Lincolnwood, IL: Scientific Software International, Inc.
Haynes, N. M., Emmons, C., & Ben-Avie, M. (1997). School climate as a factor in student adjustment and achievement. Journal of Educational & Psychological
Consulting, 8, 321–329. http://dx.doi.org/10.1207/s1532768xjepc0803_4.
Horner, R. H., Sugai, G., Smolkowski, K., Eber, L., Nakasato, J., Todd, A. W., et al. (2009). A randomized, wait-list controlled effectiveness trail assessing school-wide
positive behavior supports in elementary schools. Journal of Positive Behavior Interventions, 11, 133–144. http://dx.doi.org/10.1177/1098300709332067.
Hoy, W. K., & Woolfolk, A. E. (1993). Teachers' sense of efficacy and the organizational health of schools. The Elementary School Journal, 93, 355–372.
http://dx.doi.org/10.2307/1002017.
Irvin, L., Tobin, T., Sprague, J., Sugai, G., & Vincent, C. (2004). Validity of office discipline referral measures as indices of school-wide behavioral status and effects of
school-wide behavioral interventions. Journal of Positive Behavior Interventions, 6, 131–147. http://dx.doi.org/10.1177/10983007040060030201.
Koth, C. W., Bradshaw, C. P., & Leaf, P. J. (2008). A multilevel study of predictors of student perceptions of school climate: The effect of classroom-level factors.
Journal of Educational Psychology, 100, 96–104. http://dx.doi.org/10.1037/0022-0663.100.1.96.
Kuperminc, G. P., Leadbeater, B. J., Emmons, C., & Blatt, S. J. (1997). Perceived school climate and difficulties in the social adjustment of middle school students.
Applied Developmental Science, 1, 76–88. http://dx.doi.org/10.1207/s1532480xads0102_2.
Lewis, R. (2001). Classroom discipline and student responsibility: The students' view. Teaching and Teacher Education, 17, 307–319. http://dx.doi.org/10.1016/
S0742-051X(00)00059-7.
Lewis, R., Romi, S., Qui, X., & Katz, Y. J. (2005). Teachers' classroom discipline and student misbehavior in Australia, China and Israel. Teaching and Teacher
Education, 21, 729–741. http://dx.doi.org/10.1016/j.tate.2005.05.008.
Little, S. G., & Akin-Little, A. (2008). Psychology's contributions to classroom management. Psychology in the Schools, 45, 227–234. http://dx.doi.org/10.1002/pits.20293.
Little, T. D., Lindenberger, U., & Nesselroade, J. R. (1999). On selecting indicators for multivariate measurement and modeling with latent variables: When “good”
indicators are bad and “bad” indicators are good. Psychological Methods, 4, 192–211. http://dx.doi.org/10.1037//1082-989X.4.2.192.
Luke, P. A. (2004). Multilevel modeling. Thousand Oaks, California: Sage.
Marsh, H. W., Lüdtke, O., Robitzsch, A., Trautwein, U., Asparouhov, T., Muthén, B., et al. (2009). Doubly-latent models of school contextual effects:
Integrating multilevel and structural equation approaches to control measurement and sampling error. Multivariate Behavioral Research, 44, 764–802.
http://dx.doi.org/10.1080/00273170903333665.
McIntosh, K., Campbell, A. L., Carter, D. R., & Zumbo, B. D. (2009). Concurrent validity of office discipline referrals and cut points used in schoolwide positive
behavior support. Behavioral Disorders, 34, 100–113.
Mehta, P. D., & Neale, M. C. (2005). People are variables too: Multilevel structural equations modeling. Psychological Methods, 10, 259–284. http://dx.doi.org/
10.1037/1082-989X.10.3.259.
Mijanovich, T., & Weitzman, B. C. (2003). Which “broken windows” matter?: School, neighborhood, and family characteristics associated with youths' feelings of
unsafety. Journal of Urban Health, 80, 400–415. http://dx.doi.org/10.1093/jurban/jtg045.
Miller, A., Ferguson, E., & Byrne, I. (2000). Pupils' causal attributions for difficult classroom behaviour. British Journal of Educational Psychology, 70, 85–96.
http://dx.doi.org/10.1348/000709900157985.
610 M.M. Mitchell, C.P. Bradshaw / Journal of School Psychology 51 (2013) 599–610

Mitchell, M. M., Bradshaw, C. P., & Leaf, P. J. (2010). Student and teacher perceptions of school climate: A multilevel exploration of patterns of discrepancy. Journal
of School Health, 80, 271–279. http://dx.doi.org/10.1111/j.1746-1561.2010.00501.x.
Muthén, B. O. (2002a). Beyond SEM: General latent variable modeling. Behaviormetrika, 29, 81–117. http://dx.doi.org/10.2333/bhmk.29.81.
Muthén, L. K. (2002b). Mplus discussion board: Intraclass correlations. Available at: http://www.statmodel.com/discussion/messages/12/18.html?1344285647
(Retrieved on October 16, 2012).
Muthén, L. K., & Muthén, B. O. (1998–2010). Mplus user's guide (5th ed.)Los Angeles, CA: Muthén & Muthén.
Nelson, J. R., Benner, G. J., Reid, R. C., Epstein, M. H., & Currin, D. (2002). The convergent validity of office discipline referrals with the CBCL-TRF. Journal of Emotional
and Behavioral Disorders, 10, 181–188. http://dx.doi.org/10.1177/10634266020100030601.
Nelson, J. R., & Roberts, M. (2000). Ongoing reciprocal teacher–student interactions involving disruptive behavior in general education classrooms. Journal of
Emotional and Behavior Disorders, 8, 27–37. http://dx.doi.org/10.1177/106342660000800104.
Pas, E., Bradshaw, C. P., & Mitchell, M. M. (2011). Examining the validity of office discipline referrals as an indicator of student behavior problems. Psychology in the
Schools, 48(6), 541–555. http://dx.doi.org/10.1002/pits.20577.
Plank, S. B., Bradshaw, C. P., & Young, H. (2009). An application of “broken windows” and related theories to the study of disorder, fear, and collective efficacy in
schools. American Journal of Education, 115, 227–247. http://dx.doi.org/10.1086/595669.
Preacher, K. J., Zyphur, M. J., & Zhang, Z. (2010). A general multilevel SEM framework for assessing multilevel mediation. Psychological Methods, 15(3), 209–233.
http://dx.doi.org/10.1037/a0020141.
Rabe-Hesketh, S., Skrondal, A., & Zheng, X. (2012). Multilevel structural equation modeling. In R. Hoyle (Ed.), Handbook of Structural Equation Modeling
(pp. 512–531). New York: Guilford.
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods (2nd ed.)Thousand Oaks, CA: Sage.
Raudenbush, S. W., Bryk, A. S., Cheong, A. S., Fai, Y. F., Congdon, R. T., & du Toit, M. (2011). HLM 7: Hierarchical linear and nonlinear modeling. Lincolnwood, IL:
Scientific Software International.
Safran, S. P. (2006). Using the Effective Behavior Supports Survey to guide development of Schoolwide Positive Behavior Support. Journal of Positive Behavior
Interventions, 8(1), 3–9. http://dx.doi.org/10.1177/10983007060080010201.
Shochet, I. M., Dadds, M. R., Ham, D., & Montague, R. (2006). School connectedness is an underemphasized parameter in adolescent mental health: Results of a
community prediction study. Journal of Clinical Child and Adolescent Psychology, 35, 170–179. http://dx.doi.org/10.1207/s15374424jccp3502_1.
Sugai, G., & Horner, R. (2006). A promising approach for expanding and sustaining school-wide positive behavior support. School Psychology Review, 35, 245–259.
Sugai, G., Horner, R. H., Dunlap, G., Hieneman, M., Lewis, T. J., & Nelson, C. M. (1999). Applying positive behavioral support and functional behavioral assessment in
schools (technical assistance guide 1, version 1.4.4). Eugene: University of Oregon, Center on Positive Behavioral Interventions and Support.
Sugai, G., Sprague, J. R., Horner, R. H., & Walker, H. M. (2000a). Preventing school violence: The use of office discipline referrals to assess and monitor school-wide
discipline interventions. Journal of Emotional and Behavioral Disorders, 8(2), 94–102. http://dx.doi.org/10.1177/106342660000800205.
Sugai, G., Todd, A. W., & Horner, R. H. (2000b). Effective Behavior Support (EBS) Survey: Assessing and planning behavior supports in schools. Eugene, OR: University
of Oregon.
Waasdorp, T. E., Bradshaw, C. P., & Leaf, P. J. (2012). The impact of School-wide Positive Behavioral Interventions and Supports (SWPBIS) on bullying and peer rejection:
A randomized controlled effectiveness trial. Archives of Pediatrics & Adolescent Medicine, 116, 149–156. http://dx.doi.org/10.1001/archpediatrics.2011.755.
Walker, B., Cheney, D., & Stage, S. (2009). The validity and reliability of the self-assessment and program review: Assessing school progress in Schoolwide Positive
Behavior Support. Journal of Positive Behavior Interventions, 11, 94–109. http://dx.doi.org/10.1177/1098300708322445.
Way, N., Reddy, R., & Rhodes, J. (2007). Students' perceptions of school climate during the middle school years: Associations with trajectories of psychological and
behavioral adjustment. American Journal of Community Psychology, 40, 194–213. http://dx.doi.org/10.1007/s10464-007-9143-y.
Wilson, D. (2004). The interface of school climate and school connectedness and relationships with aggression and victimization. Journal of School Health, 74,
293–299. http://dx.doi.org/10.1111/j.1746-1561.2004.tb08286.x.
Zucker, D. M. (1990). An analysis of variance pitfall: The fixed effects analysis in a nested design. Educational and Psychological Measurement, 50, 731–738.
http://dx.doi.org/10.1177/0013164490504002.