School_climate_family_structu.pdf

School Climate, Family Structure, and Academic Achievement:
A Study of Moderation Effects

Meagan O’Malley and Adam Voight
WestEd, Los Alamitos, California

Tyler L. Renshaw
Louisiana State University

Katie Eklund
University of Arizona

School climate has been lauded for its relationship to a host of desirable academic,
behavioral, and social– emotional outcomes for youth. The present study tested the
hypothesis that school climate counteracts youths’ home–school risk by examining the
moderating effects of students’ school climate perceptions on the relationship between
family structure (i.e., two-parent, one-parent, foster-care, and homeless households),
and academic performance (i.e., self-reported [grade point average] GPA). The present
sample consisted of 902 California public high schools, including responses from over
490,000 students in Grades 9 and 11. Results indicated that, regardless of family
structure, students with more positive school climate perceptions self-reported higher
GPAs. Youths with two-parent, one-parent, and homeless family structures displayed
stepwise, linear improvements in self-reported GPA as perceptions of climate im-
proved. Foster-care students’ positive school climate perceptions had a weaker effect
on their self-reported GPA compared with students living in other family structures. A
unique curvilinear trend was found for homeless students, as the relationship between
their school climate perceptions and self-reported GPA was stronger at lower levels.
Overall, the moderation effect of positive school climate perceptions on self-reported
GPA was strongest for homeless youth and youth from one-parent homes, suggesting
that school climate has a protective effect for students living in these family structures.
A protective effect was not found for youth in foster-care. Implications for research and
practice are discussed.

Keywords: school climate, family structure, academic achievement, foster care, homeless

A child’s home and school represent distinct
microsystems that shape important develop-
mental outcomes. Each microsystem presents
its own set of potential risks, or threats to de-
velopment, as well as protective and promotive

factors that counteract those threats and facili-
tate wellbeing (Masten, Herbers, Cutuli, &
Lafavor, 2008). Promotive factors are those in-
ternal or external assets (e.g., supportive adult
relationships at home) that improve outcomes
for all youths, but that do not help close the
academic and social outcome gaps between at-
risk students and their peers. Protective factors,
on the other hand, are those assets (e.g., positive
school relationships) that disproportionately
bolster the outcomes of youth who are identified
as at-risk, thereby decreasing the outcome gaps
between them and their peers (Furlong, Shar-
key, Quirk, & Dowdy, 2011). Moreover, risk
factors, which are the negative counterparts of
promotive and protective factors, are those in-
ternal or external stressors (e.g., poverty) that
hinder or worsen youths’ outcomes. The accu-
mulation of multiple risk factors across micro-

This article was published Online First August 11, 2014.
Meagan O’Malley and Adam Voight, Health and Human

Development Program, WestEd, Los Alamitos, California;
Tyler L. Renshaw, Department of Psychology, Louisiana
State University; Katie Eklund, College of Education, Uni-
versity of Arizona.

We thank the California Department of Education for
encouraging the collection and analysis of these California
Healthy Kids Survey data, which yield important insights
into the wellbeing of youth.

Correspondence concerning this article should be ad-
dressed to Meagan O’Malley, Health and Human Develop-
ment Program, WestEd, 4665 Lampson Avenue, Los
Alamitos, CA 90720-5139. E-mail: [email protected]

T
hi

s
do

cu
m

en
t

is
co

py
ri

gh
te

d
by

th
e

A
m

er
ic

an
P

sy
ch

ol
og

ic
al

A
ss

oc
ia

ti
on

or
on

e
of

it
s

al
li

ed
pu

bl
is

he
rs

.
T

hi
s

ar
ti

cl
e

is
in

te
nd

ed
so

le
ly

fo
r

th
e

pe
rs

on
al

us
e

of
th

e
in

di
vi

du
al

us
er

an
d

is
no

t
to

be
di

ss
em

in
at

ed
br

oa
dl

y.

School Psychology Quarterly © 2014 American Psychological Association
2015, Vol. 30, No. 1, 142–157 1045-3830/15/$12.00 http://dx.doi.org/10.1037/spq0000076

142

mailto:[email protected]

http://dx.doi.org/10.1037/spq0000076

systems—referred to as cumulative risk—is
more predictive of deleterious outcomes in
youth than exposure to any one risk alone (Ob-
radović, Shaffer, & Masten, 2012; Sameroff,
2006).

The nature of the relationship between home
and school microsystems has been examined for
decades using a unidirectional model, wherein
home environment characteristics are analyzed
for their effects on school-related outcomes.
Because of the abundance of work in this area,
scholars of economics, sociology, education,
and psychology now largely agree that parental
education, socioeconomic status, and family
structure have main effects on the academic
performance of youth (Coleman et al., 1966;
Reardon, 2011) and an aggregated effect on
system-level school performance indicators,
such as state-mandated standardized test scores
(Bryk, Sebring, Allensworth, Luppescu, & Eas-
ton, 2010). In other words, students who expe-
rience more family related advantages and re-
sources at home tend to perform better
academically, and schools that enroll higher
proportions of these students tend to be deemed
successful by state accountability standards.

More recently, this line of research has been
extended to examine the possibility that school
environments may play an important role in
counteracting or buffering youths’ home-related
risks, especially as they relate to school perfor-
mance. For instance, Hopson and Lee (2011)
found that students’ positive school climate per-
ceptions attenuated the relationship between
family poverty and negative behavioral out-
comes at school (e.g., truancy, fights with other
students, and suspensions). Moreover, students’
positive school climate perceptions are nega-
tively related to disruptive school behaviors for
all students, with the strongest effect observed
for youth living in low-income homes (Hopson
& Lee, 2011). Furthermore, in their analysis of
the relations between students’ standardized test
scores and school climate perceptions, Voight,
Austin, and Hanson (2013) demonstrated that
students’ positive school climate perceptions
were a distinguishing indicator of secondary
schools that outperformed academic achieve-
ment (indicated by state-mandated standardized
test scores) predictions based on their student
demographics.

The present study extends this line of re-
search by examining whether students’ percep-

tions of a positive school climate might function
as both a promotive and protective factor: en-
hancing the academic achievement of all youth,
while offering greater benefits to youth living in
more disadvantaged family structures. Thus, be-
cause the educational research literature is re-
plete with studies investigating factors affecting
students’ school performance—with race, eth-
nicity, and socioeconomic status being chief
among these (e.g., Coleman et al., 1966; Dun-
can & Murnane, 2011; Sirin, 2005)—the re-
mainder of the introduction focuses squarely on
the two major variables of interest in the present
study—family structure and school climate—
and their relationship with academic outcomes.

Family Structure

Family structure is a sociodemographic indi-
cator that differentiates youths’ living situations
by the number of caregiving adults with whom
they live. Family structure is sometimes mea-
sured as a proxy for family stress, as decades of
research have demonstrated that children living
in single-parent homes are more likely than
those living in two-parent homes to experience
poverty and its associated stressors (McLana-
han & Garfinkel, 2000). Closely related to fam-
ily structure, parental involvement is generally
defined as a constellation of adult behaviors that
support youths’ achievement of school-related
demands (El Nokali, Bachman, & Votruba-
Drzal, 2010). Because of the myriad demands
placed upon them, single parents may have less
time available to commit to their child’s school
life than can be provided in two-parent homes
(Jeynes, 2005). Indeed, children living in sin-
gle-parent homes receive, on average, less aca-
demic encouragement and guidance at home
than their peers living in two-parent families
(Astone & McLanahan, 1991).

Single-Parent and Two-Parent Students

Family structure is sometimes considered in
research designs investigating the relationship
between parental involvement and school per-
formance. For instance, using data from a large
national dataset with predominantly White
(69%) high school-aged students, Jeynes (2005)
found that, compared with youth living in sin-
gle-parent homes (i.e., single-parent never mar-
ried, divorced), children living in two-parent

143SCHOOL CLIMATE, FAMILY STRUCTURE, AND ACHIEVEMENT

T
hi

s
do

cu
m

en
t

is
co

py
ri

gh
te

d
by

th
e

A
m

er
ic

an
P

sy
ch

ol
og

ic
al

A
ss

oc
ia

ti
on

or
on

e
of

it
s

al
li

ed
pu

bl
is

he
rs

.
T

hi
s

ar
ti

cl
e

is
in

te
nd

ed
so

le
ly

fo
r

th
e

pe
rs

on
al

us
e

of
th

e
in

di
vi

du
al

us
er

an
d

is
no

t
to

be
di

ss
em

in
at

ed
br

oa
dl

y.

homes displayed superior academic achieve-
ment across various academic outcomes (i.e.,
reading, math, science, and social science stan-
dardized test results) than children from single-
parent homes (Jeynes, 2005). Moreover, family
structure proved to be a more powerful predic-
tor of students’ academic achievement than the
types of parental involvement behaviors (e.g.,
checking on homework) that are often encour-
aged by school personnel. In a related study, El
Nolaki, Bachman, and Votruba-Drzal (2010)
found that, after controlling for family structure
and other known covariates in a sample of ele-
mentary-aged students, parent involvement ex-
erted no meaningful effect on students’ aca-
demic achievement. Disheartening though they
may be, these findings suggest that living in a
single-parent home asserts a level of risk that
even a conscientious home adult might not be
able to overcome alone. To our knowledge,
similar research has yet to be conducted with
secondary school students.

Foster-Care Students

Approximately 1% of all children in the
United States are currently placed in the foster-
care system (Pecora et al., 2006). Placement in
foster-care settings is typically prompted by the
deterioration of caregiving in the child’s native
home setting (Lawrence, Carlson, & Egeland,
2006). Although many children placed in foster
care are there for a short period of time, nearly
half of all children remain in foster care for a
year or longer (U.S. Department of Health &
Human Services, 2005). Once placed in foster-
care settings, youths’ experiences are often
characterized by prolonged separation from par-
ents and family members, disrupted relation-
ships with foster parents, and frequent transi-
tions between foster placements (Newton,
Litrownik, & Landsverk, 2000; U.S. Depart-
ment of Health & Human Services, 2012). For
these reasons, youths in foster care often have
family structures characterized by unstable
caregiver relationships.

Current, methodologically advanced studies
examining school-related outcomes for youths
living in foster care are scant in the peer-
reviewed literature. More frequently, analyses
of foster-care youths’ academic achievement
are conducted by state and federal public health
agencies. For instance, in their analysis of fos-

ter-care youths’ academic achievement in the
state of Washington, Burley and Halpern (2001)
found that, compared with their nonfoster peers,
youths in foster-care homes scored lower on
state achievement tests, had lower graduation
rates, were more likely to be enrolled in special
education, and were more likely to have re-
peated a grade or changed schools during the
school year. In addition to suffering from more
frequent and debilitating mental health prob-
lems, children in foster care are also at greater
risk of becoming involved in criminal activities,
dropping out of school, and discontinuing post-
secondary education (Pecora et al., 2006; Rubin
et al., 2004). Taken together, these findings
suggest that youth in foster care are among
those in greatest need of thoughtfully designed,
enriched, developmentally supportive school
environments that are intended to prevent dele-
terious outcomes and promote their wellbeing.

Homeless Students

Over 1 million youth are estimated to be
homeless in the United States, although precise
rates are difficult to determine because of high
residential mobility and the variability of con-
struct definitions used for policy purposes (Perl
et al., 2013). Despite such measurement chal-
lenges, researchers have made progress in ex-
amining the experiences of homeless children
and youth. For example, studies have shown
that homeless children are at higher risk for a
variety of emotional, behavioral, and academic
challenges, although poverty likely accounts for
some proportion of these effects. Research has
also indicated that homeless children experi-
ence more mental health and behavior problems
than peers living in average-income homes, al-
though when poverty is accounted for, the mag-
nitude of the difference between groups de-
creases (Buckner, 2008; Samuels, Shinn, &
Buckner, 2010). Children who are homeless and
highly mobile have also been shown to have
lower reading and math achievement than their
housed counterparts, even after the effects of
poverty are controlled (Obradović et al., 2009;
Rafferty, 2004; Voight, Shinn, & Nation, 2012).

In addition to the poor mental health and
achievement outcomes associated with home-
lessness, the family structure of homeless chil-
dren is often fragile, characterized by increased
risk for parent– child separation. For instance,

144 O’MALLEY, VOIGHT, RENSHAW, AND EKLUND

T
hi

s
do

cu
m

en
t

is
co

py
ri

gh
te

d
by

th
e

A
m

er
ic

an
P

sy
ch

ol
og

ic
al

A
ss

oc
ia

ti
on

or
on

e
of

it
s

al
li

ed
pu

bl
is

he
rs

.
T

hi
s

ar
ti

cl
e

is
in

te
nd

ed
so

le
ly

fo
r

th
e

pe
rs

on
al

us
e

of
th

e
in

di
vi

du
al

us
er

an
d

is
no

t
to

be
di

ss
em

in
at

ed
br

oa
dl

y.

Bassuk and colleagues (1997) found that chil-
dren were more likely to be placed in foster care
when their families experienced homelessness
than when they experienced poverty. Indeed,
one study found that the likelihood of mother–
child separation is quadrupled after an experi-
ence of homelessness (Cowal, Shinn, Weitz-
man, Stojanovic, & Laba, 2002). In another
study, nearly 20% of youth who entered shelters
with their families had prior or future experi-
ences in the state’s child welfare system (Park,
Metraux, Brodbar, & Culane, 2004). What is
more, the mobility associated with homeless-
ness exposes youth to the loss of social capital
gained by youth in stable homes who benefit
from persistent relationships with teachers and
friends in school settings (Rafferty, 2004).

School Climate

Some scholars have posited that develop-
mentally supportive school environments may
counteract the effects of adverse home expe-
riences through the reduction of cumulative
risk (Masten et al., 2008). One of the primary
school-level characteristics implicated in fos-
tering student resilience is a multidimensional
construct referred to as school climate (Mas-
ten & Motti-Stefanidi, 2009). One of the most
common definitions found within the educa-
tion-related literature defines school climate as
“the quality and character of school life,” which
is derived from “patterns of people’s experi-
ences of school life and reflects norms, goals,
values, interpersonal relationships, teaching,
learning and leadership practices, and organiza-
tional structures” (Cohen, McCabe, Michelli, &
Pickeral, 2009, p. 182). Thus, social support at
school, including relationships between and
among students and adults, is considered an
essential dimension of school climate (You,
O’Malley, & Furlong, 2014). Also included in
theoretical and measurement models of school
climate are opportunities for belonging and con-
nectedness, school safety, physical school re-
sources, and discipline practices (O’Malley,
Katz, Renshaw, & Furlong, 2012; Zullig, Koop-
man, Patton, & Ubbes, 2010).

Studies measuring the school climate percep-
tions of youth and school staff have indicated
that more positive perceptions of school climate
are associated with more desirable academic,
behavioral, and social– emotional outcomes.

Teachers’ perceptions of positive school cli-
mates, for example, have been positively related
with higher student achievement in reading and
mathematics (Brand, Felner, Seitsinger, Burns,
& Bolton, 2008; Uline & Tschannen-Moran,
2008). What is more, youths who perceive that
they attend schools with positive climates en-
gage in fewer risk-taking and violent behaviors
(Resnick et al., 1997), experience fewer disci-
pline referrals and school suspensions (Nelson,
Martella, & Marchand-Martella, 2002; Welsh,
2000), and report stronger feelings of safety and
greater willingness to report safety threats at
school (Syvertsen, Flanagan, & Stout, 2009;
Welsh, 2000). Furthermore, research has dem-
onstrated that students’ school climate percep-
tions are related to their complete mental health
status, with positive climate perceptions being
associated with both increases in life satisfac-
tion and decreases in internalizing and external-
izing symptoms (Suldo, McMahan, Chappel, &
Loker, 2012). Such findings suggest that school
climate may have moderating effects on various
student wellbeing outcomes.

Purpose of the Present Study

Although cross-sectional research indicating
relations between students’ positive school cli-
mate perceptions and desirable academic, be-
havioral, and social– emotional outcomes has
burgeoned recently, the particular dynamics of
school climate, such as its potential moderating
effects on students’ exposure to various risk
factors, have gone largely unexplored. Specifi-
cally, there is little available empirical evidence
supporting the assertion that positive school cli-
mates can compensate for the risk associated
with living in disadvantaged family structures.
The present study intended to address this gap
in the literature by investigating two related
research questions:

1. Are students’ positive school climate per-
ceptions associated with improved aca-
demic outcomes for youth living in differ-
ent family structures (i.e., single-parent,
two-parent, foster-care, and homeless)?

2. Are students’ positive school climate per-
ceptions associated with a reduction in the
academic achievement gap commonly ob-
served between at-risk youth and their
peers?

145SCHOOL CLIMATE, FAMILY STRUCTURE, AND ACHIEVEMENT

T
hi

s
do

cu
m

en
t

is
co

py
ri

gh
te

d
by

th
e

A
m

er
ic

an
P

sy
ch

ol
og

ic
al

A
ss

oc
ia

ti
on

or
on

e
of

it
s

al
li

ed
pu

bl
is

he
rs

.
T

hi
s

ar
ti

cl
e

is
in

te
nd

ed
so

le
ly

fo
r

th
e

pe
rs

on
al

us
e

of
th

e
in

di
vi

du
al

us
er

an
d

is
no

t
to

be
di

ss
em

in
at

ed
br

oa
dl

y.

Given these questions, we hypothesized that
students’ school climate perceptions would op-
erate as a protective factor, yielding greater
moderation effects on academic achievement
for youth living in higher-risk family structures
(i.e., single-parent, foster-care, and homeless
households) compared with youth living in low-
er-risk family structures (i.e., two-parent house-
holds).

Method

Sampling Procedure and Student
Demographics

This study is based on existing survey data
collected from �1.5 million high school stu-
dents in 902 California public high schools
from 2008 to 2010. The sample was limited to
schools that administered the California
Healthy Kids Survey (CHKS; WestEd, 2014), a
state-adopted measure of student risk and pro-
tective factors, during the 2008 –2009 and
2009 –2010 school years. In a recent psycho-
metric analysis of the CHKS conducted for the
California Department of Education (2014),
Hanson (2012) reported that the measure was
characterized by a nine-factor latent structure
(i.e., school connectedness, relationships with
school adults, opportunities for meaningful par-
ticipation, perceived school safety, positive
learning environment, low racial or ethnic ten-
sion, low substance use, low violence victim-
ization, and low violence perpetration). A single
administration of the CHKS was required of
California public schools during the 2008 –2009
to 2009 –2010 period as a condition of the Safe
and Drug-Free School and Communities (Title
IV) funding or the state Tobacco Use Preven-
tion Education (TUPE) program. Approxi-
mately two thirds of all comprehensive public
high schools in the state had students complete
the survey. However, in one large district, only
a small sample of the entire population of
schools completed the survey. Other schools did
not administer the survey because of not receiv-
ing Title IV or TUPE funding, being exempt
from this requirement under the Rural Educa-
tion Achievement Program, or for unknown rea-
sons. Parental permission was collected via pas-
sive or active consent procedures, which were
determined at the discretion of individual
school districts. Survey proctor instructions

specified that students were to be informed that
their participation in the survey was both vol-
untary and anonymous.

At the high-school level, the CHKS is only
administered to students in Grades 9 and 11.
The present sample was comprised of 305,956
students in Grade 9 (61.6%) and 190,946 stu-
dents in Grade 11 (38.4%). In each of the 2
years of data collection, there were slightly over
one million public school students in Grades 9
and 11 combined in California (California De-
partment of Education, 2014); thus, the study
sample constitutes approximately half of all stu-
dents in Grades 9 and 11 statewide. The average
age of sample students was 15.09 years, and
52.4% of the sample was female. In terms of
race, 45.4% of sample students identified as
Latino/a, 33.7% as White, 17.2% as Asian,
7.1% as Black, 3.8% as American Indian, and
8.5% as another race. For comparison, accord-
ing to the California Department of Education
(2014), in the 2009 –2010 school year, 48.8% of
all students in Grades 9 and 11 statewide were
Latino/a; 28.4% were White; 11.9% were
Asian; 7.4% were Black; and 0.8% were Amer-
ican Indian. See Table 2 for a full presentation
of available student demographics data.

Measures

Family structure. The family structure of
sample students was measured using a single
survey question on the CHKS, “What best
describes where you live?,” which had 12
response options, including “A home with
both parents,” “A home with only one par-
ent,” “Foster home, group care, or waiting
placement,” and “On the street (no fixed hous-
ing), car or van, park campground or abandoned
building.” Given students’ response to this item,
they were classified into one of four family
structure categories: (a) two-parent, (b) one-
parent, (c) foster-care, and (d) homeless (that
included all home statuses that meet the defini-
tion of homelessness under the federal Housing
and Urban Development authority). Using this
classification system, 69.6% of participants re-
ported living in two-parent homes, 26.6% in
one-parent homes, 0.5% in foster-care homes,
and 1.8% in homeless households. One of the
unique advantages of this dataset for studying
homeless and foster-care students was that, de-
spite the low proportions of students repre-

146 O’MALLEY, VOIGHT, RENSHAW, AND EKLUND

T
hi

s
do

cu
m

en
t

is
co

py
ri

gh
te

d
by

th
e

A
m

er
ic

an
P

sy
ch

ol
og

ic
al

A
ss

oc
ia

ti
on

or
on

e
of

it
s

al
li

ed
pu

bl
is

he
rs

.
T

hi
s

ar
ti

cl
e

is
in

te
nd

ed
so

le
ly

fo
r

th
e

pe
rs

on
al

us
e

of
th

e
in

di
vi

du
al

us
er

an
d

is
no

t
to

be
di

ss
em

in
at

ed
br

oa
dl

y.

sented in these categories, there were still suf-
ficient overall group counts (n � 8,582 and n �
2,310, respectively) to power inferential analy-
ses.

School climate. Students’ perception of
school climate was calculated as the average of
four CHKS constructs identified by Hanson
(2012): (a) school connectedness (four items;
Cronbach’s � � .78 in the present sample), (b)
relationships with adults at school (six items;
� � .88), (c) opportunities for meaningful par-
ticipation in school (three items; � � .75), and
(d) perceived school safety (two items; � �
.69). Each of these first- constructs was the
average of two or more standardized survey
items (shown in Table 1), and each of the first-
construct scores were then averaged to
create a second- construct of school cli-
mate perceptions (� � .73), which was, in turn,
standardized to create z-scores. Descriptive sta-
tistics for this standardized school climate per-
ceptions variable indicated that students living
in two-parent homes had the most positive
school climate perceptions (M � 0.08, SD �
0.98), followed by students in one-parent homes
(M � �0.13, SD � 0.99), foster-care homes

(M � �0.31, SD � 1.14), and homeless house-
holds (M � �0.80, SD � 1.33).

Academic achievement. The outcome
variable in the present study was students’ self-
reported academic achievement, which was
measured using a single item from the CHKS,
“During the past 12 months, how would you
describe the grades you mostly received in
school?” This item had eight potential re-
sponse options, ranging from “Mostly A’s” to
“Mostly F’s.” Student responses were re-
coded such that a response of “Mostly A’s”
was scored 4.0, a response of “A’s and B’s”
was scored 3.5, and so on, with a response of
“Mostly F’s” being scored 0.0. In this sense,
this self-reported academic achievement met-
ric was conceptualized analogous to the stan-
dard four-point metric for grade point average
(GPA). Thus, for analog purposes, students’
self-reported academic achievement is re-
ferred to hereafter as GPA. Within the present
sample, students’ average GPA was 2.92
(SD � 0.96), with the highest mean GPA
observed for students living in two-parent
homes (M � 3.01, SD � 0.91), followed by
students in one-parent homes (M � 2.73,

Table 1
Survey Items Used in the Construction of the School Climate
Perceptions Variable

Relationships with adults at school
1. At my school there is an adult who really cares about me.
2. At my school there is an adult who tells me when I do a good job.
3. At my school there is an adult who notices when I am not there.
4. At my school there is an adult who always wants me to do my best.
5. At my school there is an adult who listens to me when I have something to say.
6. At my school there is an adult who believes I will be a success.

Opportunities for meaningful participation in school
7. At school, I do interesting activities.
8. At school, I help decide things like class activities or rules.
9. At school, I do things that make a difference.

Perceived school safety
10. I feel safe in my school.
11. How safe do you feel when you are at school?

School connectedness
12. I feel close to people at this school.
13. I am happy to be at this school.
14. I feel like I am a part of this school.
15. The teachers at this school treat students fairly.

Note. All items, with the exception of #11, use the response options: (1) “Strongly Dis-
agree”; (2) “Disagree”; (3) “Neither Disagree Nor Agree”; (4) “Agree”; or (5) “Strongly
Agree.” Item #11 uses the response options” (1) “Very Safe”; (2) “Safe”; (3) “Neither Safe
Nor Unsafe”; (4) “Unsafe”; (5) “Very Unsafe.” Item #11 was reverse coded. All item
responses were standardized before scale construction.

147SCHOOL CLIMATE, FAMILY STRUCTURE, AND ACHIEVEMENT

T
hi

s
do

cu
m

en
t

is
co

py
ri

gh
te

d
by

th
e

A
m

er
ic

an
P

sy
ch

ol
og

ic
al

A
ss

oc
ia

ti
on

or
on

e
of

it
s

al
li

ed
pu

bl
is

he
rs

.
T

hi
s

ar
ti

cl
e

is
in

te
nd

ed
so

le
ly

fo
r

th
e

pe
rs

on
al

us
e

of
th

e
in

di
vi

du
al

us
er

an
d

is
no

t
to

be
di

ss
em

in
at

ed
br

oa
dl

y.

SD � 1.01), homeless households (M � 2.60,
SD � 1.33), and foster-care homes (M �
2.53, SD � 1.18) (Table 2).

Analytic Approach

To examine the association of students’
school climate perceptions with their family
structure and GPA, we estimated a series of
regression models. All regression models were
estimated using a multilevel, random-intercept
approach that accounted for the clustered nature
of the data (i.e., students in schools). No school-
level variables were modeled, but the multilevel
approach renders …