Vous êtes sur la page 1sur 12

This article was downloaded by: [Assam University, Silchar]

On: 27 April 2015, At: 00:30


Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Journal of Clinical Child & Adolescent Psychology


Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/hcap20

Schools, Parents, and Youth Violence: A Multilevel,


Ecological Analysis
Kathryn A. Brookmeyer , Kostas A. Fanti & Christopher C. Henrich
Published online: 07 Jun 2010.

To cite this article: Kathryn A. Brookmeyer , Kostas A. Fanti & Christopher C. Henrich (2006) Schools, Parents, and Youth
Violence: A Multilevel, Ecological Analysis, Journal of Clinical Child & Adolescent Psychology, 35:4, 504-514, DOI: 10.1207/
s15374424jccp3504_2
To link to this article: http://dx.doi.org/10.1207/s15374424jccp3504_2

PLEASE SCROLL DOWN FOR ARTICLE


Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions

Journal of Clinical Child and Adolescent Psychology


2006, Vol. 35, No. 4, 504514

Copyright 2006 by
Lawrence Erlbaum Associates, Inc.

Schools, Parents, and Youth Violence: A Multilevel, Ecological Analysis


Kathryn A. Brookmeyer, Kostas A. Fanti, and Christopher C. Henrich

Downloaded by [Assam University, Silchar] at 00:30 27 April 2015

Georgia State University


Using data from the National Longitudinal Study of Adolescent Health (Add Health),
this study utilized an ecological approach to investigate the joint contribution of parents and schools on changes in violent behavior over time among a sample of 6,397
students (54% female) from 125 schools. This study examined the main and interactive effects of parent and school connectedness as buffers of violent behavior within a
hierarchical linear model, focusing on both students and schools as the unit of analysis. Results show that students who feel more connected to their schools demonstrate
reductions in violent behavior over time. On the school level, our findings suggest that
school climate serves as a protective factor for student violent behavior. Finally, parent and school connectedness appear to work together to buffer adolescents from the
effects of violence exposure on subsequent violent behavior.
adulthood (Lipsey & Derzon, 1998) and for violent
victimization over time (Henrich, Brookmeyer, Shahar,
& Kidd, 2005). Further, violence exposure, such as
witnessing or being victimized by violence, is a critical
risk factor for subsequent violent behavior among nonreferred youth (Farrell & Bruce, 1997; Miller, Wasserman, Neugebauer, Gorman-Smith, & Kamboukos,
1999), even after controlling for prior levels of aggression (Brookmeyer, Henrich, & Schwab-Stone, 2005;
Gorman-Smith & Tolan, 1998).

In 2001 the Surgeon General emphasized an urgent


need to confront the problem of youth violence in the
United States (U.S. Department of Health and Human
Services, 2001), and recent studies have confirmed
alarmingly high rates of adolescent violent behavior
(Farrington, 2004; Mendel, 2003). The frequency of
violent crime increases during the adolescent years,
with 27% of all serious violent victimizations in the
United States committed by youth under the age of 18
(Federal Bureau of Investigation, 1999; Office of Juvenile Justice and Delinquency Prevention, 1999). Just as
alarming as the high rate of adolescent violence among
community samples are the clinical consequences of
violent behavior on youth adjustment outcomes. Research among nonreferred samples finds that violent
behavior in adolescence places youth at greater risk for
continued high levels of violent behavior and other
externalizing problems in late adolescence and early

Understanding Protective Processes


Despite such well-established effects, many youth,
regardless of risk exposure, do not become violent
(Luthar, 1991; Masten, 1999). Understanding protective processes contributing to positive youth outcomes
and reduced violent behavior can be just as vital as understanding the compromised functioning that results
from violence (Sroufe & Rutter, 1984). In understanding protection, one must consider not only adolescent individual characteristics but also how the environment affords youth opportunities to succeed
(Masten & Coatsworth, 1998). Youth are embedded
within multiple social contexts simultaneously (Bronfenbrenner, 1979), and an ecological approach considers how these contexts work together to influence
youth adjustment.
In this study, we investigated how the intersecting
social ecologies of family and school interacted with
one another and in turn how they were related to youth
violent behavior. Specifically, we investigated the
paths by which adolescents feelings of school and parent connectedness, plus differences among schools in
climate and parental involvement, may work independently and in concert to lessen the risk of adolescents en-

This research was based on data from the National Longitudinal


Study of Adolescent Health (Add Health), a program project designed by J. Richard Udry (principal investigator) and Peter Bearman and funded by grant P01HD31921 from the National Institute
of Child Health and Human Development to the Carolina Population
Center, University of North Carolina at Chapel Hill, with cooperative funding participation by the National Cancer Institute; the National Institute of Alcohol Abuse and Alcoholism; the National Institute on Deafness and Other Communication Disorders; the National
Institute of Drug Abuse; the National Institute of General Medical
Sciences; the National Institute of Mental Health; the National Institute of Nursing Research; the Office of AIDS Research, National
Institute of Health; the Office of Research on Womens Health, National Institute of Health; the Office of Population Affairs, Department of Health and Human Services; the National Center for
Health Statistics, Centers for Disease Control and Prevention, Department of Health and Human Services; the Office of the Assistant
Secretary for Planning and Evaluation, Department of Health and
Human Services; and the National Science Foundation.
Correspondence should be addressed to Kathryn Brookmeyer,
Georgia State University, Department of Psychology, P.O. Box 5010,
Atlanta, GA 30302-5010. E-mail: psykxbx@langate.gsu.edu

504

SCHOOLS, PARENTS, AND VIOLENCE

gaging in violent behavior and mitigate the link between violence exposure and violent behavior.

Downloaded by [Assam University, Silchar] at 00:30 27 April 2015

School Connectedness and Climate


In the United States, school-age children and adolescents spend nearly half of their waking hours in
school, and schools are widely considered to be a critical setting for developing youth (Smith, Boutte, Zigler,
& Finn-Stevenson, 2004; U.S. Department of Health
and Human Services, 2001). Among community samples of youth, school climate is an aspect of the school
setting found to be crucial in relation to student aggression and behavior (Haynes, Comer, & Hamilton-Lee,
1989; Haynes, Emmons, & Ben-Avie, 1997). In this
study, we conceptualized school climate in two distinct
ways: (a) as individual students feelings of connectedness to their school and (b) as school-level climate,
defined as differences among schools in composite
school climate ratings. In studies of the protective effects of school connectedness on adolescent adjustment outcomes among nonreferred samples, the importance of taking student perceptions into account has
been recognized (Kuperminc, Leadbeater, Emmons, &
Blatt, 1997; Resnick et al., 1997). Specifically, when
students feel connected to their school, they may be
more likely to confide in teachers about the experience
of violence exposure, which may lead to better coping
skills and decreased violent behavior. Yet, the role of
school connectedness as a buffer against the effects of
violence among community samples remains unclear
(Ozer, 2005; Ozer & Weinstein, 2004).
Focusing on school climate from a school-level perspective allows school climate to be viewed as a school
characteristic, which has also shown to be integral in
understanding school effects on adolescent student behavior among community samples (Khoury-Kassabri,
Benbenishty, Astor, & Zeira, 2004). When viewed from
the school level, the proportion of students at school
who feel connected to school may be associated with
more effective learning environments within the classroom and linked to reduced disorder and fighting behavior. Our dual conceptualization of school climate
allows setting- and individual-level effects to be teased
apart (Raudenbush, Rowan, & Kang, 1991) and serves
to investigate the unique and interactive effects of both
students and schools on violent behavior.

Parent Connectedness and National


Parent Teacher Association (PTA)
Involvement
A critical feature of the ecological approach is its
emphasis on linking youth and their social ecologies
to foster successful adolescent adjustment outcomes

(Smith et al., 2004). The role of parents in particular is


an ecological factor widely cited as having the potential to promote protection for youth (Garmezy, 1991;
U.S. Department of Health and Human Services,
2001). Specifically, parentchild relationships characterized by parent involvement and connectedness have
been associated with reduced externalizing behavior,
including violence (Blum, Ireland, & Blum, 2003;
Resnick et al., 1997). Recent community-based studies
examining the buffering effect of parents have illustrated that supportive parenting may attenuate the link
between violence exposure and violent behavior over
time for adolescents (Brookmeyer et al., 2005; Gorman-Smith, Henry, & Tolan, 2004).
Parent involvement within the school, such as attending school programs, has been linked to a host
of beneficial student outcomes, including increased
academic achievement and improved youth behavior
(Haynes et al., 1989). Further, the onus of responsibility for parents school involvement is shared jointly by
the parent and the school in promoting and facilitating
involvement (Epstein, 1991; Hoover-Dempsey, Bassler, & Brissie, 1987). We recognized this shared responsibility for facilitating parent involvement by operationalizing it as a school-level variable, examining
variability in percentage of parent participation in parentteacher alliances such as the PTA. PTA participation is also examined as a school-level construct because parent involvement in schools has been found to
exert a greater effect on student behavior when the adolescent attends a school in which a large proportion of
other students parents are involved as well (Darling &
Steinberg, 1997; Pong, 1998).

This Study
This study utilized an ecological approach by taking
into account the joint contribution of parents, schools,
and adolescent characteristics on changes in violent
behavior over time (Bronfenbrenner & Morris, 1998).
We examined the main and interactive effects of parent
and school connectedness on changes in violent behavior over time and as buffers of violence exposure on violent behavior. In doing so, we took a comprehensive
view of school climate and used hierarchical linear
modeling to tease apart the effects of school-level climate and school characteristics from student perceptions of school connectedness and parent connectedness on violent behavior over time. Such a multilevel
design focuses on both students and schools as the
units of analysis (Raudenbush & Bryk, 2002).
Data for this study came from the National Longitudinal Study of Adolescent Health (Add Health). Prior
research has used the Add Health data to examine the
roles of parent and school connectedness in protecting
youth from exposure to and use of weapons in acts of
505

Downloaded by [Assam University, Silchar] at 00:30 27 April 2015

BROOKMEYER, FANTI, HENRICH

violence over time (Henrich et al., 2005). Henrich and


colleagues found that youth who were more connected
to their parents and school were less likely to be exposed to or commit weapon violence over time, but
they did not find that feelings of connectedness buffered youth from the effects of violence. Building
on Henrich et al. and other extant research (e.g.,
Resnick, Ireland, & Borowsky, 2004), this study makes
unique contributions by operationalizing violence exposure and violent behavior broadly (e.g., fighting,
getting jumped) and through its multilevel examination of school-level characteristics, climate, and parent
involvement.
We hypothesized that adolescents would engage in
less violent behavior over time if they perceived high
levels of parent and school connectedness and that
youth who have been exposed to violence would engage in more violent behavior over time. Further, we
examined the additive and interactive effects of schooland individual-level protective factors on violent behavior and as buffers of the effects of violence exposure on violent behavior. Finally, we hypothesized
cross-level interactions among individual- and schoollevel protective factors as working together to mitigate
the effects of violent behavior.
Additional school characteristics have been linked
to violence outcomes in nonreferred student samples
(Hunt et al., 2002; Khaoury-Kassabri et al., 2004), and
these factors were covaried. Specifically, we controlled
for class and school size (Olweus, 1993; Walker &
Gresham, 1997), dropout rates (Hunt et al., 2002),
school type (private or public), school urbanicity (Anderman, 2002), and school level (middle or high
school).
Last, research consistently finds that boys are more
involved than girls in violence (Fitzpatrick & Boldizar,
1993; Office of Juvenile Justice and Delinquency Prevention, 1999) and that the experience of violence and
associated processes of protection may work differently by gender (Brookmeyer et al., 2005; GormanSmith et al., 2004). This study builds on these findings
to examine potential gender differences in the protective processes of parent and school connectedness,
PTA involvement, and school climate.

Method
Participants
Data for this study were taken from the first two
waves of the Add Health restricted-use contractual dataset. Add Health was designed with the understanding that families, friends, schools, and communities play important roles in the lives of adolescents.
To facilitate such analyses, a randomized clustered
sampling design based around 144 middle and high
506

schools was used to recruit from a national sample of


American schools (Bearman, Jones, & Udry, 1997).
Add Health includes a core sample of 12,105 students
in Grades 7 through 12 at the onset of the study. Students through Grade 11 were followed up 1 year later.
For the in-school assessment, passive parental consent
was used in addition to adolescent assent, unless
schools mandated active parental consent. For the inhome interview, active informed consent was obtained
from the parent or legal guardian. Participation rates
were 78.9% in Wave 1 and 88.2% in Wave 2 (Harris et
al., 2003).
The sample for this study included 6,397 students
from 125 schools who were still enrolled in the school
from which they were recruited (96%) and who had
complete data for both time points (11% attrition). The
sample was split roughly equally by gender (54%
girls). The mean age was 15.54 (SD = 1.56). Ten percent of the sample reported being of Hispanic origin,
71% White, 20% African American, 6% Native American, and 5% Asian. Due to the range of ages, age was
controlled in the analyses. Ethnicity was not controlled
because we wished to examine our research questions
across the full range of diversity of the sample.
Measures and Procedures
In home interview. Between April and December 1995 (Wave 1), adolescents were administered an
extensive survey during in-home interviews, and their
parents completed a questionnaire. One year later,
in 1996 (Wave 2), adolescents were administered the
same in-home survey a second time. In the interview,
youth were asked about their relationships with family
and their feelings about school. For questions about
sensitive, illegal behaviors, youth directly entered their
responses into interviewers laptops, with the screen
facing away from the interviewer (Blum et al., 2003).
Items from the Add Health interview section on
fighting and violence were used to form violence exposure1 and violent behavior constructs, both of which
were adapted from Resnick et al. (2004). For the violence exposure items assessed in Wave 1, adolescents
were asked five questions, which included whether
during the past 12 months someone had pulled a knife
on them (10%), they were jumped (9%), they saw
someone shoot or stab another person (7%), someone
shot them (1%), or someone stabbed them (5%) one or
more times. In Wave 1, 21% of the sample reported be1An exploratory factor analysis indicated that the witnessing violence item loaded on the same factor as the victimization items. The
goodness of fit of a one-factor construct of violence exposure was
verified via confirmatory factor analysis, 2(5) = 40.42, root mean
square error of approximation = .03, Nonnormed Fit Index = .99,
comparative fit index = .99. (Contact the authors for more details on
the fit statistics.) Therefore, violence exposure was examined as a total construct of five items in our analyses.

Downloaded by [Assam University, Silchar] at 00:30 27 April 2015

SCHOOLS, PARENTS, AND VIOLENCE

ing exposed to one or more forms of violence. Due to


low exposure rates, the construct was analyzed as a binary variable in the analyses.
For the violent behavior items, assessed in both
Waves 1 and 2, youth were asked eight questions, including whether during the past 12 months they used or
threatened another person with a weapon (4% in Wave
1; 3% in Wave 2), took part in a group fight (19% in
Wave 1; 17% in Wave 2), got into a physical fight (31%
in Wave 1; 19% in Wave 2), pulled a knife or gun on
someone (4% in Wave 1; 4% in Wave 2), shot or
stabbed someone (1% in Wave 1; 1% in Wave 2), used
a weapon in a fight (5% in Wave 1; 3% in Wave 2), sustained a serious injury from a fight (17% in Wave 1; 8%
in Wave 2), or carried a weapon to school (5% in Wave
1; 3% in Wave 2) one or more times. Due to the low
prevalence rates of violent behavior, each item was
scored 0 (none) or 1 (one or more times) and summed,
with total scores ranging from zero to a maximum of
eight behaviors. In Wave 1, 44.3% of youth reported
engaging in at least one type of violent behavior, and in
Wave 2, 30.1% reported engaging in at least one type
of violent behavior, with scores ranging to a maximum
of all eight behaviors in both waves. Cronbachs alpha
was .71 in Wave 1 and .75 in Wave 2.
The in-home interview included a 12-item parent
connectedness scale assessed in Wave 1 (Resnick et al.,
1997) for which youth reported on their perceptions on
the quality of their relationships with their parents and
the extent to which parents support them. Items for
the scale included youths average perceptions of connectedness to each parent and to both parents together.
Sample items included, Most of the time your mother
is warm and loving toward you and How much do
you feel your parents care about you? Youth responded to items on 5-point scales from 1 (strongly
disagree) to 5 (strongly agree) and from 1 (not at all) to
5 (very much). Items were averaged into an internally
consistent scale, and Cronbachs alpha was equal to
.87.
The Protective Factors and Academics sections included seven items on adolescents feelings of connectedness to their school, assessed in Wave 1 (Resnick et al., 1997). Sample school connectedness items
included How much do you feel your teachers care
about you? and How close do you feel to people at
your school? Adolescents responded to items on 5point scales ranging from 1 (not at all) to 5 (very much)
and 1 (strongly disagree) to 5 (strongly agree). Items
formed an internally consistent scale, and Cronbachs
alpha was equal to .77.
In-school survey. A 45-min long in-school survey was administered between September 1994 and
April 1995 to a sample of 90,118 students in all of
the schools from which the in-home participants were
sampled. The median number of students completing

the survey per school was 470.5, with a range of 25 to


2,325 per school. The Protective Factors and Academics sections of the in-school survey included seven
items assessing school climate and sample items, including Teachers treat students fairly and I feel safe
at this school. Adolescents responded to items on
5-point scales from ranging from 1 (not at all) to 5
(very much) and 1 (strongly disagree) to 5 (strongly
agree). Items formed an internally consistent scale,
and Cronbachs alpha was equal to .73. These items
were parallel to the in-home school connectedness
scale with the exception of two items that were asked in
the in-school survey but not in the in-home assessment
and vice versa. Students mean scores were averaged
with other students mean scores from the same school
to create an average school climate rating for each
school in the sample. Although this scale was similar to
the in-home school connectedness scale, it differed in
that most of the respondents were not from the core
sample, so there was minimal redundancy in variance
between in-home school connectedness, which measured individual differences in feelings about school,
and in-school school climate, which measured climate
as an aggregated school characteristic.
School characteristics. Five school characteristics were assessed independently from the school administrator questionnaire code book. Average daily
attendance, average school size, average class size,
percentage dropout of students, and percentage of parents in parentschool organizations were assessed (see
Table 1). Also, the type of school (public or private);
whether the school was a middle school, high school,
or comprehensive (K12) school; and whether the
school was located within a rural or a suburban location were assessed through dummy-coded variables. In
our sample of schools in Wave 1, 84.9% had students
that attended 90% or more of the time, 45.3% were categorized as medium (4011,000 students), 90.1% were
public, 54.7% were suburban, and 45.9% were high
schools.

Results
Descriptive Statistics
Means and standard deviations of the variables investigated are shown in Table 1.
Table 2 shows the correlations among all the Level 1
variables. Gender was significantly correlated with violence exposure, parent connectedness, and violent behavior at both waves. Boys were more likely to be exposed to violence, committed more types of violence,
and felt more connected to their parents than girls. Further, older adolescents felt less connected to their
schools and were exposed to higher levels of violence.
507

BROOKMEYER, FANTI, HENRICH

Violent behavior during Wave 1 was positively related


to violent behavior during Wave 2 and violence exposure. Violent behavior during Wave 2 was positively
related to violence exposure. In addition, adolescents
who committed more types of violence during Wave 1
and Wave 2 felt less connected to their parents and their
schools. Finally, parent connectedness was positively
related to school connectedness.

Table 3 shows the correlations among all the Level 2


variables. School climate was significantly positively
related to attendance rates and significantly negatively
related to dropout rates, attending high school, and
class and school size, such that greater numbers of students in classes and schools were associated with more
negative school climate. Also, school climate was significantly positively related to PTA and school type,
such that attending a private school was associated
with a more positive school climate.

Downloaded by [Assam University, Silchar] at 00:30 27 April 2015

Table 1. Means and Standard Deviations

Level 2a
School climate Wave 1
Average class size Wave 1
Percent student drop-out Wave 1
Percent parents in parent-school
organizations Wave 1
Level 1b
Parent connectedness Wave 1
School connectedness Wave 1
Violent behavior Wave 1
Violent behavior Wave 2
Violence exposure Wave 1

SD

2.05
25.35
2.34
27.15

0.18
5.41
3.38
22.77

4.28
3.15
0.86
0.58
0.21

0.58
0.65
1.34
1.18
0.41

Hierarchical Linear Modeling Analysis


Multilevel modeling with HLM 5 (Raudenbush,
Bryk, Cheong, & Congdon, 2000) was used to investigate the unique and interactive effects of violence exposure, parenting variables, and school variables on violent behavior over time. The analyses employed a
two-level hierarchical linear model with parent and
school connectedness on the first level nested within
schools and estimates of school characteristics and
school climate on the second level.
Attrition. The final sample represented 11% attrition from Wave 1 to Wave 2. Prior to the main analyses, a two-level logistic regression was conducted on
the Wave 1 data to determine the extent to which attrition was selective across the key variables. A binary

Note: Ranges of possible scores for each study variable are as follows: school climate (15), parent connectedness (15), school connectedness (15), violent behavior (08), and violence exposure
(01).
an = 125. bn = 6,397.

Table 2. Correlations Between Level-1 Variables

Age Wave 1
Violent behavior Wave 1
Violent behavior Wave 2
Violence exposure Wave 1
Parent connectedness Wave 1
School connectedness Wave 1

Gender

Age

Violent Behavior
Wave 1

Violent Behavior
Wave 2

Violence
Exposure

Parent
Connectedness

.05**
.22**
.18**
.17**
.08**
.02

.00
.01
.06**
.17**
.06**

.54**
.50**
.16**
.28**

.36**
.14**
.21**

.12**
.22**

.36**

Note: N = 6,397.
*p .05. **p .01.

Table 3. Correlations Between Level-2 Variables

Class Size
PTA
School size
School type
Dropout
Suburban
Rural
Middle school
High school
School climate

Attendance

Class Size

PTA

School
Size

School
Type

Dropout

Suburban

Rural

Middle
School

High
School

.23**
.33**
.21**
.16*
.43**
.17*
.07
.17*
.20**
.40**

.28**
.32**
.25**
.24**
.09
.20**
.26**
.07
.35**

.20**
.33**
.31**
.11
.02
.09
.28**
.26**

.26**
.41**
.12
.22**
.41**
.60**
.37**

.22**
.13
.07
.12
.11
.34**

.01
.08
.38**
.51**
.36**

.42**
.02
.21**
.01

.07
.18**
.06

.68**
.02

.27**

Note: Wave 1, N = 125.


*p .05. **p .01.

508

SCHOOLS, PARENTS, AND VIOLENCE

Downloaded by [Assam University, Silchar] at 00:30 27 April 2015

variable indicating whether students dropped out of the


study by Wave 2 was regressed on the studys key
Level 1 and Level 2 variables. According to the results,
boys were less likely to drop out, B = .15, SE = .08, p =
.05, and older students were more likely to drop out, B
= .11, SE = .03, p < .001. Attrition was not selective
across any of the other Level 1 variables, although students who attended schools with higher PTA participation (Level 2) were slightly more likely to drop out, B =
0.002, SE = .001, p = .03.
Level 1. On the first level of the model, Wave 2
violent behavior was regressed on the following Wave
1 variables: gender, age, and violent behavior as controls; violence exposure, parent connectedness, and
school connectedness; and interactions of gender with
violence exposure and with parent connectedness and
school connectedness to test for gender interactions;
interactions of violence exposure with school connectedness and parent connectedness to test for buffering effects; three-way interactions of Gender Violence Exposure each of the connectedness variables
to test for gender differences in buffering effects; and a
three-way School Connectedness Parent Connectedness Violence Exposure interaction. All continuous independent variables were centered to facilitate
ease of interpretation in the presence of interaction
terms, and the product interaction terms were computed from centered variables to reduce mutlicollinearity (Jaccard & Turrisi, 2003). The Level 1 variables accounted for 43% of the variance in Wave 2
violent behavior.
Level 2 random effects (analysis of variance).
When just random effects were estimated on the second level, 16.3% of the remaining variance in violent
behavior was among schools, controlling for Level 1
effects, p < .01. The estimation of variance components indicated that several slopes did not vary across
schools (Age, Gender Climate, and Parent Connectedness School Connectedness), p > .50. A more parsimonious model in which these variables random effects were not estimated fit the data equally well,
2(45) = 23.21, p > .50.
Level 2 fixed effects. In the fixed-effects model,
Wave 2 violent behavior was regressed on school attendance, class size, school size, school type, dropout rate,
urbanicity, middle versus high school, parent involvement, school climate, and the interaction between parent involvement and school climate, controlling for the
Level 1 model, with its variables group-centered.
Level 2 slopes as outcomes. To examine whether school-level variables interacted with Level 1 variables to reduce violent behavior over time and buffer
adolescents from the effects of violence exposure on

violent behavior, the Level 1 slopes of violence exposure, school connectedness, parent connectedness,
and the two-way buffering interactions (except for Parent Connectedness School Connectedness, which
did not vary across school) were regressed on Level 2
parent involvement and school climate. None of the
slopes-as-outcomes effects was significant. As a result,
the effects of the Level 2 fixed-effects model were
interpreted.
These results are presented in Table 4. The intercept
presented in Table 4 indicates that average number of
types of violent behavior during Wave 2 equaled 0.59
(SE = .09). Table 4 also presents the Level 1 main effects and the interactions among the Level 1 variables,
as well as the Level 2 effects.
Level 1 effects. Gender was uniquely negatively
related to violent behavior during Wave 2, suggesting
that boys committed more types of violence over time.
Older adolescents committed fewer types of violence
over time. Violent behavior and violence exposure during Wave 1 were positively related to violent behavior
Table 4. HLM Analysis Predicting Violent Behavior
at Wave 2
Wave 1 Variables

SE

Intercept
Level 1
Gender
Age
Violent behavior
Violence exposure
Parent connectedness
School connectedness
Exposure School Connectedness
Exposure Parent Connectedness
Exposure Gender
Parent Connectedness Gender
School Connectedness Gender
School Connectedness Parent
Connectedness
Exposure Parent Connectedness
Gender
Exposure School Connectedness
Gender
Exposure School Connectedness
Parent Connectedness
Level 2
Attendance
Class size
School size
School type
Dropout
Suburban
Rural
Middle school
High school
PTA
Climate
PTA Climate

.59

.09

.00

.13
.03
.37
.49
.03
.17
.04
.30
.16
.01
.07
.00

.03
.01
.02
.13
.10
.08
.18
.27
.08
.06
.04
.03

.00
.00
.00
.00
.75
.03
.83
.27
.05
.94
.12
.90

.15

.17

.62

.00

.12

.99

.23

.12

.05

.03
.00
.08
.08
.01
.02
.04
.09
.04
.00
.57
.00

.02
.00
.03
.05
.00
.04
.05
.05
.05
.00
.10
.00

.13
.29
.01
.13
.11
.66
.43
.07
.36
.14
.00
.34

Note: N = 6,397.

509

Downloaded by [Assam University, Silchar] at 00:30 27 April 2015

BROOKMEYER, FANTI, HENRICH

during Wave 2, and there was a significant Gender


Exposure interaction, indicating that being exposed to
violence was more strongly associated with violent behavior over time for girls. Further, adolescents who reported higher school connectedness committed fewer
types of violence over time. Because there was also a
significant three-way School Connectedness Parent
Connectedness Violence Exposure interaction, the
effect of school connectedness in Table 4 applies to adolescents with average levels of parent connectedness
who had not been exposed to violence.
The three-way School Connectedness Parent Connectedness Violence Exposure interaction uniquely
explained 2% of the variance in Wave 2 violent behavior, and it is probed in Figure 1. When school connectedness was low (1 SD below the mean), the Exposure Parent Connectedness interaction was not
significant (B = .03, SE = .13, p = .84). However,

when school connectedness was high (1 SD above the


mean), the Exposure Parent Connectedness interaction was significant (B = .34, SE = .10, p = .001). According to Figure 1, parent connectedness buffered adolescents from the effects of violence exposure on
violent behavior when school connectedness was also
high. As shown in Table 4, the Exposure Parent
Connectedness interaction was also significant at average levels of school connectedness, although it was not
as strong as at high levels.
Level-2 effects. School size was negatively related to violent behavior, suggesting that students
from smaller schools commit more types of violence.
School climate was negatively related to violent behavior but PTA participation was not. The interaction between PTA and school climate was not significant. Although statistically significant, the school size and

Figure 1. Level 1 interaction of Wave 1 levels of violence exposure and parent connectedness predicting Wave 2 violent behavior when
school connectedness is at high and low levels.

510

SCHOOLS, PARENTS, AND VIOLENCE

school climate effects were small, accounting for less


than 1% of the variance in violent behavior.
Residual among-school variance. Table 5 shows
the residual variability among schools in the Level 1
slopes after taking into account the Level 2 variables.
According to Table 5, there still remained significant variability in number of types of violent behavior
among schools. There also remained significant variability in the majority of the slopes of violence exposure and most of the interactions under investigation.
These findings suggest that there were additional
school-level factors associated with violent behavior
not accounted for in these analyses.

School and Parent Connectedness:


Joint Contributions

Downloaded by [Assam University, Silchar] at 00:30 27 April 2015

Discussion
Within the framework of ecological theory (Bronfenbrenner, 1979), the findings of this study begin to
elucidate the potential roles of parents and schools in
protecting youth from violence. Using a multilevel perspective, we found that when students felt connected to
their schools, they reported engaging in fewer types of
violent behavior over time, an effect consistent with
previous research (Ozer, 2005). Additionally, consistent with prior research, our results found that violence
exposure is a predictor of subsequent increases in violent behavior (Brookmeyer et al., 2005; Gorman-Smith
et al., 2004), and our findings also suggest that parent
and school connectedness can interact to buffer adolescents from these effects of violence exposure. The effects of violence exposure may be stronger for girls.
This finding could lend support for the premise that the
Table 5. Residual Between-School Variance
Wave 1 Variables
Intercept
Gender
Violent Behavior
Violence Exposure
Parent Connectedness
School Connectedness
Exposure School
Connectedness
Exposure Parent
Connectedness
Exposure Gender
Parent Connectedness Gender
Gender Exposure Parent
Connectedness
Gender Exposure School
Connectedness
School Connectedness
Exposure Parent
Connectedness

experience of violence may operate differently by gender (Brookmeyer et al., 2005; Gorman-Smith et al.,
2004).
This study suggests that school-level indexes of
school climate served as a protective factor for student
violent behavior. That is, adolescents who attended
schools with higher levels of aggregated school climate engaged in fewer types of violence over time,
though the effect of school-level climate was small
in magnitude. Although few previous studies have
viewed school climate as a school construct, those that
have also found main effects of school climate on violent behavior outcomes (Khoury-Kassabri et al., 2004).

Variance

df

0.15
0.02
0.03
0.87
0.31
0.16
1.45

57
69
69
69
69
69
69

160.74
90.58
203.85
206.62
86.41
68.98
139.34

.00
.04
.00
.00
.08
>.50
.00

3.46

69

154.20

.00

0.32
0.08
1.38

69
69
69

200.15
78.60
142.13

.00
.20
.00

0.56

69

134.39

.00

0.77

69

168.21

.00

Note: N = 6,397. The chi-square statistics are based on only 70 of the


125 schools with sufficient data for computation. Fixed effects and
variance components are based on all the data.

The interactive contribution of school and parent


connectedness emerged as a potentially protective process in buffering adolescents from the effects of violence exposure. Specifically, results indicate that when
students perceived high levels of connectedness to both
their schools and their parents, the effects of violence
exposure on violent behavior were mitigated. In this
way, our findings suggest that perceived school and
parent connectedness may work together to buffer the
effects of violence exposure on violent behavior. Conversely, when students lacked feelings of connectedness to their schools, parent connectedness did not exhibit a protective effect. Consistent with these findings,
recent research has found that even exceptionally positive family functioning was not able to mitigate the effects of pervasive and severe violence exposure (Gorman-Smith et al., 2004), presumably because there was
too much stress and risk within the youths environment to counter and that additional proximal resources
were needed (Gorman-Smith, Tolan, & Henry, 1999).
Based on our results, school connectedness appears to
be one such resource.
The findings of this study, which examined violent
behavior and fighting outcomes, stand in contrast to
prior Add Health research that found no buffering effects of school or parent connectedness from weapon
violence (Henrich et al., 2005). It may be that the potential protective effects of school and parent connectedness break down in the face of the most extreme
forms of violence. Further, it should be noted that violent behavior in adolescence is comorbid with a host of
other problem behaviors, including delinquency and
difficulties with impulse control (e.g., Jessor & Jessor,
1977; Lipsey & Derzon, 1998), which may make it difficult for these youth to forge positive connections to
parents and schools. Indeed, engaging in violent behavior is typically met with punitive actions by parents
and schools (e.g., Kupersmidt, Coie, & Howell, 2004).
This comorbidity was statistically controlled in the
analyses; in reality such conditions may create obsta511

BROOKMEYER, FANTI, HENRICH

cles to forming supportive relationships with parents


and perceiving strong ties to the school environment in
the face of violence exposure and violent behavior.

Downloaded by [Assam University, Silchar] at 00:30 27 April 2015

School-Level Effects
This study suggests that school climate may operate
above and beyond the individual level to promote reduced youth fighting and violent behavior. However,
aside from school climate, the second level of our
model revealed only smaller school size to be a predictor of student violent behavior, and the effects of each
were small. This result is inconsistent with previous research, which has found that additional school characteristics (e.g., Anderman, 2002; Hunt et al., 2002)
are linked to youth behavior. This relative lack of
school-level effects may be related to the fact that our
model examined school-level factors controlling for
individual-level factors, which were more proximal.
Key to the ecological model is the premise that although youth are embedded within multiple intersecting contexts, those environmental factors that are the
most proximal are considered to exert the greatest,
most pronounced effects (Bronfenbrenner, 1979). In
addition, Add Healths measure of PTA consists of percentages of parents involved in PTA on the school
level, in contrast to previous research, which measured
individuals rates of attendance at PTA meetings.
Additionally, results indicated that other schoollevel factors not included in the model accounted
for variability in students violent behavior over time.
Among these potentially important factors are constructs not measured by Add Healths administrator or
in-school surveys, such as school-level violent behaviors, discipline problems, and academic failure. Under
the No Child Left Behind Act (2001), data on schools
safety and their performance relative to state standards
are required to be collected and made publicly available. The inclusion of these data in future research may
further elucidate school-level effects on student behavior as well as the success of public policies aimed at attenuating negative school-level effects.
Implications
A critical feature of the ecological approach is its
emphasis on linking youth and families to promote
successful adolescent adjustment outcomes (Smith et
al., 2004), and this study suggests the importance of
understanding how multiple levels of the adolescents
environment (e.g., family and school contexts) can
work together to afford youth opportunities to succeed.
Indeed, our findings underscore the necessity of looking beyond the unique effects of individual social settings and understanding how those social settings interact with one another to influence youth outcomes
(Bronfenbrenner, 1979). Our findings also serve to em512

phasize the importance of considering the presence of


risk and protective factors as indicative of a complex
adjustment process (Cicchetti & Toth, 1997). Therefore, rather than depicting risk and protection as independent, static factors, there is a need to consider risk
and protection as occurring within an interactive, contextually dependent framework (Masten, 1999).
Finally, our findings suggest implications for interventions to foster adolescent connectedness to both
schools and parents and highlight the potential for psychologists and mental health experts to make community-scale impacts on adolescent well-being through
consultation with school personnel on ways to improve
schools climates (Haynes et al., 1989; Schwab-Stone,
Henrich, & Armbruster, 2002). Promoting partnerships
between parents and schools may also have implications for increased connectedness to both social ecologies. Haynes and colleagues suggested that one of the
most effective ways to improve students perception of
school climate is to make parents an integral element
within the school ecology.
Strengths, Limitations,
and Future Directions
Strengths of this investigation include a nationally
representative and large number of middle and high
schools within a longitudinal design controlling for
students previous violent behavior. Additionally, this
study is unique in offering an ecological and multilevel
approach to investigate the protective processes of
schools and parents on youth violence. A key ecological strength of the study is that it demonstrates protective effects of schools that can generalize beyond
building walls in that the measures of violence exposure and violent behavior do not pertain just to school
violence. However, this strength is accompanied by
limitations inherent in Add Healths measures of violence. Response categories were not sensitive to the
frequency of violent behaviors, limiting analyses to
measure variability in types of violent behavior. The
exposure construct included only a single witnessing
item, so that the experience of witnessing and victimization by violence were assessed as a combined risk
for youth, therefore limiting the generalizability of the
violence exposure construct. Also, because the exposure variable is limited to the most extreme forms of
violence, the prevalence rates reported here may be
lower than in other studies in which a wider array of violence severity was assessed. Supplementary questions providing additional context about the nature of
the violence reported would provide more robust information about the meaning of adolescents experiences
with violence.
Overall, this study represents progress in understanding the link between processes of protection and
youth violence from an ecological and multilevel per-

Downloaded by [Assam University, Silchar] at 00:30 27 April 2015

SCHOOLS, PARENTS, AND VIOLENCE

spective, and it does so in a sample that is representative of the diversity among American schools. Future
avenues of research, in addition to addressing the limitations previously mentioned, could include further
exploration of how cultural and sociodemographic differences between adolescents may relate to distinct
protective processes for youth violent behavior as well
as how additional individual-level constructs, such as
deviant peers, can affect violent behavior outcomes.
In addition, the examination of mediational models,
which have the ability to focus on the process by which
school characteristics affect students behavioral outcomes, would increase understanding of how school
climate exerts its effects. Incorporating additional
social ecologies, such as neighborhood efficacy and
neighborhood danger, would have important implications for adolescents processes of risk and protection.
Finally, greater exploration of how school-level change
can influence adolescent change could offer researchers the possibility of avenues to prevent violence and
support adolescent adjustment through promoting feelings of connectedness.

References
Anderman, E. (2002). School effects on psychological outcomes
during adolescence. Journal of Educational Psychology, 94,
795809.
Bearman P. S, Jones J., & Udry J. R. (1997). National Longitudinal
Study of Adolescent Health: Research design. Retrieved May
15, 2002, from http://www.cpc.unc.edu/addhealth
Blum, J., Ireland, M., & Blum, R. W. (2003). Gender differences in
juvenile violence: A report from Add Health. Journal of Adolescent Health, 32, 234240.
Bronfenbrenner, U. (1979). The ecology of human development: Experiments by nature and design. Cambridge, MA: Harvard University Press.
Bronfenbrenner, U., & Morris, P. (1998). The ecology of developmental processes. In W. Damon & R. Lerner (Eds.), Handbook
of child psychology (5th ed., pp. 9931027). New York: Wiley.
Brookmeyer, K. A., Henrich, C. C., & Schwab-Stone, M. (2005).
Adolescents who witness community violence: Can parent support and prosocial cognitions protect them from committing violence? Child Development, 76, 917929.
Cicchetti, D., & Toth, S. L. (1997). Transactional ecological systems
in developmental psychopathology. In S. S. Luthar, J. A.
Burack, D. Cicchetti, & J. R. Weisz (Eds.), Developmental
psychopathology: Perspectives on adjustment, risk, and disorder (pp. 317349). New York: Cambridge University Press.
Darling, N., & Steinberg, L. (1997). Community influences on adolescent achievement and deviance. In J. Brooks-Gunn, G. Duncan, & L. Aber (Eds.), Neighborhood poverty: Contexts and
consequences for children. Volume 2: Conceptual, methodological, and policy approaches to studying neighborhoods (pp.
120131). New York: Russell Sage Foundation.
Epstein, J. L. (1991). Effects on student achievement of teachers
practices of parent involvement. Advances in Reading/Language Research, 5, 261276.
Farrell, A. D., & Bruce, S. E. (1997). Impact of exposure to community violence on violent behavior and emotional distress among
urban adolescents. American Journal of Preventative Medicine,
12, 1321.

Farrington, D. (2004). Conduct disorder, aggression, and delinquency. In R. Lerner & S. Steinberg (Eds.), Handbook of adolescent psychology (pp. 627664). New York: Wiley.
Federal Bureau of Investigation. (1999). Uniform crime reports for
the United States. Washington DC: U.S. Government Printing
Office.
Fitzpatrick, K. M., & Boldizar, J. P. (1993). The prevalence and consequences of exposure to violence among African-American
youth. Journal of the American Academy of Child & Adolescent
Psychiatry, 32, 424430.
Garmezy, N. (1991). Resilience and vulnerability to adverse developmental outcomes associated with poverty. American Behavioral Scientist, 34, 416430.
Gorman-Smith, D., Henry, D. B., & Tolan, P. H. (2004). Exposure to
community violence and violence perpetration: The protective
effects of family functioning. Journal of Clinical Child and Adolescent Psychology, 33, 439449.
Gorman-Smith, D., & Tolan, P. H. (1998). The role of exposure
to community violence and developmental problems among
inner-city youth. Development and Psychopathology, 10,
101116.
Gorman-Smith, D., Tolan, P. H., & Henry, D. B. (1999). The relation
of community and family to risk among urban-poor adolescents. In P. Cohen, C. Slomkowski, & L. Robbins (Eds.), Historical and geographical influences on psychopathology (pp.
349367). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Harris, K. M., Florey, F., Tabor, J., Bearman P. S., Jones J., & Udry,
R. J. (2003). The National Longitudinal Study of Adolescent
Health: Research design. Retrieved January 3, 2006, from
http://www.cpc.unc.edu/projects/addhealth/design
Haynes, N. M., Comer, J. P., & Hamilton-Lee, M. (1989). School climate enhancement through parental involvement. Journal of
School Psychology, 27, 8790.
Haynes, N. M., Emmons, C., & Ben-Avie, M. (1997). School climate
as a factoring student adjustment and achievement. Journal of
Educational and Psychological Consultation, 8, 321329.
Henrich, C. C., Brookmeyer, K. A., Shahar, G., & Kidd, S. (2005).
Weapon violence in adolescence: Parent and school connectedness as protective factors. Journal of Adolescent Health, 37,
306312.
Hoover-Dempsey, K. V., Bassler, O. C., & Brissie, J. S. (1987). Parent involvement: Contributions of teacher efficacy, school socioeconomic status, and other school characteristics. American
Educational Research Journal, 24, 417435.
Hunt, M. H., Meyers, J., Davies, G., Meyers, B., Rogers Grogg, K.,
& Neel, J. (2002). A comprehensive needs assessment to facilitate prevention of school drop out and violence. Psychology in
the Schools, 39, 399416.
Jaccard, J., & Turrisi, R. (2003). Interaction effects in multiple regression (2nd ed.). Thousand Oaks, CA: Sage.
Jessor, R., & Jessor, S. (1977). Problem behavior and psychosocial
development: A longitudinal study of youth. New York: Academic.
Khoury-Kassabri, M., Benbenishty, R., Astor, R. A., & Zeira, A.
(2004). The contributions of community, family, and school
variables to student victimization. American Journal of Community Psychology, 34, 187204.
Kuperminc, G. P., Leadbeater, B. J., Emmons, C., & Blatt, S. (1997).
Perceived school climate and difficulties in the social adjustment of middle school students. Applied Developmental Science, 1, 7688.
Kupersmidt, J. B., Coie, J. D., & Howell, J. C. (2004). Resilience
in children exposed to negative peer influences. In K. Maton
(Ed.), Investing in children, youth, families and communities (pp. 251268). Washington DC: American Psychological
Association.
Lipsey M. W., & Derzon, J. H. (1998). Predictors of violent or serious delinquency in adolescence and early adulthood. In R.

513

Downloaded by [Assam University, Silchar] at 00:30 27 April 2015

BROOKMEYER, FANTI, HENRICH


Loeber & D. P. Farrington (Eds.), Serious and violent juvenile
offenders: Risk factors and successful interventions (pp.
86105). Thousand Oaks, CA: Sage.
Luthar, S. S. (1991). Vulnerability and resilience: A study of
high-risk adolescents. Child Development, 62, 600616.
Masten, A. S. (1999). Resilience comes of age: Reflections on the
past and outlook for the next generation of research. In M. D.
Glantz & J. L. Johnson (Eds.), Resilience and development:
Positive life adaptations (pp. 281296). Dordrecht, The Netherlands: Kluwer Academic.
Masten, A. S., & Coatsworth, J. D. (1998). The development of competence in favorable and unfavorable environments: Lessons
from research on successful children. American Psychologist,
53, 205220.
Mendel, D. (2003). And the walls keep tumbling down. AdvoCasey,
5, 1827.
Miller, L. S., Wasserman, G. A., Neugebauer, R., Gorman-Smith, D.,
& Kamboukos, D. (1999). Witnessed community violence and
antisocial behavior in high-risk, urban boys. Journal of Clinical
Child Psychology, 28, 211.
No Child Left Behind Act of 2001, Pub. L. No. 107110,
11111112, 115 Stat. 1425 (2002).
Office of Juvenile Justice and Delinquency Prevention. (1999). Juvenile offenders and victims: 1999 national report. Retrieved June
1, 2005, from http://ncjrs.org/html/ojjdp/nationalreport99/
index.html
Olweus, D. (1993). Bullying at school: What we know and what we
can do. Maiden, MA: Blackwell.
Ozer, E. J. (2005). The impact of violence on urban adolescents:
Longitudinal effects of perceived school connectedness and
family support. Journal of Adolescent Research, 20, 167192.
Ozer, E. J., & Weinstein, R. S. (2004). Urban adolescents exposure
to community violence: The role of support, school safety, and
social constraints in a school-based sample of boys and girls.
Journal of Clinical Child and Adolescent Psychology, 33,
463476.
Pong, S. (1998). The school compositional effect of single parenthood on 10th grade achievement. Sociology of Education, 71,
2342.

514

Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods. Newbury Park,
CA: Sage.
Raudenbush, S. W., & Bryk, A. S., Cheong, Y., & Congdon, R. T.
(2000). HLM 5: Hierarchical linear and nonlinear modeling.
Chicago: Scientific Software International.
Raudenbush, S. W., Rowan, B., & Kang, S. J. (1991). A multivariate
model for studying school climate with estimation via the EM
algorithm and application to U.S. high-school data. Journal of
Education Studies, 16, 295330.
Resnick, M. D., Bearman, P. S., Blum, R. W., Bauman, K. E., Harris,
K. M., Jones, J., et al. (1997). Protecting adolescents from
harm: Findings from the National Longitudinal Study on Adolescent Health. Journal of the American Medical Association,
287, 823832.
Resnick, M. D., Ireland, M., & Borowsky, I. (2004). Youth violence:
What protects? What predicts? Findings from the National
Longitudinal Study of Adolescent Health. Journal of Adolescent Health, 35, 424434.
Schwab-Stone, M. E., Henrich, C., & Armbruster, P. (2002). School
consultation. In M. Lewis (Ed.), Child and adolescent psychiatry: A comprehensive textbook (3rd ed., pp. 13611370). Baltimore: Lippincott, Williams & Wilkins.
Smith, E. P., Boutte, G. S., Zigler, E., & Finn-Stevenson, M. (2004).
Opportunities for schools to promote resilience in children and
youth. In K. Maton (Ed.). Investing in children, youth, families
and communities (pp. 213233). Washington DC: American
Psychological Association.
Sroufe, L. A., & Rutter, M. (1984). The domain of developmental
psychopathology. Child Development, 55, 1729.
U.S. Department of Health and Human Services. (2001). Youth violence: A report of the Surgeon General. Retrieved June 1, 2005,
from http://www.surgeongeneral.gov/library/youthviolence
Walker, H. M., & Gresham, F. M. (1997). Making schools safer and
violence free. Intervention in School and Clinic, 32, 199204.

Received October 7, 2005


Accepted May 8, 2006

Vous aimerez peut-être aussi