NIH Public Access
Author Manuscript
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
NIH-PA Author Manuscript
Published in final edited form as:
J Hum Behav Soc Environ. 2010 October 1; 20(7): 839–856. doi:10.1080/10911351003751918.
An Ecological Approach to Understanding Youth Violence: The
Mediating Role of Substance Use
Sung Seek Moon, PhD,
School of Social Work, University of Texas at Arlington
Joy Patton, PhD, and
School of Social Work, University of Arkansas at Little Rock
Uma Rao, MD
Department of Psychiatry, University of Texas Southwestern Medical Center
Abstract
NIH-PA Author Manuscript
The authors tested an ecological model that posits mediating variables (substance use and mental
health) in the association between ecological factors (family closeness, school closeness, and peer
closeness) and youth violence in a sample of 4,783 adolescents. Model including substance use
present significantly less total effect between ecological factors and youth violence than do models
without substance use. Additional probing of significant mediation effect using the Sobel test was
performed and suggested that substance use did function as a mediator in the hypothesized path.
Considerations of adolescent violence should recognize the possible role of ecological factors and
how their influence may vary by substance use.
Keywords
Family closeness; school closeness; peer closeness; substance use; violence; ecological model
NIH-PA Author Manuscript
Adolescent violence has been a great concern in American society for over two decades.
According to Marcus (2005), youth are more vulnerable to violence than any other age
group. Wiesner and Windle (2004) state, “adolescence is known as the peak period for
delinquent activity on the aggregate level and characterized by changes and transitions in
multiple domains (e.g. pubertal development, increasing engagement in or experimentation
with various deviant behaviors, identity development, initiation of romantic relationships)”
(p. 432). Data collected through the National Crime Victimization Survey in 1991 show that
the most likely age group to be involved in serious violent crimes were adolescents (Snyder
& Sickmund, 1995).
Many statistics have been published about adolescent violence, previously from research in
juvenile justice and school settings. Juvenile justice research shows that violent adolescent
crimes increased significantly from the 1980s to the mid 1990s, reaching its peak in 1994. In
2000, there were over two million arrests of adolescents under the age of 18, and of these
arrests, nearly 100,000 were for violent offenses (Snyder, 2002). By 2001, “juveniles
accounted for 17% of all arrests and 15% of all violent crime arrests” (Snyder, 2003, p.1).
Correspondence should be directed to Dr. Sung Seek Moon, School of Social Work, University of Texas at Arlington, 211 S. Cooper
St., Arlington, TX 76019; sungmoon@uta.edu.
Moon et al.
Page 2
NIH-PA Author Manuscript
Adolescents who are violent frequently spend time in a juvenile correctional facility and, in
some cases, spend time in adult correctional facilities depending on their age, the severity of
the offense and the state’s policy concerning transfers to criminal court for adolescents.
While male adolescents are more likely to participate in violence, there has been an increase
in the number of female adolescents participating in violence. Mullis, Cornille, Mullis, and
Huber (2004), report an alarming trend for female juveniles between 1989 and 1993 that
show “arrests for violent crimes increased 55% for juvenile females and by 33% for juvenile
males” (p. 206). Also, according to the U. S. Department of Justice (2003), there was a 59%
increase in female adolescents’ involvement with the juvenile justice system for all juvenile
offenses between 1990 and 1999.
NIH-PA Author Manuscript
Data from the Add Health Survey in 2001 showed that as much as 33% of high school
students had been involved in at least one physical fight and 7.5% of these fights resulted in
injury (Marcus, 2005). Data collected through the Center for Disease Control (2002) showed
that, of those students who were involved in at least one physical fight in the 12 months
prior to the survey, 54% had also been involved in fights in two or more instances, and of
the students who had reported being injured during a fight, 37% also reported that they had
been injured in fights in two or more instances. Furthermore, a World Health Organization
international study of youth violence revealed that nearly one half (47.8%) of adolescent
boys and one fourth (25%) of adolescent girls in the United States reported being involved
in at least one physical fight in 2000. Of youth surveyed in European and North American
countries, these results rank the United States as 29th in percentage of male youths involved
in at least one physical fight in 2000 and 11th in percentage of female youths involved in at
least one physical fight in 2000 (Krug, 2002).
Literature Review
Age and Violence
Research studies have shown that the earlier an adolescent begins a life of violence, the
longer the adolescent tends to offend, turning into a chronic offender with violent and
aggressive tendencies that extends into adulthood (Kethini, Blimling, Madden-Bozarth &
Gaines, 2004; Stone, 2003). Nagin, Farrington and Moffitt (1995) conducted a longitudinal
study and found that middle adolescents who unremittingly and repeatedly engage in violent
behavior are highly likely to have begun this behavior in early adolescence. Engaging in
violent behaviors at an early age has also been connected to a higher risk of developing
secondary problems such as poor academic achievement, substance abuse and mental health
issues such as depression (Capaldi & Patterson 1996).
NIH-PA Author Manuscript
Family Closeness
Additionally, several research studies have shown that factors such as family closeness,
school closeness, peer closeness, mental health issues and substance abuse have an effect on
adolescent violence. Family factors involving parenting practices of infrequent reasoning
and frequent spanking seem to escalate adolescent violence (Harbin & Madden, 1979;
Henry, Tolan & Gorman-Smith, 2001). Adolescents who reported lack of parent-child
interactions and communication were more likely to participate in delinquent acts, including
violence (Davalos, Chavez, & Guardiola, 2005; Henry et al., 2001; Kratcoski, 1985;
Whitebeck, Hoyt, & Ackley, 1997). Those adolescents who had a strong bond or attachment
during preadolescence extending into adolescence were less likely to be involved in violence
(Hoffmann, 2003; Paschall, Ennett, & Flewelling, 1996; Smith, Flay, Bell, & Weissberg,
2001). Lastly, parents who have abusive behavior, either physical or verbal, were more
likely to have children with these behaviors (Schreiber & Schreiber, 2002)
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
Moon et al.
Page 3
School and Peer Context
NIH-PA Author Manuscript
Outside the immediate family, the school setting has been identified as the most consistent
institution in the lives of children (Cicchetti, Toth, & Maughan, 2000). According to Maguin
and Loeber (1996), factors such as low academic achievement, academic failure and low
commitment to school are related to delinquent behaviors and chronic offending (Wiesner &
Windle, 2004). Also, adolescents who drop out of school are two and one half times more
likely to participate in any type of delinquent act (Davalos et al., 2005; Ellickson, Saner, &
McGuigan, 1997). Teacher support was also a significant predictor of pro-social behavior
and student interest in school activities (Charlebois et al., 2004; Wentzel, 1998). Prior
studies have provided empirical evidence that teacher support and good child-school
relationship are protective factors for adolescents’ behavioral and mental health outcomes
(Bowen et al., 1998; Fottland & Matre, 2005; Murdock, 1999; Vedder, Boekaertes, &
Seegers, 2005).
Additional social support in the school context can come from peers who are disapproving
of violence. In the National Youth Survey, respondents who reported associating with peers
who disapproved of violence reported less violent behavior by age 18 (Elliot, 1994).
Mental Health, Substance use, and Violence
NIH-PA Author Manuscript
There are also associations between adolescent violence and mental health, and adolescent
violence and substance abuse. Mental health issues like depression (Moss & Lynch, 2001;
Capaldi, 1992) have been found to be prevalent among adolescents who are violent, and
these adolescents are one and one half times more likely to have poor mental health overall
(Ellickson & McGuigan, 2000). Likewise, drug use was found to be a predictor of future
violence among adolescents (Sussman, Simon, Dent, Steinberg, & Stacy, 1999). In a metaanalysis of 66 longitudinal studies related to violent behavior, substance use and general
offenses (including status offenses and property crimes) had the strongest relationship to
future offending in children (Lipsey & Derzon, 1998). Adolescents who are violent are ten
times more likely to sell drugs and two to three times more likely to use alcohol, cigarettes
or marijuana on a weekly basis than their nonviolent counterparts (Ellickson et al., 1997).
These statistics reveal the magnitude and prevalence of violence among both male and
female adolescents. Previous research also points out the complexity of the violence
problem. While understanding direct effects of family closeness, school closeness, mental
health issues and substance abuse on adolescent violence is important, examining the
influence of mediating factors is equally important. In identifying mediating relationships,
the nature of the direct effects can be clarified.
NIH-PA Author Manuscript
Conceptual framework
Wilcox (2003) asserted that individual-level characteristics cannot account fully for the
behavior among adolescents because there appear to be persistent, non-random patterns of
adolescent problem behaviors across communities defined by geographical areas and groups
defined by gender, race, education level, income, occupation and marital status in the past
epidemiological study. This implies that we need a complimentary ecological model to focus
on factors in children’s lives that may place them at risk for, or help protect them from,
undesired outcomes, as well as individuals’ biological and psychological factors. Human
ecology focuses on “the relationship between environmental and demographic
characteristics of the population and the impact of these two broad sets of variables upon
human behavior” (McBride & McCoy, 1981, p. 284). Ecology model looks at a child’s
development and behavior within the system of the interactions between a child, immediate
environment (family, school, and peers) and larger social environment (community, society,
culture), as well as interactions among different levels of the environment (Bronfenbrenner,
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
Moon et al.
Page 4
NIH-PA Author Manuscript
1977; Kumpfer & Turner, 1990). A considerable number of studies have shown that
children’s ecological contexts may serve as risk or protective factors for substance use
(Robbins et al., 2006; Hawkins et al., 1992), mental health (Fotti et al., 2006) and problem
behaviors (Cookston, 1999; Gorman-Smith et al., 1996; Tarter et al., 2002; Windle, 2000).
Much of the extant research has focused on direct influence of contextual factors on youth
problem behaviors; however, considerably fewer studies have examined indirect or
interactional (mediating or moderating) effects of various factors across different
environmental contexts. Thus, we developed an integrated model including family
closeness, school closeness, and peer closeness as distal factors, as well as individuals’
mental health status and substance use as proximal factors on youth violence as shown in
Figure 1.
We developed hypotheses linking the exogenous constructs of family closeness, peer
closeness, and school closeness with the endogenous constructs of mental health status,
substance use, and youth violence. Specifically, Figure 1 hypothesizes that ecological
factors (family closeness, peer closeness, and school closeness) all directly affect mental
health and substance use. Also, mental health and substance use have mediating functions in
the relationship between ecological factors and youth violence.
NIH-PA Author Manuscript
Methods
Sample
To address these hypotheses, we used data from the first wave of the National Longitudinal
Study of Adolescent Health (Add Health). Add Health is a multi-survey, multi-wave study
of U.S. adolescents, their parents, and their schools (Bearman et al., 1998). In the initial inschool survey, conducted in 1994–1995, all students attending each of 134 high schools and
their “feeder” middle schools (grades 7 through 12) were interviewed (N = 90,118). A
randomly sampled subset of about 20,000 of these students, and in most cases was one of
their parents, was subsequently interviewed at home. For this analysis, we select respondents
who were attending school at the time of the initial in-home survey. The final sample
included 2,287 boys (47.8%) and 2,496 girls (52.2%), with a mean age of 16.01 (SD = 1.62,
range = 11-21).
Measures
NIH-PA Author Manuscript
Family closeness—Family closeness refers to the degree of close relationship with
family members. It was assessed with three items (e.g., family pays attention to you) on
five-point scale. Response categories ranged from 1 = “not at all” to 5 = “very much”.
Higher scores on the scale indicate more family closeness. Cronbach’s alpha was 0.79.
Peer closeness—Child-peer closeness was measured with an item, “friends care about
you”. A higher score on the item indicates more peer closeness. Response categories ranged
from 1 = “not at all” to 5 = “very much”.
School closeness—School closeness refers to the degree of close relationship with
teachers, and students, and feelings about school. It was measured with five items (e.g., feel
close to people at school) on a five-point scale. Response categories ranged from 1 =
“strongly agree” to 5 = “strongly disagree”. Items on this scale were reversed, with the
exception of item 3 (Students at school are prejudiced) so scores of the scale items would
reflect 1 equal to “strongly disagree” and 5 equal to “strongly agree” in order to correspond
with the previous mentioned scales of family closeness and peer closeness. Higher scores on
the scale indicate more school closeness. Cronbach’s alpha was 0.79.
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
Moon et al.
Page 5
NIH-PA Author Manuscript
Mental health—Mental health had three indicators, including depression, anxiety, and
self-esteem. Depression was assessed with a 19-item, modified Center for Epidemiologic
Studies-Depression Scale (CES-D). The original 20-item scale has been widely used as a
measure of depressive symptoms in epidemiologic research and as a first-stage screening
tool for clinical depression in community samples. In the modified scale available in Add
Health, 2 original CES-D items were dropped, namely, “My sleep was restless,” and “I had
crying spells.” One item was added; “I felt that life was not worth living.” Two additional
items were rephrased. Item scores on the modified scale correspond to symptom frequency
and range from 0 (never or rarely) to 3 (most or all of the time); overall scale scores thus
range from 0 to 56. The total score of the scale was used for the analyses. The scoring of
four positive statements on the scale (4, 8, 11, and 15) was reversed. Higher scores on the
scale indicate more depressive symptoms. Cronbach’s alpha for the 19-item scale was 0.78.
Anxiety was measured with two items (e.g., past year how often trouble relaxing) on fivepoint scale. Response categories ranged from 0 = “never” to 4 = “everyday”. Cronbach’s
alpha was 0.42. Self-esteem was measured with 7 items (e.g., have lots of good qualities) on
five-point scale. Response categories ranged from 1 = “strongly agree” to 5 = “strongly
disagree”. Cronbach’s alpha was 0.85.
NIH-PA Author Manuscript
Substance use—Substance use was measured with three items (e.g., smoked cigarette
regularly) with two response categories consisting of 0 = “no” and 1 = “yes”. Higher scores
on the scale indicate more substance use.
Violence—Youth violence was measured with four items (e.g., past year how often serious
physical fight). Responses were anchored on 4-point scales, with 0 = “not at all” and 3 = “5
or more times”. Higher scores on the scale indicate more violence. Cronbach’s alpha was
0.59. All measurement items are presented in Appendix A along with item-to-total
correlations for each measure.
Analyses and Results
NIH-PA Author Manuscript
The four indicators of violence were ordered categorical variables. Spearman correlations (a
nonparametric analysis used for ordinal data) and Pearson correlations between these
variables and the predictors were nearly identical. In addition, we ran tests of skewness and
kurtosis on all of our variables and found them to be generally normally distributed (see
Table 1). Given the similarity of the Spearman and Pearson correlations, the normal
distribution of the variables and the large sample size, we chose to use Pearson correlations
and structural equation modeling (SEM) using AMOS 6.0 (Arbuckle & Wothke, 2005).
These robust parametric procedures were selected to increase power and protect against
Type II error.
SEM was employed to evaluate the relationship between three distal factors (family
closeness, peer closeness, school closeness) and youth violence, as well as the mediating
role of proximal factors (mental health and substance use). A latent variables approach
offers potential advantages in SEM analysis as it enables measurement error to be taken into
account (Bollen, 1989). Typically, multiple items or measures are used to assess each latent
variable (i.e., measurement model). In the present application, however, the three
subconstructs of mental health (depression, anxiety, and self-esteem) were used as observed
variables in the estimation of the latent mental health variable.
Descriptive statistics
As an initial step prior to testing structural models, bivariate correlations between family
closeness, peer closeness, school closeness, mental health, substance use, and violence were
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
Moon et al.
Page 6
NIH-PA Author Manuscript
computed (see Table 2). All indicators of family closeness, school closeness, and peer
closeness were negatively correlated with the measures of mental health, substance use, and
violence. However, mental health and substance use were positively correlated with the
measures of violence.
Structural equation modeling
In addition to chi-square, model fit was evaluated via the Comparative Fit Index (CFI), the
Tucke-Lewis Index (TLI), and the root mean square error of approximation (RMSEA).
Values of .90 and above have been recommended for CFI and TLI, and values of .08 and
less have been suggested for RMSEA (Browne & Cudeck, 1993; Hu & Bentler, 1999).
Indices of overall fit were generally good for the final model (χ2 (129) = 1699.5, p < .001,
CFI = .94, TLI = .90, RMSEA = .04).
NIH-PA Author Manuscript
In the resulting model (Figure 1), the latent variable youth violence sufficiently represents
the three items with the standardized regression weights between 0.25 and 0.64. The
variable mental health also represents the three indicators with the standardized regression
weights between 0.36 and 0.71. Tobacco use, alcohol use, and marijuana use were
represented by the latent variable substance use (standardized regression weights between
0.42 and 0.63). Also, two exogenous variables (family closeness and school closeness)
significantly represent their indicators with the standardized regression weights from
between 0.25 and 0.79.
Mediation model
According to Baron and Kenny (1986), a mediating model must meet the following criteria:
the mediating variable (e.g., substance use) must significantly relate to the independent
variable (e.g., family closeness), the dependent variable (e.g., violence) must significantly
relate to the independent variable, and the dependent variable must significantly relate to the
mediating variable. A mediating role is suspected when a previously significant relationship
between the independent variable (e.g., family closeness) and dependent variable (e.g.,
violence) is substantially reduced when the mediation variable (e.g., substance use) is
entered into the equation (Baron & Kenny, 1986).
NIH-PA Author Manuscript
We hypothesized that higher family closeness, school closeness, and peer closeness are
related to lower health risk behaviors (i.e., violence composite). Higher closeness with
family, peer, and school was also expected to relate to better mental health and lower
substance use, which reasonably could be attributed to the adolescents’ behavior, which in
turn were then hypothesized to relate to engagement in more violence. Once the variable of
substance use is added to the structural equation model, the direct relationship between
ecological factors and violence should weaken. Substance use, therefore, was expected to
function as a partial mediator between ecological factors and violence among adolescents.
However, the variable mental health did not change the total effect in the association
between ecological factors and violence, so we did not test mediating function of mental
health. In addition, we did not test another possible mediation model of the association
between peer closeness and violence because peer closeness did not significantly predict
violence when tested directly.
As predicted for the direct path, family closeness had a direct effect on violence
(standardized β weight = -.09, p < .05). For the indirect path, family closeness predicted
substance use as expected (standardized β weight = -.38, p < .001). In addition, substance
use predicted violence (standardized β weight = .51, p < .001). Post-hoc probing of
significant mediation effect was performed as recommended by Holmbeck (2002) to
determine if the drop in the total effect (i.e., family closeness to violence) was significant
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
Moon et al.
Page 7
NIH-PA Author Manuscript
upon inclusion of the mediator (substance use) in the model. The direct effect of family
closeness to violence in the mediating model (standardized β weight = -.09, p <.05) was
significantly decreased when compared to the direct effect of the model when not including
substance use (standardized β weight = -.27, p <.001). The Soble test indicated that
substance use did function as a mediator (z = -6.65, p <.001) (Preacher & Leonardelli,
2001).
Also, substance use partially mediated the relationship between school closeness and
violence. For the direct path, school closeness had an effect on violence (standardized β
weight = -.10, p < .001). For the indirect path, school closeness predicted substance use
(standardized β weight = -.25, p < .001). In addition, substance use predicted violence
(standardized β weight = .51, p < .001). The direct effect of school closeness to violence in
the mediating model (standardized β weight = -.10, p <.05) was significantly decreased
when compared to the direct effect of the model when not including substance use
(standardized β weight = -.23, p <.001). The Sobel test indicated that substance use did
function as a mediator (z = -6.14, p <.001).
Discussion
NIH-PA Author Manuscript
Additional knowledge was gained in this study regarding the relationship between
ecological factors and youth violence. Existing literature examining relationships between
ecological predictors and youth violence has mostly relied on correlational descriptive
methods. These approaches are limited insofar as they cannot be used to model causal
pathways between predictors and outcome variable. In addition, the lack of theoretical
guidance is a common criticism of studies examining ecological influences on adolescent
outcomes. Thus, the structural equation modeling approach to test theoretically-based
hypothesized model for this paper fills a key gap in the current literature regarding specific
pathways of influence for family closeness, school closeness and peer closeness on youth
violence. Results of the study supported the proposed hypothesized structural equation
model. Consistent with ecological theory, environmental factors continued to uniquely
account for part of the path to adolescent violence, but substance use emerged as a
significant mediator as well. The result implies that ecological factors (family closeness,
school closeness, and peer closeness) affect violence directly and independently, but it is
also affected by substance use. As the family closeness and school closeness decreased,
adolescent’s substance involvement increased, and as adolescent’s substance use
involvement increased, violent behaviors increased.
NIH-PA Author Manuscript
Most prior studies of the association between ecological variables and adolescent problem
behaviors, however, examined main effects and did not consider the possibility that these
associations could be affected by substance use. Prior research has already shown a
reciprocal relationship between substance use and violence (Lipsey & Derzon, 1998;
Sussman et al., 1999). The results of our current study are consistent with a role for
substance use as a modifiable target area for interventions to reduce youth violence.
Our findings suggest that programs designed to prevent substance use may be appropriate
approaches to maximize the effects of protective ecological factors on youth violence.
Although our study focused on the mediating function of substance use, the findings also
underscored the direct effects of ecological factors on youth violence, suggesting that
interventions should equally emphasize improving children’s relationship with family,
school, and peers, as well as to prevent substance use.
A strength of the present study includes its large nationwide community sample of
participants including an ethnically and socioeconomically diverse population of youth. It
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
Moon et al.
Page 8
NIH-PA Author Manuscript
allows for generalization to larger populations, exploration of patterns across race/ethnicities
and socioeconomic status, and identification of groups at potentially high risk for substance
abuse and violence.
NIH-PA Author Manuscript
However, the present results need to be placed in the context of methodological limitations.
First, the items assessing each variable were brief, and prone to source variance due to selfreport by only one reporter (adolescent). Second, though the effects of several ecological
factors on youth violence were tested, other community factors, such as school quality,
gang-related activity, prevalence of substance use and abuse, violent crime statistics, or
community involvement should be included in the model to explain more fully how the
physical and social environments of children influence their behavioral development. Third,
we only tested the models on one age group. Future research is warranted to further
investigate these models and the age at which a shift takes place, or if it exists at all.
Longitudinal studies addressing how violence develops over time will be useful in
examining individual trajectories, and thus, whether the salience of ecological factors varies
developmentally. Fourth, we discussed the mediating role of substance use in the
relationship between ecological factors and youth violence in the current study, but some
possible moderators (potential buffers or exacerbating factors) of the effect of ecological
factors on youth violence should be recognized and they also should be included in the
model for a more comprehensive understanding of youth violence. Finally, we did not test
the variability among racial/ethnic groups. Distinct differences in the rates of violent
offending within different racial/ethnic groups have led to questions regarding the
relationship between race/ethnicity and violence.
There are implications on two levels that can be derived from the results of this study. These
implications are: (a) preventative strategies for children and families who are at risk for
future substance use and, (b) violent behavior patterns and issues that should be addressed in
adolescents who are already involved in using substances and exhibit violent behaviors.
NIH-PA Author Manuscript
First, preventative measures should be taken to avoid adolescent substance use and violent
behavior. The family environment plays a vital role in prediction of adolescent violence and
future substance use. One strategy to aid in prevention of both substance use and violence
could be to enact early childhood home visitations. Previous research has indicated that
children who grow up in abusive home environments are more likely to be abusive. Morantz
and Torrey (2004) report that, “approximately 40% of episodes of child maltreatment may
be prevented through early childhood home visits” (p. 997). Early childhood home visitation
programs should focus on a variety of training programs for parents such as child abuse/
neglect prevention, parenting skills, family planning assistance and problem-solving/life
skills training (Morantz & Torrey, 2004).
Second, since family connectedness and substance use share underlying risks of violent
behavior, for adolescents who are already involved in substance use and violent behaviors,
programs should target both behaviors. Working with adolescents to change unhealthy
behavior patterns of substance use and working to help modify their thoughts toward
violence by teaching prosocial problem-solving skills and conflict resolution while, at the
same time, working to strengthen the family bond could be an effective strategy
(Herrenkohl, McMorris, Catalano, Abbott, Hemphill, & Toumbourou, 2007).
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
Moon et al.
Page 9
Appendix A
NIH-PA Author Manuscript
Construction of Indices
Family closeness (4 items)
Answers range: 1 (Not at all) to 5 (Very much)
1.
Parents care about you.
2.
Family understands you.
3.
Family has fun together.
4.
Family pays attention to you.
School closeness (6 items)
Answers range: 1 (Strongly agree) to 5 (Strongly disagree)
NIH-PA Author Manuscript
1.
Feel close to people at school.*
2.
Feel part of your school.*
3.
Students at school are prejudiced.
4.
Happy at your school.*
5.
Teachers treat students fairly.*
6.
Feel safe in your school.*
Peer closeness (1 item)
Answers range: 1 (Not at all) to 5 (Very much)
1.
Friends care about you
Depressive symptoms (19 items)
Answers range: 1 (rarely or none of the time) to 5 (most or all of the time)
NIH-PA Author Manuscript
1.
In past week bothered by things.
2.
In past week had poor appetite.
3.
In past week had the blues.
4.
In past week felt just as good as other people.*
5.
In past week had trouble keeping mind focused.
6.
In past week felt depressed.
7.
In past week too tired to do things.
8.
In past week hopeful about the future.*
9.
In past week felt life had been a failure.
10. In past week felt fearful.
11. In past week felt happy.*
12. In past week talked less than usual.
13. In past week felt lonely.
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
Moon et al.
Page 10
14. In past week people unfriendly to you.
15. In past week enjoyed life.*
NIH-PA Author Manuscript
16. In past week felt sad.
17. In past week felt people dislike you.
18. In past week hard to start doing things.
19. In past week felt life not worth living.
Anxiety (2 items)
Answers range: 0 (never) to 4 (everyday)
1.
Past year how often-trouble relaxing.
2.
Past year how often-fearfulness.
Self-Esteem (7 items)
Answers range: 1 (strongly agree) to 5 (strongly disagree)
NIH-PA Author Manuscript
1.
Have lots of good qualities.*
2.
Physically fit.*
3.
Have a lot to be proud of.*
4.
Like self as are.*
5.
Do everything just right.*
6.
Feel socially accepted.*
7.
Feel loved and wanted.*
Substance use (3 items)
1.
Smoked cigarettes regularly.
2.
Drink alcohol more than 2-3 times.
3.
Age first tried marijuana.
Violence (3items)
NIH-PA Author Manuscript
1.
Past year-how often serious physical fight.
2.
Past year-how often take part in a group fight.
3.
Past year-how often loud or rowdy in a public place.
* The scoring of positive items is reversed.
Acknowledgments
This work was supported in part by grants from the National Institutes of Health (DA14037, DA15131, DA17804,
DA17805, MH62464 and MH68391), and the Sarah M. and Charles E. Seay Endowed Chair in Child Psychiatry at
UT Southwestern Medical Center.
References
Arbuckle, J.; Worhke, W. AMOS 6.0 [Computer software]. Chicago: Smallwaters; 2005.
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
Moon et al.
Page 11
NIH-PA Author Manuscript
NIH-PA Author Manuscript
NIH-PA Author Manuscript
Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research:
Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology
1986;51:1173–1182. [PubMed: 3806354]
Bearman, PS.; Jones, J.; Udry, JR. The national longitudinal study of adolescent health: Research
design. 1998. Retrieved January 15, 1999, from
http://www.cpc.unc.edu/projects/addhealth/design.html
Bollen, K. Structural equations with latent variables. New York: Wiley; 1989.
Bowen GL, Richman JM, Brewster A, Bowen N. Sense of school coherence, perceptions of danger at
school, and teacher support among youth at risk of school failure. Child & Adolescent Social Work
Journal 15:273–286.
Bronfenbrenner U. Toward an experimental ecology of human development. American Psychologist
1977;32:513–531.
Browne, MW.; Cudeck, R. Alternative ways of assessing model fit. In: Bollen, KA.; Lond, JS., editors.
Testing structural equation models. Newbury Park, CA: Sage; 1993. p. 136-162.
Capaldi DM, Patterson GR. Can violent offenders be distinguished from frequents offenders:
Predictions from childhood to adolescence. Journal of Research in Crime and Delinquency
1996;33:206–231.
Center for Disease Control and Prevention. National youth risk behavior survey: Trends in the
prevalence of behaviors that contribute to violence. Washington, DC: U S Department of Health and
Human Services; 2002.
Charlebois P, Vitaro F, Normandeau S, Brendgen M, Rondeau N. Trainers’ behavior and participants’
persistence in a longitudinal preventive intervention for disruptive boys. Journal of Primary
Prevention 2004;25:375–388.
Cicchetti, D.; Toth, SL.; Maughan, A. An ecological-transactional model of child maltreatment. In:
Sameroff, AJ.; Lewis, M.; Miller, SM., editors. Handbook of developmental psychopathology. 2.
New York: Plenum; 2000. p. 689-722.
Cookston JT. Parental supervision and family structure: Effects on adolescent problem behaviors.
Journal of Divorce and Remarriage 1999;32:107–122.
Davalos DB, Chavez EL, Guardiola RJ. Effects of perceived parental school support and family
communication on delinquent behaviors in Latinos and white non-Latinos. Cultural diversity &
ethnic minority psychology 2005;11(1):57–68. Retrieved February 18, 2007, from CSA
Sociological Abstracts database. [PubMed: 15727495]
Ellickson P, McGuigan K. Early predictors of adolescent violence. American Journal of Public Health
2000;90:566–572. [PubMed: 10754971]
Ellickson P, Saner H, McGuigan K. Profiles of violent youth: Substance use and other concurrent
problems. American Journal of Public Health 1997;87:985–990. [PubMed: 9224181]
Elliot DS. Serious violent offenders: Onset, developmental course, and termination. Criminology
1994;32:1–21.
Fotti SA, Katz LY, Afifi TO, Cox BJ. The associations between peer and parental relationships and
suicidal behaviors in early adolescents. Canadian Journal of Psychiatry 2006;51:698–703.
Fottland H, Matre S. Assessment from a sociocultural perspective: Narratives from a first grade
classroom. Scandinavian Journal of Educational Research 2005;49:503–321.
Gorman-Smith D, Tolan PH, Zelli A, Huesmann LR. The relation of family functioning to violence
among inner-city minority youths. Journal of Family Psychology 1996;10:115–129.
Harbin, HT.; Madden, DJ. Battered parents: A new syndrome. Baltimore: University of Medical
School Press; 1979.
Hawkins JD, Catalano RF, Miller JY. Risk and protective factors for alcohol and other drug problems
in adolescence and early adulthood: Implications for substance abuse prevention. Psychological
Bulletin 1992;112:64–105. [PubMed: 1529040]
Henry DB, Tolan PH, Gorman-Smith D. Longitudinal family and peer group effects on violence and
nonviolent delinquency. Journal of Clinical Child Psychology 2001;30:172–186. [PubMed:
11393918]
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
Moon et al.
Page 12
NIH-PA Author Manuscript
NIH-PA Author Manuscript
NIH-PA Author Manuscript
Herrenkohl TI, McMorris BJ, Catalano RF, Abbott RD, Hemphill SA, Toumbourou JW. Risk factors
for violence and relational aggression in adolescence. Journal of Interpersonal Violence
2007;22:386–405. [PubMed: 17369443]
Hoffmann JP. A contextual analysis of differential association, social control, and strain theories of
delinquency. Social Forces 2003;81:753–785.
Holmbeck GN. Post-hoc probing of significant moderational and mediational effects in studies of
pediatric populations. Journal of Pediatric Psychology 2002;27:87–96. [PubMed: 11726683]
Hu L, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria
versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal 1999;6:1–55.
Kratcoski PC. Youth violence directed toward significant others. Journal of Adolescence 1985;8:145–
157. [PubMed: 4019876]
Kethini S, Blimling L, Madden-Bozarth J, Gaines C. Youth violence: An exploratory study of a
treatment program in Central Illinois County. International Journal of Offender Therapy and
Comparative Criminology 2004;48(6):697–720. [PubMed: 15538027]
Krug, EG. World report on violence and health. Geneva: World Health Organization; 2002.
Kumpfer KL, Turner CW. The social ecology model of adolescent substance abuse: Implications for
prevention. International Journal of addictions 1990;25:435–463.
Lipsey, MW.; Derzon, JH. Predictors of violent or serious delinquency in adolescence and early
adulthood: A synthesis of longitudinal research. In: Loeber, R.; Farrington, DP., editors. Serious
and violent juvenile offenders. Thousand Oaks, CA: Sage; 1998. p. 86-105.
Maguin, E.; Loeber, R. Academic performance and delinquency. In: Tonry, M., editor. Crime and
Justice. Chicago: University of Chicago Press; 1996.
Marcus RF. Youth violence in everyday life. Journal of Interpersonal Violence 2005;20:442–447.
[PubMed: 15722499]
McBride DC, McCoy CB. Crime and drug-using behavior. Criminology 1981;19:281–302.
Morantz C, Torrey B. Strategies to prevent juvenile violence. American Family Physician
2004;69:997.
Moss HB, Lynch KG. Comorbid disruptive behavior disorder symptoms and their relationship to
adolescent alcohol disorders. Drug and Alcohol Dependence 2001;64:75–83. [PubMed: 11470343]
Mullis RL, Cornille TA, Mullis AK, Huber J. Female juvenile offending: A review of characteristics
and contexts. Journal of Child and Family Studies 2004;13:205–218.
Murdock TB. The social context of risk: Status and motivational predictors of alienation in middle
school. Journal of Educational Psychology 1999;91:62–75.
Nagin DS, Farrington DP, Moffitt TE. Life-course trajectories of different types of offenders.
Criminology 1995;33:111–139.
Paschall MJ, Ennett ST, Flewelling RL. Relationships among family characteristics and violent
behavior by Black and White male adolescents. Journal of Youth and Adolescence 1996;25:177–
197.
Preacher, KJ.; Leonardelli, GJ. Calculation for the Sobel test: An interactive calculation tool for
mediation tests [Computer software]. 2001. Retrieved from
http://www.unc.edu/~preacher/sobel/sobel.htm
Robbins MS, Briones E, Schwartz SJ, Dillon FR, Mitrani VB. Differences in family functioning in
grandparent and parent-headed households in a clinical sample of drug-using African American
adolescents. Cultural Diversity & Ethnic Minority Psychology 2006;12:84–100. [PubMed:
16594857]
Schreiber EH, Schreiber KN. A study of parents of violent children. Psychological Reports
2002;90:101–104. [PubMed: 11898970]
Smith P, Flay BR, Bell CC, Weissberg RP. The protective influence of parents and peers in violence
avoidance among African American youth. Maternal and Child Health 2001;5:245–252.
Snyder, HN. Juvenile arrests 2001: Juvenile justice bulletin. Washington, DC: U.S. Department of
Justice, Office of Justice Programs; 2003.
Snyder, HN. Juvenile arrests 2000. Washington, DC: Office of Juvenile Justice Delinquency and
Prevention; 2002.
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
Moon et al.
Page 13
NIH-PA Author Manuscript
NIH-PA Author Manuscript
Snyder, HN.; Sickmund, M. Juvenile offenders and victims: A national report. Washington, DC:
Office of Juvenile Justice and Delinquency Prevention; 1995.
Sussman S, Simon T, Dent C, Steinberg J, Stacy A. One-year prediction of violence perpetration
among high-risk youth. American Journal of Health and Behavior 1999;23:332–344.
Tarter RE, Kirisci L, Vanyukov M, Cornelius J, Pajer K, Shoal GD, Giancola PR. Predicting
adolescent violence: Impact of family history, substance use, psychiatric history, and social
adjustment. American Journal of Psychiatry 2002;159:1541–1547. [PubMed: 12202275]
Vedder P, Boekaerts M, Seegers G. Perceived social support and well being in school: The role of
students’ ethnicity. Journal of Youth and Adolescence 2005;34:269–278.
Wentzel KR. Social relationships and motivation in middle school: The role of parents, teachers, and
peers. Journal of Educational Psychology 1998;90:202–209.
Whitebeck LB, Hoyt DR, Ackley KA. Families of homeless and runaway adolescents: A comparison
of parent/caretaker and adolescent perspectives on parenting, family violence and adolescent
conduct. Child Abuse and Neglect 1997;21:517–528. [PubMed: 9192141]
Wiesner M, Windle M. Assessing covariates of adolescent delinquent trajectories: A latent growth
mixture modeling approach. Journal of Youth and Adolescence 2004;33:431–442.
Wilcox P. An ecological approach to understanding youth smoking trajectories: Problems and
prospects. Addiction 2003;98:57–77. [PubMed: 12752362]
Windle M. A latent growth curve model of delinquent activity among adolescents. Applied
Developmental Science 2000;4:193–207.
U. S. Department of Justice. Juvenile Court Statistics 1999. Washington, D.C: Office of Juvenile
Justice and Delinquency Prevention; 2003.
NIH-PA Author Manuscript
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
Moon et al.
Page 14
NIH-PA Author Manuscript
Figure 1.
Structural Equation Model.
Note. FC: Family closeness. PC: Peer closeness. SC: School closeness. V: Violence. F1:
Understand. F2: Fun together. F3: Attention. S1: Feel close to people at school. S2: Feel part
of your school. S3: Students at school are prejudiced. S4: Happy at your school. S5:
Teachers treat students fairly. S6: Feel safe in your school. T: Smoked cigarettes regularly.
A: Drink alcohol more than 2-3 times. M: Age first tried marijuana. D: Depression. X:
Anxiety. E: Self-esteem. V1: Physical Fight. V2: Group fight. V3: Loud in a public place.
NIH-PA Author Manuscript
NIH-PA Author Manuscript
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
Moon et al.
Page 15
Table 1
Mean, Standard Deviation, Skewness, and Kurtosis.
NIH-PA Author Manuscript
Mean
SD
Skewness
Kurtosis
1. Family Closeness (FC)
Understand (FC1)
3.61
1.01
-.40
-.29
Fun together (FC2)
3.75
1.03
-.55
-.20
Attention (FC3)
3.93
.94
-.68
.13
Feel Closeness (SC1)*
2.29
1.01
.79
.23
Feel Part of School (SC2)*
2.29
1.01
.96
.53
Prejudiced (SC3)
2.91
1.21
.06
-1.00
Happy (SC4)*
2.32
1.14
.77
-.16
Teacher’s Fairness (SC5)*
2.52
1.09
.54
-.44
Feel Safeness (SC6)*
2.19
1.02
.87
.33
21.21
6.96
.89
1.12
2. School Closeness (SC)
3. Mental Health (MH)
NIH-PA Author Manuscript
Depression (MH1)
Anxiety (MH2)
1.21
1.29
.52
.53
26.30
4.23
-.41
.25
Tobacco (SU1)
.56
.50
-.23
-1.95
Alcohol (SU2)
.55
.50
-.21
-1.96
3.71
6.32
1.17
-.52
Self-Esteem (MH3)*
4. Substance Use (SU)
Marijuana (SU3)
5. Violence (V)
Physical Fight (V1)
.45
.77
1.83
2.86
Group Fight (V2)
.25
.60
2.76
8.08
Loud in a public place (V3)
.69
.89
1.26
.78
*
Reversed scores
NIH-PA Author Manuscript
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
NIH-PA Author Manuscript
NIH-PA Author Manuscript
NIH-PA Author Manuscript
Table 2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
J Hum Behav Soc Environ. Author manuscript; available in PMC 2011 April 1.
1. FC1
–
2. FC2
.52**
–
3. FC3
.55**
.61**
–
4. SC1
.21**
.18**
.18**
–
5. SC2
.23**
.23**
.23**
.58**
–
6. SC3
.13**
.11**
.10**
.11**
.13**
–
7. SC4
.23**
.24**
.22**
.46**
.53**
.18**
–
8. SC5
.22**
.21**
.23**
.26**
.28**
.17**
.39**
–
9. SC6
.17**
.14**
.18**
.31**
.34**
.17**
.40**
.38**
–
10. MH1
-.32**
-.27**
-.31**
-.21**
-.24**
-.16**
-.25**
-.21**
-.21**
–
11. MH2
-.36**
-.32**
-.34**
-.26**
-.34**
-.11**
-.30**
-.20**
-.22**
.44**
–
12. MH3
.20**
.15**
.17**
.08**
.11**
.14**
.11**
.11**
.09**
-.40**
-.26**
–
13. SU1
-.14**
-.14**
-.12**
-.09**
-.15**
-.09**
-.18**
-.10**
-.07**
.10**
.10**
.03
–
14. SU2
-.20**
-.22**
-.18**
-.09**
-.10**
-.15**
-.16**
-.17**
-.08**
.19**
.13**
.10**
.14**
–
15. SU3
-.19**
-.21**
-.15**
-.11**
-.15**
-.08**
-.19**
-.16**
-.07**
.17**
.13**
.11**
.23**
.41**
–
16. V1
-.09**
-.07**
-.10**
-.09**
-.11**
-.02
-.14**
-.12**
-.13**
.13**
.05**
.03*
.06**
.12**
.15**
–
17. V2
-.09**
-.09**
-.10**
-.03**
-.05**
-.05**
-.09**
-.12**
-.09**
.13**
.05**
.04**
.02
.17**
.20**
.43**
–
18. V3
-.19**
-.18**
-.16**
-.06**
-.07**
-.08**
-.13**
-.16**
-.07**
.17**
.12**
.13**
.06**
.25**
.21**
.23**
.30**
18
Moon et al.
Correlations.
–
Note. FC: Family closeness. SC: School closeness. MH: Mental health. SU: Substance use. V: Violence.
**
p < 0.01.
Page 16