Links between Family Conflict and Substance-Use in Early Adolescence:
Mediating Role of Bully Perpetration
Mrinalini A. Rao, M.S., & Dorothy L. Espelage, Ph.D. , University of Illinois, Urbana-Champaign
Todd D. Little, Ph.D., University of Kansas, Lawrence
Introduction
Research has demonstrated an association between
early exposure to family conflict and substance use in
adolescence and early adulthood (Skeer, McCormick,
Normand, Buka, & Gilman, 2009). The family is the
primary social context in which children develop and
continues to be a salient developmental context in early
adolescence.
Family conflict has been associated with an increase in
both internalizing and externalizing behaviors in
adolescence (Shelton & Harold, 2008). Family conflict
has also been found to increase the risk of bully
perpetration for adolescents (Baldry & Farrington, 2000;
Loeber & Dishion, 1984), potentially through social
learning processes that demonstrate that aggression and
violence are acceptable and effective ways of interacting
with others as postulated in the social interaction learning
theory (e.g., Conger & Rueter, 1996; Vakalahi, 2001, for a
review) . Thus, interpersonal processes in the family are
hypothesized to spill over into the peer context and
increase the risk for bully perpetration.
Both family conflict and bully perpetration have been
found to predict substance-use in adolescence, possibly
as an extension of engagement with externalizing
behaviors (Skeer et al., 2009; Pepler, Craig, Connolly, &
Henderson, 2002). The present study examines whether
such transference of interpersonal aggressive behaviors
occurs from the family setting to the peer setting. We
hypothesized that family stressors, including family
conflict and sibling aggression, predicts bully
perpetration, which in turn predicts substance-use using
a longitudinal design.
.
RESEARCH QUESTION
Does bully perpetration mediate the
relation between family conflict and
substance use in a middle-school
sample?
Research was supported by Centers for Disease
Control & Prevention (#1U01/CE001677) to Dorothy
Espelage (PI).
Method
Design & Sample
Participants included 1,132 middle school students
(grades 5-7) from four schools in a mid-western city (34%
Caucasian, 51% Black, 3% Hispanic, 3%Asian, 1%
Native American/Indian, 8% Other). The sample was
evenly distributed among males and females. Data were
collected over three waves including Spring 2008, Fall
2008 and Spring 2009 and included three cohorts (5th
graders in 2008 – 7th graders in 2009; 6th graders in 2008
– 8th graders in 2009; 7thth graders in 2008 – 9th graders
in 2009).
Measures
Family conflict was measured using both The Family
Conflict and Hostility Scale (Thornberry, Krohn, Lizotte,
Smith, & Tobin, 2003) and a Sibling Aggression
Perpetration scale that was created for the study. The
Family Conflict and Hostility Scale measures the level of
self-reported conflict and hostility in the family
environment. Respondents indicated on a 4-point scale
how often hostile situations had occurred in their families
in the past 30 days. Responses range from 1 (Often)
through 4 (Never). Higher scores indicating higher levels
of family conflict and hostility. In the development of the
Sibling Aggression Perpetration Scale, five items
emerged as a scale in factor analysis (i.e., I upset my
brother or sister for the fun of it; I got into a physical fight
with my brother or sister; I started arguments with my
brother or sister; I hit back when a sibling hit me first; and
I teased my siblings for the fun of it). A Cronbach alpha
coefficient of .81 was found for this study.
Bullying perpetration was measured using the nine-item
Illinois Bully Scale (Espelage & Holt, 2001) which
assesses the frequency of teasing, name-calling, social
exclusion, and rumor spreading. Students are asked how
often in the past 30 days they teased other students,
upset other students for the fun of it, excluded others
from their group of friends, and helped harass other
students etc. Response options include “Never”, “1 or 2
times”, “3 or 4 times”, “5 or 6 times”, and “7 or more
times.” A Cronbach alpha coefficient of .86 was found for
this sample.
Substance use was assessed with an 8-item scale
(Farrell, Kung, White, & Valois, 2000) asking students to
report how many times in the past year they used alcohol
and/or drugs. The scale consists of items such as,
“Smoked cigarettes”, “Drunk liquor”, and “Used
inhalants”. Responses are recorded on a 5-point Likerttype scale with options ranging from 1 (Never) to 5 (10 or
more times). In the current study, the Cronbach’s alpha
was .90.
Data Analysis & Results
•
•
•
•
•
•
Longitudinal structural equation modeling was conducted using LISREL 8.8.
Confirmatory factor analysis was used to test measurement model.
Factorial invariance was tested (using effects-coding method).
The mediation model was tested using the Cole and Maxwell (2003) framework.
The overall indirect effect of mediation was .020.
The significance of this estimate was tested using the Monte Carlo simulation approach developed by MacKinnon and
colleagues (2004; see also Preacher & Selig, 2012).
• The confidence interval for this effect was .004 – .033 indicating that the overall indirect effect is significant.
Table 1
Fit Indices for the Single Group Confirmatory Factor Analysis and Final SEM Model
Model
χ2
df
Configural Invariance 1237.059 474
Weak Invariance
1522.646 492
Strong Invariance
1639.181 510
Final SEM Model
2287.475 548
0.68
Family
Conflict
Time 1
p
RMSEA
<.001 0. 0368
<.001 0. 0424
<.001 0. 0434
<.001 0. 0510
Family
Conflict
Time 2
90% CI RMSEA
0.0443; 0.0520
0.0538; 0.0612
0.0550; 0.0621
0.0540; 0.0609
0.73
Bully
Perpetration
Time 1
Bully
Perpetration
Time 2
0.50
CFI
Constraint Tenable?
0.987
-0.982
Yes
0.980
Yes
0.970
Yes
0.70
Family
Conflict
Time 4
0.35
0.25
0.24
0.37
Family
Conflict
Time 3
NNFI
0. 982
0 .977
0. 976
0. 965
Bully
Perpetration
Time 3
Conclusions
0.30
Bully
Perpetration
Time 4
•Majority of sexual harassment perpetration is perpetrated
0.23
-0.05
by boys to girls; boys to boys
•No differences in cross-sex bullying
•Involvement in bullying and homophobic teasing
of sexual Substance
harassment perpetration
0.38 perpetration
0.75
Substance
Substance
Substancepredictive0.74
over time
Use
Use
Use
Use
Time 1
Time 2
3
•PredictTime
bullying
perpetration moreTime
than4 sexual harassment
perpetration with established risk & protective factors
•Bullying
is CAUSALLY linked to homophobic
Figure 1. Mediational
Model perpetration
Tested
bullying.
The results of the model tested are shown in the figure above. These
•School-based
results indicate
bullying
bullyprevention
perpetration
programs
fully mediates
during middle
the
relation between family conflict and subsequent AOD use. It is possible
school
that
willyouth
only have
who are
an impact
targets on
of bullying
reducingexperience
sexual violence
lower
self-esteem which in turn might lead to their using alcohol and other
perpetration
drugs to cope
when
emotionally.
homophobic teasing is included and
other SH-specific variables are considered.
0.15
Discussion
Findings from this study provide strong support for the links among family conflict, bully perpetration, and initiation and
continued alcohol and other drug (AOD) use in middle school. Family conflict emerged as a precursor to increased bully
perpetration, which in turn predicted increased AOD use. It provides support for the social interaction learning theory, and the
hypothesis that conflictual relationships in the family result in externalizing behaviors such as bully perpetration, which in turn
puts youth at risk for substance use. This finding sheds light on a potential point of intervention, and both bullying and
substance-use prevention programs would do well to consider how conflictual relationships within the family and among peers
can put youth at an increased risk for substance use in early adolescence. Future research must account for additional
variables that impact and influence this crucial and risky association.