Transcript School Aggression and Bullying

School bullying as a risk
factor for violence,
depression and other
adverse outcomes later in
life: Implications for
protecting school youth.
David P. Farrington and Maria M. Ttofi
Campbell Collaboration Colloquium,
Copenhagen, May 31, 2012
Pernille Due et al.(Denmark)
• Due et al. (2005) carried out perhaps the largest study of
the prevalence of being bullied (sometimes or more often
during this school term) among nationally representative
samples of 11–15 year olds in 28 western industrialised
countries (surveying over 4,000 students per country on
average). (later extended to 66 countries!)
• Overall, 18 per cent of boys and 15 per cent of girls were
bullied according to this fairly demanding criterion.
• In the United States 16 per cent of boys and 11 per cent
of girls were bullied.
• In Denmark, 26 per cent of boys and 24 per cent of girls
were bullied!
Indications from previous research…
Short term effects:
>depressive symptomatology (Bosworth et al, 1999; Van
der Wal et al, 2003)
=>increased risk for suicidal ideation and self-injurious
behaviour (Kaltiala-Heino et al, 1999)
=>eating disorders (Kaltiala-Heino et al, 2000)
• Long-term effects:
=>later offending (Farrington, 1993; Sourander et al,
2006; Losel, 2008)
=> difficulties in trust/intimacy in opposite-sex and
friendly relationships in adulthood (Gilmartin, 1987;
Dietz, 1994)
School bullying: Is it really related to
later adverse outcomes?
Aims of the review:
1. To conduct a systematic review (minimizing bias)
2. Meta-analyze data; calculate standardized effect sizes with the
final aim of:
a) Establishing whether there is indeed an association between
bullying perpetration/victimization and health/criminal
outcomes
b) Establishing where the strongest effect lies; guiding future
bullying prevention initiatives (e.g. do victims suffer from low
self-esteem?; current findings on empathy: D. Jolliffe)
c) Establishing the unique contribution of school
bullying/victimization (i.e. after controlling for covariates)
across-time (predictive efficacy)
d) Relating bullying to general criminology (e.g.: bullying
prevention = crime prevention?)
Why does bullying/victimization
predict health/criminal outcomes?
• Persistence of underlying tendency?
• Or does occurrence of B/V increase probability
of later outcomes?
• E.g. if B is reinforced or B leads to decrease in
bonding to society or increase in antisocial peers
• Can’t distinguish persistent heterogeneity from
state dependence in systematic review
• Can test if B/V to H/C is driven by risk factor
e.g. impulsiveness or low socioeconomic status
Specified Risk Factors (based on Murray et al.
JEC paper on Methodological Quality)
Outcome Measures for British Academy Project
Results of Searches up to March 2012
Have obtained 600 reports screened studies in the systematic
review
Of the 436 included reports
Bullying Perpetration versus Violence
Later in Life
[based only on prospective longitudinal
studies]
51 reports on violence from 28 longitudinal studies
Measure of Effect Size (Odds Ratio)
•
• Non-Bully
• Bully
Non-Violent
225
50
Violent
75
50
• OR = (225*50)/(75*50) = 3.0
• OR > 1 indicates desirable effect
300
100
Bullying Perpetration vs Violence:
Unadjusted OR = 2.97 (95% CI: 2.25 – 3.92)
Adj. OR = 2.04 (95% CI: 1.69 – 2.45)
The meaning of OR = 2.04
This value of the OR indicates a very strong
relationship between bullying perpetration and
later violence. For example, if a quarter of
children were bullies and a quarter were violent,
this value of the OR would correspond to 35.8%
of bullies becoming violent, compared with
21.4% of non-bullies. Thus, being a bully would
increase the risk of being violent (even after
controlling for other childhood risk factors) by
about two-thirds.
School Bullying versus Violence: Metaregressions
Moderator Variables for Heterogeneity
(Q = 75.80, p = .0001)
• Age at Time 1/Bullying
(range: 8.00 – 15.54 years; M = 12.04; SD = 2.35)
• Age at Time 2/Outcome
(range: 10.00 – 24.64 years; M = 17.65; SD =
4.83)
• Length of the follow-up period
(range: 0.42 – 16.50 years; M = 5.61; SD = 4.88)
• Number of Covariates
(range: 2 – 20; M = 6.93; SD = 5.25)
Meta-Regression Results
• Age Time 1:
(B= - 0.065; SE = 0.021; p = 0.002)
• Length of Follow-Up:
(B= - 0.017; SE = 0.009; p = 0.051)
• Age Time 2:
(B= - 0.033; SE = 0.009; p = 0.0005)
• Number of Covariates
(B= - 0.013; SE = 0.010; p = 0.185)
Caveat: many uncontrolled variables
No publication bias
Some further findings…
Bullying Perpetration versus Offending
48 reports on offending from 29 longitudinal studies
Bullying Perpetration vs Offending:
Unadjusted OR = 2.54 (95% CI: 2.05 – 3.14)
Adjusted OR = 1.89 (95% CI: 1.60 – 2.23)
The meaning of OR = 1.89
This OR indicates quite a strong relationship
between bullying perpetration and later
offending. For example, if a quarter of children
were bullies and a quarter were offenders, this
value of the OR would correspond to 34.5% of
bullies becoming offenders, compared with
21.8% of non-bullies. Thus, being a bully would
increase the risk of being an offender (even after
controlling for other childhood risk factors) by
more than half.
School Bullying versus Offending: Metaregressions
Moderator Variables for Heterogeneity
(Q = 36.82, p = .001)
• Age at Time 1/Bullying
(range: 6.23 – 15.54 years; M = 11.26; SD = 2.68)
• Age at Time 2/Outcome
(range: 10.00 – 24.64 years; M = 17.10; SD = 4.91)
• Length of the follow-up period
(range: 0.42 – 16.50 years; M = 5.84; SD = 4.56)
• Number of Covariates
(range: 1 – 20; M = 7.00; SD = 5.22)
Meta-Regression Results
• Age Time 1:
(B = .019, SE = .024, p = .428)
• Length of Follow-Up:
(B = -.027, SE = .012, p = .018)
• Age Time 2:
(B = -.025, SE = .012, p = .039)
• Number of Covariates
(B = -.027, SE = .013, p = .037)
Some further findings…
Bullying Perpetration versus Drug Use
23 reports on drug use from 13 longitudinal studies
Bullying Perpetration vs. Drug Use:
Unadjusted OR = 2.44 (95% CI: 1.73 – 3.43)
Adjusted OR = 1.49 (95% CI: 1.21 – 1.84)
The meaning of OR = 1.49
• This value of the OR indicates a moderate
relationship between bullying perpetration and
later drug use. For example, if a quarter of
children were bullies and a quarter were drug
users, this value of the OR would correspond to
30.9% of bullies doing drugs, compared with
23.0% of non-bullies. Thus, being a bully would
increase the risk of drug use later in life (even
after controlling for other childhood risk factors)
by about one-third.
Some further findings…
Bullying Victimization versus Depression
75 reports on depression from 49 longitudinal studies
Bullying Victimization vs. Depression:
Unadjusted OR = 1.99 (95% CI: 1.69 – 2.33)
Adjusted OR = 1.71 (95% CI: 1.49 – 1.96)
The meaning of OR = 1.71
• This value of OR indicates a strong relationship
between bullying victimization and depression.
For example, if a quarter of children were
victims and a quarter were depressed, this value
of the OR would correspond to 33.0% of victims
becoming depressed, compared with 22.3% of
non-victims. Thus, being a victim would
increase the risk of being depressed (even after
controlling for other childhood risk factors) by
about half.
Bullying victimization versus Depression:
Meta-regressions
Moderator Variables for Heterogeneity
(Q = 50.88, p = .0001)
• Age at Time 1/Bullying
(range: 8.00 – 18.00 years; M = 12.32; SD = 2.74)
• Age at Time 2/Outcome
(range: 10.00 – 47.00 years; M = 19.45; SD =
9.64)
• Length of the follow-up period
(range: 1.00 – 36.00 years; M = 7.13; SD = 8.79)
• Number of Covariates
(range: 1 – 20; M =6.42; SD = 5.06)
Meta-Regression Results
• Age Time 1:
(B = -.028, SE = .012, p = .026)
• Length of Follow-Up:
(B = -.007, SE = .004, p = .055)
• Age Time 2:
(B = -.007, SE = .003, p = .026)
• Number of Covariates
(B = .020, SE = .008, p = .017)
Other results (adjusted ORs)
•
•
•
•
•
•
•
•
Victimization versus violence: OR = 1.42
(bullying: OR = 2.04)
Victimization versus offending: OR = 1.14
(bullying: OR = 1.89)
Victimization versus drug use: OR = 1.00
(bullying: OR = 1.49)
Bullying versus depression: OR = 1.41
(victimization: OR = 1.71)
Key references
• Farrington, D. P., Lösel, F., Ttofi, M. M. and Theodorakis, N. (2012) School
Bullying, Depression and Offending Behaviour Later in Life: An Updated
Systematic Review of Longitudinal Studies. Stockholm: Swedish National
Council for Crime Prevention.
• Ttofi, M. M., Farrington, D. P. and Lösel, F. (2012) School bullying as a
predictor of violence later in life: A systematic review and meta-analysis of
prospective longitudinal studies. Aggression and Violent Behaviour.
• Ttofi, M. M., Farrington, D. P., Lösel, F. and Loeber, R. (2011) Do the
victims of school bullies tend to become depressed later in life? A
systematic review and meta-analysis of longitudinal studies. Journal of
Aggression, Conflict and Peace Research, 3, 63-73.
• Ttofi, M. M., Farrington, D. P., Lösel, F. and Loeber, R. (2011) The
predictive efficiency of school bullying versus later offending: A
systematic/meta-analytic review of longitudinal studies. Criminal
Behaviour and Mental Health, 21, 80-89.
Conclusions
• Bullying is a strong predictor of later offending and
violence, but a weaker predictor of later drug use
• Victimization is a strong predictor of later depression
• After controlling for earlier risk factors, so can’t be
explained by these
Key Questions:
• Does bullying perpetration cause an increase in the
probability of later offending and violence?
• Or is bullying perpetration a ‘stepping stone’ on a
developmental sequence leading to offending and
violence?
• Equivalent questions for victimization vs depression …
Policy Implications
• Programmes that reduce bullying perpetration are likely
to be followed by a reduction in crime and violence
• Programmes that reduce bullying victimization are likely
to be followed by a reduction in depression
• Bullying prevention might be regarded as a method of
prevention of crime and internalizing problems and a
method of promoting health
• Our review indicates the urgent need to implement high
quality anti-bullying programmes (see earlier review)
• Need to investigate protective factors that
interrupt the continuity from school bullying to
later adverse outcomes => a new special issue
coming out on this!!!
Research Support:
For our review on health/criminal outcomes of school
bullying we are again most grateful to:
Swedish National Council for Crime Prevention
www.bra.se
For more information:
Maria M. Ttofi: [email protected]
David P. Farrington: [email protected]
Friedrich Losel: [email protected]