School bullying as a predictor of violence later in life: A systematic review and meta-analysis of prospective longitudinal studies
Highlights
► A systematic review and meta-analysis of prospective longitudinal studies. ► Bullying perpetration increased the risk of later violence by about two thirds. ► Bullying victimization increased the risk of later violence by about one-third. ► High quality bullying prevention programs should be promoted. ► They could be seen as an indirect form of preventing adult violence.
Introduction
Bullying continues to be a serious problem plaguing school youth in both developed and developing countries (Liang et al., 2007, Smith et al., 1999). Early longitudinal research highlighted the negative impact of school bullying on children's internalizing (e.g. depression) and externalizing (e.g., offending) behavior (e.g., Farrington, 1993, Olweus, 1991) and various reviews have synthesized research on the topic (e.g., Ttofi & Farrington, 2008).
Following a strong scientific interest in school bullying research in Scandinavian, and subsequently, other European countries, bullying has gradually become a topic of general public concern and has drawn a lot of media attention, with articles in major newspapers and magazines reporting cases of children who committed (or attempted) suicide because of severe bullying victimization at school, and parents suing school authorities for their failure to protect their offspring from continued bullying victimization (e.g., BBC Online, 2010, Boston Globe, 2010, Daily Mail, 2009; see relevant web links at the end of references).
Two special issues of journals have been organized on this topic based on our project on ‘Health and criminal outcomes of children involved in school bullying’ (Farrington et al., 2011, Ttofi et al., 2011). Within this project, peer-reviewed papers with longitudinal data analyses are being analyzed to study the association between school bullying (perpetration and victimization) and various internalizing (e.g., anxiety, depression) and externalizing (e.g., delinquency, violence) problems. The two special issues followed the activities of a two-year international research network, in which principal investigators and research groups of 29 longitudinal studies participated by providing unpublished data in line with the aims of our project (see Farrington, Losel, Ttofi, Loeber, & Theodorakis, 2012, Table 4 for a list of participants). Both special issues of journals offered up-to-date scientific evidence on the topic and two meta-analyses were published on the efficiency of school bullying in predicting depression (Ttofi, Farrington, Lösel, & Loeber, 2011a) and offending (Ttofi, Farrington, Lösel, & Loeber, 2011b) later in life.
In an update of the aforementioned systematic reviews (Farrington et al., 2012), it was found that bullying perpetration at school was a significant predictor of offending up to about six years later (range in years: 0.42–16.50; M = 5.84; SD = 4.56), even after controlling for other major childhood risk factors (Adjusted OR: 1.89; 95% CI: 1.60–2.23). Interestingly, school bullies also had a significantly higher probability of being depressed later in life (range in years: 0.42–24.00; M = 6.17; SD = 6.67) compared with non-involved children (Adjusted OR: 1.41; 95% CI: 1.22–1.64).
Results were equally disheartening for the victims of school bullying. Farrington et al. (2012) found that bullying victimization was a significant predictor of depression up to about seven years later (range in years: 1.00–36.00; M = 7.13; SD = 8.79), even after controlling for other major childhood risk factors (Adjusted OR: 1.71; 95% CI: 1.49–1.96). Victims of school bullying also had a significantly higher probability of being involved in offending (Adjusted OR: 1.14, 95% CI: 1.00–1.31) later in life (range in years: 0.42–16.50; M = 5.55; SD = 4.85) although, admittedly, the magnitude of the summary effect size was quite small.
Longitudinal studies have shown that adult violent criminals frequently have school records of bullying and other forms of aggressive behavior (Luukkonen, Riala, Hakko, & Rasanen, 2011), suggesting the intra-generational continuity of externalizing behavior. Prospective studies have also pointed out the inter-generational continuity of school bullying. In the Cambridge Study in Delinquent Development, for example, the bullies at age 14 tended, at age 32, to have children who were also bullies (Farrington, 1993). No previous systematic review, however, has ever been conducted to calculate an unbiased standardized effect size on the association of school bullying with aggression and violence in adult life.
This paper addresses this gap in research literature and aims to examine whether bullying perpetration at school can significantly predict aggression and violence later in life. Given the overlap between violence victims and violent offenders (Singer, 1986), this research will focus not only on bullying perpetration but also on bullying victimization. Notably, Rivara, Shepherd, Farrington, Richmond, and Cannon (1995) found that young males treated in an Accident and Emergency Department for assault-related injuries were more likely to have a criminal record than young males treated for accidental and sporting injuries. Following this line of research, within this manuscript it is investigated whether victims of school bullying are also significantly more likely to be involved in violence later in life. We focus on prospective longitudinal studies because of the scientific superiority of longitudinal research for the holistic understanding of antisocial and criminal behavior (Blumstein, Cohen, Roth, & Visher, 1986).
A significant association between school bullying and violence later in life would have many implications both at a theoretical and a practical level. On the one hand, significant results would support the argument of a persistent underlying violent tendency (i.e., specific continuity of violent behavior) and would, therefore, highlight the importance of early intervention research in targeting school aggressive behavior at root before it develops into more serious forms of aggression and violence later in life. If there is, indeed, a specific and persistent continuity of aggressive and violent behavior, then special attention should be paid in designing early intervention programs specifically for aggressive and violent individuals. The argument in favor of a persistent underlying violent tendency can be supported if the summary effect size for bullying perpetration versus violence is substantially larger than the summary effect size that was found in our pervious meta-analyses on bullying perpetration versus offending.
On the other hand however, and given the efficacy of school bullying in significantly predicting delinquency and offending behavior later in life (Farrington et al., 2012, Ttofi et al., 2011b), significant results arising from the current systematic review would possibly favor the argument of a more general underlying antisocial tendency. The argument of a more general underlying antisocial tendency can be supported only if the summary effect size for bullying perpetration versus offending is more or less of the same magnitude as the summary effect size for bullying perpetration versus violence. In this case, it could be suggested that anti-bullying programs could be seen as an indirect form of both crime and violence prevention. If research directs towards a more general underlying antisocial tendency, it would then also be reasonable to suggest that both anti-bullying and more general multiple-component programs – that address both bullying and other externalizing problems – may be equally beneficial in interrupting the continuity from school bullying to violence and offending in adult life.
Existing research supports the comorbidity of aggressive, violent and other externalizing problem behaviors and the marked shared variance in risk factors predicting these behaviors (Farrington, 2002). With regard to the argument of comorbidity of externalizing problem behaviors, based on analyses from three independent samples in the Pittsburgh Youth Study, Loeber, Farrington, Stouthamer-Loeber, and van Kammen (1998) showed that aggression, delinquency, conduct problems, and various other challenging behaviors were significantly intercorrelated, a finding which is in line with our own research on bullying perpetration at school predicting delinquency and offending later in life (Farrington et al., 2012, Ttofi et al., 2011b).
With regard to the argument of the marked shared variance in risk factors predicting these externalizing behaviors, the same study (Loeber et al., 1998) also showed that these problem behaviors tended to share many risk factors. Specifically, they discovered that, of the significant risk factors for physical aggression, 84% were also significant risk factors for delinquency in the youngest sample, 82% were significant for delinquency in the middle sample, and 68% were significant for delinquency in the oldest sample.
Within the body of school research, a number of studies suggest that school bullying is also highly correlated with other conduct problems such as violent and oppositional behavior. For example, a representative study of 15,686 American students in grades 6 through 10 in public and private schools (Nansel, Overpeck, Haynie, Ruan, & Scheidt, 2003) found that both perpetrators and victims of bullying behavior had a higher probability of weapon carrying compared with non-involved children, not just in school (adjusted ORs of 2.6 and 1.5 accordingly), but also away from school (adjusted ORs of 5.9 and 4.1 respectively). Another study (Kokkinos & Panayiotou, 2004) found that CD symptomatology was predictive of bullying perpetration whereas ODD was predictive of bullying victimization.
Notably, school bullying shares many risk factors with offending (Lösel & Bliesener, 2003), behavioral conduct (Boulton, Smith, & Cowie, 2010), and other externalizing behaviors, such as hyperactivity. For instance, longitudinal data analyses of the Cambridge Study in Delinquent Development show that low non-verbal intelligence (OR: 2.0), low attainment (OR: 1.8), hyperactivity (OR: 2.1), and having a convicted parent (OR: 1.7) at ages 8–10 significantly predicted bullying perpetration at age 14 (Farrington & Baldry, 2010). In the same study, bullying at age 14 significantly predicted antisocial behavior at age 18 (OR: 2.2).
To conclude, the issue of whether school bullying and other externalizing behaviors later in life are different age- and context-related manifestations of the same underlying antisocial dispositions has many implications at both the theoretical and practical level. Previous research has not given an answer to this matter and it is hoped that this paper will address adequately this issue.
Another important question is whether involvement in bullying influences the individual in such a way as to engender more aggressive and violent behavior later in life. Establishing a causal long-term link between school bullying and violence in adult life, or establishing that bullying is in some way a ‘stepping stone’ towards violence, is a very demanding task and beyond the aims of the current research. A similar question is central in an ongoing debate within the field of criminology regarding the explanation of the strong correlation between prior and future criminal activity.
Some researchers have argued in favor of the theoretical model of ‘state dependence’, suggesting that the experience of crime materially influences the individual in such a way as to engender more crime in the future because of its undesirable effect on social bonds, conventional attachments and so forth. Other researchers have argued in favor of ‘population heterogeneity’, suggesting that the correlation between past and future criminal behavior could be attributed to the persistence of time-stable individual differences in an anti-social characteristic between persons which can take various forms across life. Finally, others have suggested a combination of the above two explanations (for a summary of research, see Nagin & Paternoster, 2000).
Transferring the above theoretical model into the area of research on school aggression, one may ask the following question: Could school bullying have a causal effect on later violence (i.e., ‘state dependence’ argument) because, for example, involvement in bullying results in poorer school bonding, which in turn facilitates the increasing intensity of aggressive behaviors? Or is it the case that a strong association between school bullying and violence later in life could merely reflect the persistence of an underlying anti-social tendency (i.e., ‘population heterogeneity’ argument)? Although this matter is beyond the aims of the current research, it is of great importance and it is essential to articulate.
To date, systematic reviews in criminology (and also in other fields) have focused primarily on intervention research, failing to address naturally occurring (i.e., non-manipulated) causes of delinquency or other externalizing and internalizing problems. Examples of naturally occurring/non-manipulated causes of internalizing and externalizing problems could include divorce, brain damage, parental incarceration or, with reference to the current research, school bullying (Murray, Farrington, & Eisner, 2009).
In risk factor research, many naturally occurring probabilistic events (such as broken homes or getting married) cannot be randomly assigned due to ethical or other considerations (Petrosino, 2003). Even though children are not allocated to victim versus non-victim status in a planned manner, bullying victimization can be regarded as a ‘social intervention’ with specific outcomes. Our analyses will also include bullying perpetration.
Systematic reviews on risk factors could advance theory and also help to develop effective prevention programs (Murray et al., 2009). For example, it would be interesting to examine whether victims of bullying suffer from low self-esteem or whether school bullies lack empathy. Such findings, based on relevant systematic reviews, could guide future intervention initiatives, while also refining theories about possible causes of bullying behavior.
Establishing a significant long-term link between school bullying and violence in adult life could potentially have important implications for policy and practice. It would give a stronger voice to anti-bullying agencies and would also re-establish the moral imperative of school communities to create an appropriate violence-free school climate for all youngsters.
Section snippets
Objectives and main questions to be addressed
Our objective is to conduct a systematic review and calculate standardized effect sizes with the final aim of:
- •
Establishing whether there is indeed an association between bullying at school (perpetration and victimization) and aggressive or violent behavior later in life.
- •
Establishing the unique contribution (i.e., after controlling for covariates) of school bullying across time (predictive efficacy).
- •
Establishing what covariates (e.g., age at which bullying was measured; length of follow-up
Bullying perpetration at school and violence later in life: unadjusted and adjusted effect sizes
A total of 15 studies presented data on the association of bullying perpetration with aggression and violence later in life. The unadjusted summary effect size across these studies was OR = 3.09 (95% CI: 2.35–4.07; z = 8.10). For one study (i.e., Swedish Community Samples Study), only an unadjusted effect size was available. The unadjusted effect size for the remaining 14 studies was OR = 2.97 (95% CI: 2.25–3.92; z = 7.71). All individual studies yielded a significant effect size as shown in the forest
Discussion and directions for future research
The present meta-analyses have found a significant link between school bullying (perpetration and victimization) and aggressive/violent behavior later in life. Notably, these associations remained significant even after controlling for other major childhood risk factors. Therefore, it can be suggested that school bullying is a risk factor with a unique contribution to later violence, although this does not necessarily imply any causal or ‘stepping stone’ relationship between bullying and later
Acknowledgments
This study was supported by the British Academy (Small Research Grant Scheme) and the Swedish National Council for Crime Prevention. We would like to thank all longitudinal researchers who have conducted special analyses for our wider systematic review for the Campbell Collaboration. Finally, we are most appreciative of the valuable methodological advice we received from Professor David B. Wilson.
References (83)
- et al.
Joint development of bullying and victimization in adolescence: Relations to delinquency and self-harm
Journal of the American Academy of Child and Adolescent Psychiatry
(2008) - et al.
School neighborhood, and family factors are associated with children's bullying involvement: A nationally representative longitudinal study
Journal of the American Academy of Child and Adolescent Psychiatry
(2009) - et al.
Bullying perpetration and subsequent sexual violence perpetration among middle school students
Journal of Adolescent Health
(2012) - et al.
Bullying, violence, and risk behavior in South African school students
Child Abuse & Neglect
(2007) - et al.
Bullying behaviour and criminality: A population-based follow-up study of adolescent psychiatric inpatients in Northern Finland
Forensic Science International
(2011) A longitudinal study of heterosexual relationships, aggression, and sexual harassment during the transition from primary school through middle school
Applied Developmental Psychology
(2001)- et al.
Victim as offender in youth violence
Annals of Emergency Medicine
(1995) - et al.
Childhood predictors of male criminality: A prospective population-based follow-up study from age 8 to late adolescence
Journal of the American Academy of Child and Adolescent Psychiatry
(2006) - *Arseneault, L. (2011). Adjusted effect sizes for the E-Risk Longitudinal Study. Email communication, January 21,...
- *Azzuzi, S., & Killias, M. (2010). Long-term outcomes of active and passive bullying during childhood: Suicide...