Aggressive behavior at school is a complex phenomenon, and bullying is probably the most common expression of it. Bullying is a major health problem in schools worldwide (National Academies of Sciences, Engeering, and Medicine, 2016). The most consensual definition of bullying includes the following elements (Farrington, 1993; Smith et al., 1999; Olweus, 2013; Menesini and Salmivalli, 2017): (i) Intentionality, i.e., the behavior has the intention to cause harm to another; (ii) Imbalance of power, i.e., a situation of unequal psychological or physical power, in favor of the bully; and (iii) Repetitiveness, i.e., a repeated action over time. Even though there is a general agreement in these elements of the definition, it is important to highlight some disagreements. For example, the Centers for Disease Control (CDC) explicitly excludes from the definition aggressive behaviors that occur between siblings, dating partners, or committed by adults, even if the acts take place in the school context.

The prevalence of bullying in schools varies from study to study, depending on the instruments used to measure it. Seminal research was conducted by Dan Olweus, who carried out an investigation including 750 representative schools (around 130,000 students) in Norway. This study showed that almost 15% of the students were involved in bullying (Olweus, 1993). Since then, many studies have been conducted, but mainly in high-income countries (Zych et al., 2015). A recent review across 80 studies found a mean prevalence rate of 35% for traditional bullying involvement and 15% for cyber bullying involvement (Modecki et al., 2014). In another large, cross-country comparison study, important differences were found between eastern and western countries (Sittichai and Smith, 2015), likely related to how students understand the meaning of bullying, and linguistic and cultural influences on its conceptualization (Menesini and Salmivalli, 2017). Overall, we can consider that between 10 and 30% of children and adolescents directly participate as bullies, victims, or bully-victims (Smith et al., 2002; Menesini and Salmivalli, 2017).

However, as was mentioned earlier, most bullying research has been conducted in developed countries (Zych et al., 2015), with more than three quarters of the most cited articles in bullying being published in North America and Northern Europe. Considering that some reports already show that violence and bullying are higher in less developed countries, especially in Latin America (Muula et al., 2009; Oliveros et al., 2009; Brito and Oliveira, 2013; Cassiani-Miranda et al., 2014; Lister et al., 2015), a future challenge within the field is to promote research in areas outside of North America and Northern Europe (Zych et al., 2015).

Bullying research is relatively new in Latin America (Wolf and Esteffan, 2008). In this context, the situation in Chile is not very different from the rest of the countries of the region. The first national study on school climate and school violence in Chile was conducted in 2005 (Ministerio De Salud, 2005). The results showed that 28% of students from 7 to 11th grades reported that they had been frequently insulted at school, compared with 9% who reported being frequently harmed. In addition, the results showed that such experiences were more prevalent in schools attended by students of lower socio-economic status, and more common among boys than girls. In 2009, the Chilean Government conducted another study to describe violent behaviors within school contexts. The results showed that 23.3% of students from the 7 to 12th grade reported having been a victim of aggressive behavior in schools and in 14.5%, this aggressiveness was repeated over time. Additionally, 28% of students admitted conducting an aggressive act toward others in school. Furthermore, this study reported an increase in more complex and severe aggressions, such us sexual abuse, weapon injuries, robberies, and thefts (Ministerio Del Interior, 2009). More recently, the Ministry of Education conducted a survey of 8th grade students in 5,855 schools. One out of five surveyed declared that harassment and threats were very frequent among school peers, 10% reported being victims of different types of aggression, and one fourth declared that they experienced this violence daily (Ministerio De Educacion, 2012).

There are typically at least three groups of students involved in bullying: bullies, victims, and bystanders (Salmivalli et al., 2011). Bullies are those students who play the leading role and commence the bullying. Victims are those who suffer the harassment from bullies. Some bullies may be also be victimized by other students; they are so-called bully-victims. Finally, bystanders are those students who witness the bullying. Studies have shown that ~85% of bullying episodes are witnessed by peers (Lynn Hawkins et al., 2001; Pepler et al., 2010). It is important to note that within each of these main groups, there are different types of students (National Academies of Sciences, Engeering, and Medicine, 2016).

Several factors associated with bullying have been reviewed by other authors (Olweus, 2013; National Academies of Sciences, Engeering, and Medicine, 2016; Menesini and Salmivalli, 2017). For the purposes of this study, we will focus on studies exploring the association between bullying experience and substance use.

Several studies have found an association between bullying behaviors and substance use among adolescents (Kaltiala-Heino et al., 2000; Haynie et al., 2001; Morris et al., 2006; Liang et al., 2007; Carlyle and Steinman, 2007; Wolf and Esteffan, 2008; Tharp-Taylor et al., 2009; Luk et al., 2010, 2012; Centers for Disease Control Prevention, 2011; Vieno et al., 2011; Radliff et al., 2012; Gámez-Guadix et al., 2013; Espelage et al., 2014). Research has found that there is a strong association between tobacco use and bullies or bully-victims (Morris et al., 2006; Vieno et al., 2011), or victims (Tharp-Taylor et al., 2009). In a similar manner, alcohol use is strongly associated with those students having aggressive behaviors, such as bullies and bully-victims (Kaltiala-Heino et al., 2000; Vieno et al., 2011), and other studies have found that being a victim is an important risk factor for alcohol consumption (Sullivan et al., 2006; Tharp-Taylor et al., 2009; Luk et al., 2010; Vieno et al., 2011; Radliff et al., 2012). Finally, cannabis use is strongly associated with being a bully or a bully-victim (Radliff et al., 2012). Moreover, longitudinal studies have also shown the association between bullying and substance use. For instance, self-reported bullying at Grade 5 (when students are 11 years-old) predicts violent behavior and problematic consumption of alcohol and drugs 6 years later (Kim et al., 2011), with similar results reported in another study (Niemelä et al., 2011). However, there is less information about these associations and participation as bystander in the bullying experience (Rivers et al., 2009; Durand et al., 2013).

Considering the hierarchical structure of school-based data (students being nested within classes nested within schools) multilevel modeling is the most appropriate approach to analyze it. However, few studies have explored these associations using multi-level modeling. For instance, Bradshaw et al. (2013) developed three-level hierarchical linear models with the purpose of exploring the association between the different roles involved in bullying behavior (victim, bully, bully-victim, and no involvement) and three health-risk behaviors: violence, substance use (alcohol, cigarette, cannabis, and prescription drug), and academic problems. In general, the results showed that all students involved in bullying behavior were more likely to consume different kinds of substances. The group at greatest risk consisted of bully-victims, followed by bullies and victims, respectively.

Much less information can be found in Latin American countries (Rodríguez et al., 2006; Andrade et al., 2012; Córdova Alcaráz et al., 2012; Cassiani-Miranda et al., 2014). In the 2009 National School Health Survey in Brazil, results showed that both boys and girls who were bullies or victims were at a heightened risk of consuming illegal drugs and alcohol (Andrade et al., 2012). Similarly, in Peru, the Second National Study of Drug Consumption and Prevention reported that being a victim of bullying was associated with the abuse of legal, illegal, and medical drugs (Romaní and Gutiérrez, 2010).

The explanation of this association needs further investigation. Research has shown that bullies may be involved in drug use as part of a more general involvement in antisocial behavior, or as a byproduct of their engagement with groups of peers presenting deviant behaviors (Ttofi and Farrington, 2011). Fighting or verbal and relational aggression, seem to be part of a cluster of problem behaviors, insofar as substance use and bullying show reciprocal relations (Espelage et al., 2014). Several studies show that this association is stronger among bullies than victims (Kaltiala-Heino et al., 2000; Nansel et al., 2001; Carlyle and Steinman, 2007; Radliff et al., 2012; Gámez-Guadix et al., 2013). For instance, data from the 2009 Massachusetts Youth Health Survey suggests that middle school students who are bullies or bully-victims are three to six times more likely to admit substance use than their victims or those not involved in bullying (Centers for Disease Control Prevention, 2011). It is possible that victims may use substances as a coping mechanism to deal with the anxiety produced by the attacks and rejection by their peers (Carlyle and Steinman, 2007; Continente et al., 2010; Romaní and Gutiérrez, 2010). Even though, much less information is available exploring the association between being a bystander (with no direct involvement in bullying actions) and substance use (Rivers et al., 2009), it is possible to hypothesize that these students may have a higher risk of substance use when compared to students with no involvement. A recent study that explored the effect of adolescent exposure to violence (witnessing or awareness of) on subsequent adult illicit drug use found that witnessing parental violence and exposure to neighborhood violence during adolescence was significantly associated with adult drug use (Menard et al., 2015). This study also suggested that the exposure to violence in adolescence is especially hazardous because this is the period of life when illicit drug use is typically initiated (Elliott et al., 1989). In support of this, the general strain theory (Agnew, 1985) proposes that strains are present when individuals are exposed to noxious stimuli, which may include witnessing of violence. This strain may lead to different modes of adaptation, such as engagement in deviant behavior or delinquency (Agnew, 1985; Broidy, 2001), or using drugs as a form of self-medication (Harrison et al., 1997). Several studies have shown that children and adolescents exposed to violence often experience anxiety and social concerns (Groves, 1999; Kitzmann et al., 2003; Meltzer et al., 2009), which in turn may lead them to use drugs to reduce those symptoms.

Finally, substance use in Chile is an important public health problem. In the last national study surveying students from 8 to 12th grade, 22.9% reported any alcohol consumption in the last month and 8.4% reported tobacco use every day (SENDA, 2015). In relation to cannabis, cocaine, and cocaine base (Pasta base), the monthly prevalence in 2015 was found to be 13.5, 1.6, and 0.3%, respectively. Additionally, there has been an increase in the prevalence reported since the previous survey (2012) at least for alcohol and cannabis use (SENDA, 2015). Therefore, exploring the factors associated to substance misuse, will provide valuable information in the implementation of preventive measures, as there is already some evidence showing that anti-bullying preventive programs may also reduce substance use among adolescents (Olweus, 1993; Durand et al., 2013). Several individual, family, school, peer, and community factors have been associated with substance use among adolescents (Tyas and Pederson, 1998; Monasterio, 2014). For example, students with behavioral problems have a higher risk for drinking and cannabis use, and peer smoking and drinking have also been associated with the use of multiple drugs (Harakeh et al., 2012; Tomczyk et al., 2015). Additionally, the children of parents who provide support and clear limits and norms are less likely to drink (Berge et al., 2016). Similarly, poor parental monitoring is associated with tobacco, alcohol and cannabis use (Gaete and Araya, 2017). Regarding school factors, poor school bonding increases the likelihood of substance use (Gaete and Araya, 2017), and strong anti-tobacco polices at school have been associated with less smoking (Wiium et al., 2011; Galan et al., 2012; Paek et al., 2013). Therefore, adjusting for these variables in the analyses of the association between bullying experience and substance use seems a reasonable approach.”

Therefore, this study aimed to explore the association between different degrees of bullying involvement and alcohol, tobacco, and cannabis use at the individual level, taking into account personal (e.g., school membership), family (parental monitoring), and community factors (perception of violence in the neighborhood) at the individual level, and contextual factors (e.g., school type) at the school level. To our knowledge, this is the first multilevel study in Latin-America exploring this association, and this is one of the few studies exploring these associations in the different groups that play a role in bullying. Specifically, individuals with no experience of bullying, bystander only, victim only, bully only, and bully-victim. Our general hypothesis is that the exposure to bullying in adolescence, as perpetrator, victims or as bystander, will increase the risk for substance use.