Abstract
Over the last decade, cyberbullying has emerged as a public health concern among young people. Cyberbullying refers to intentional harmful behaviors and communication carried out repeatedly using electronic media. Considerable research has demonstrated the detrimental and long-lasting effects of cyberbullying involvement. This article draws on a social–ecological perspective to identify protective health assets from across the multiple environmental domains of the adolescent that may mitigate against experiencing cyberbullying. Data were collected from 5,335 students aged 11, 13, and 15 years who participated in the 2014 World Health Organization Health Behaviour in School-aged Children Study for England. Protective health assets were identified at the family (family communication), school (school sense of belonging and teacher support), and neighborhood (neighborhood sense of belonging) levels. In particular, the findings draw attention to the protective role fathers can play in supporting young people.
Introduction
Bullying is widely acknowledged as a public health concern, with cross-national analysis identifying that one in three young people were victimized in the past 2 months (Chester et al., 2015). While variation in definitions exist, bullying is commonly defined as an individual or group of individuals intentionally inflicting harm, repeatedly and over time, against someone who is unable to defend themselves (Olweus, 1993). Bullying behaviors can be physical, verbal, relational, or cyber in nature. Longitudinal studies have demonstrated the detrimental effects of bullying on both physical and psychological health, as well as social outcomes including school attainment (Kowalski & Limber, 2013; Zwierzynska, Wolke, & Lereya, 2013). Moving from an individual behaviorist model, more recently, the social–ecological model has provided a valuable framework for the study of traditional forms of bullying (physical, verbal, and relational), acknowledging that bullying is a complex social phenomenon that is cultivated or inhibited by the environment (Espelage, 2014). However, less research has examined cyberbullying within the framework of the social–ecological model.
Cyberbullying
The current generation of young people inhabits a virtual world that spans the domains of adolescent life. Moreover, the development of smart phones has increased accessibility to the Internet for young people, allowing online activity to shift from being primarily home based to openly available in public spaces; every week nearly half of young people in the United Kingdom access the Internet outside of the home (Livingstone, Mascheroni, Ólafsson, & Haddon, 2014).
With young people conducting a significant amount of their social interaction in virtual environments (Brooks, Magnusson, Klemera, Spencer, & Morgan, 2011), it is unsurprising that negative forms of interaction and communication are also being played out online. Cyberbullying, the online aspect of bullying, can take many different forms including sending abusive or threatening messages, uploading embarrassing photographs, sharing personal information, or exclusion from online groups. With ongoing technological developments, the nature of cyberbullying is likely to be in flux, constantly evolving and changing, including both the platforms and methods adopted.
To date, reports of cyberbullying prevalence have varied; a recent systematic review identified lows of 3% and highs of 72% for cyberbullying victimization in the United States (Selkie, Fales, & Moreno, 2015). The variation can be attributed in part to differences in operationalising and defining cyberbullying (see Kowalski, Giumetti, Schroeder, & Lattanner, 2014, for an extensive record of research definitions). The notion of intent to cause harm via electronic means is widely accepted, but the concepts of repetition and a power imbalance underpinning traditional forms of bullying have been queried in relation to the virtual world; for example, when a single post can be viewed multiple times, and additionally shared by other individuals, it is difficult to quantify repetition (Smith, 2012; Waasdorp & Bradshaw, 2015). Moreover, the concept of a power imbalance differs in a virtual world where physical or social strength is less apparent (Dooley, Pyżalski, & Cross, 2009; Smith, 2012). The varying reference periods, for example, lifetime, past 12 months, or past month, also contribute to the ambiguity of cyberbullying prevalence rates (Kowalski et al., 2014), along with differing measurement approaches, for example, behavioral check lists versus cyberbullying definitions (Modecki, Minchin, Harbaugh, Guerra, & Runions, 2014). Although it is difficult to ascertain the true extent of cyberbullying, a recent cross-national study found 21.4% of 14- to 17-year-old respondents had been a victim of cyberbullying in the previous year (Tsitsika et al., 2015).
As with the more traditional forms of bullying, research has demonstrated the detrimental effect of cyberbullying on health and well-being. Studies to date have explored the emotional well-being implications, including depression (Bauman, Toomey, & Walker, 2013; Wang, Nansel, & Iannotti, 2011), anxiety (Juvonen & Gross, 2008; Rose & Tynes, 2015), loneliness (Olenik-Shemesh, Heiman, & Eden, 2012), and suicidal ideation (van Geel, Vedder, & Tanilon, 2014). Moreover, longitudinal studies have demonstrated the causal nature of these relationships (Gámez-Guadix, Orue, Smith, & Calvete, 2013; Rose & Tynes, 2015). The consequences also extend beyond victims; being a cyberbully is associated with lower quality of life, increased psychological difficulties, and suicide attempts (Bauman et al., 2013; Fletcher et al., 2014).
Studies have begun examining whether cyberbullying is an extension of the more traditional forms of bullying or functionally different (Law, Shapka, Hymel, Olson, & Waterhouse, 2012), and this extends to the relative consequences of each form. While the true possibility to match bullying behaviors online and offline has been called into question (Bauman & Newman, 2013), two large-scale studies both identified that victims of cyberbullying had increased odds of internalizing and externalizing symptoms compared with victims of traditional bullying alone (Schneider, O’Donnell, Stueve, & Coulter, 2012; Waasdorp & Bradshaw, 2015). It is important to acknowledge that there is considerable overlap between cyberbullying and traditional bullying. Victims of bullying are likely to be subjected to a number of different bullying behaviors: Schneider et al. (2012) found that 60% of young people that had been cyberbullied also experienced bullying of a traditional form. Overall, those who experience both cyberbullying and traditional bullying appear to have the worst health outcomes when compared with young people who experience only either cyber or traditional bullying (Schneider et al., 2012), indicating that cyberbullying has a unique effect on top of the impact of just traditional bullying (Bonanno & Hymel, 2013).
It has been speculated that the experience of cyberbullying may be more traumatic and result in greater harm due to contextual differences between the two forms of bullying, most notably issues relating to time and place. Unlike traditional bullying behaviors, which tend to occur primarily in the school environment, cyberbullying can be experienced in any context where the victim is accessing electronic media (Patchin & Hinduja, 2006; Slonje & Smith, 2008). With more than 80% of young people aged 12 to 15 years in the United Kingdom possessing a mobile phone (Ofcom, 2013), exposure to virtual communication and social interactions is ever present and largely unavoidable. Cyberbullying has also been distinguished from more traditional forms of bullying due to the breadth of the audience, as the bystanders of cyberbullying often outnumber those of traditional bullying (Kowalski et al., 2014). It is thought the effects of cyberbullying may be heightened due to the anonymity of the bully; not only may the victim feel helpless not knowing the perpetrator but also the sense of anonymity can create a disinhibition effect among the perpetrator resulting in increased hostility and reduced empathy (Aboujaoude, Savage, Starcevic, & Salame, 2015). Removing online material can be difficult and can result in the victim being exposed to cyberbullying repeatedly, this permanent nature is a distinctive and unique feature of virtual interaction, where instances are recorded and stored online (van Geel et al., 2014).
Social–Ecological Framework
A number of scholars have advocated the use of the social–ecological theory for advancing current understanding of school bullying (Espelage, 2014; Swearer & Hymel, 2015). A social–ecological perspective situates the development of young people in their social context, acknowledging the bi-directional interaction between an individual and the multiple domains in which they inhabit (Swearer & Hymel, 2015). The traditional ecological model of development proposed by Bronfenbrenner (1977) contains five elements: the individual, the micro-, meso-, exo-, and macro systems. The individual is placed at the center of the model, interacting with and shaped by the different ecological systems as opposed to just individual character traits. The microsystem describes the immediate setting with which the individual has direct contact including school and family. The mesosystem describes interactions between elements of microsystem such as school and parents; the exosystem is an extension of the mesosystem which contain interactions in which the individual is not an active participant. The overarching level, macrosystem, describes the broader societal context including culture, economy, and politics.
The social–ecological framework is not unique to young people, but refers to human development in the broader sense. However, Bronfenbrenner (1994) acknowledged the importance of the environment during early development in particular. The ecological systems are likely to evolve and shift throughout the life course; research highlights the following domains as particularly relevant to the development of young people.
Family
The family is a fundamental microsystem in which young people’s primary development and socialization is fostered. Research has demonstrated family structure and dynamics within the family, particularly parent–child communication, as important influences on young people’s health and well-being (Moreno et al., 2009) and engagement with risk behaviors (Bell, Forthun, & Sun, 2015; Brooks, Magnusson, Spencer, & Morgan, 2012).
Friends
Traditional perspectives assume friends become of greater relevance during adolescence while the influence of the family diminishes. Subsequent theories, for example, the continuity/cognitive model, describe the complementary role friends play in young people’s lives (Cooper & Cooper, 1992). Friendships have been established as particularly salient for adolescent identity development (Heaven, 1994).
School
With young people spending a substantial amount of time at school, the school environment is an integral part of young people’s lives. Young people acquire knowledge and life skills at school that will affect later life chances in adulthood and also encourage identity development and socialization (Eccles & Roeser, 2011). Furthermore, student’s perception of the school environment, including feelings of belonging and teacher connectedness, has been associated with young people’s health and well-being (Fenton, Brooks, Spencer, & Morgan, 2010; García-Moya, Suominen, & Moreno, 2014).
nNeighborhood
The local community has received less research interest compared with other domains of young people’s lives; however, the neighborhood has been identified as an important exosystem for young people’s development (Morrow, 2001, 2003). Young people who feel included and a sense of belong in their local community are less likely to engage in risk behaviors (Brooks et al., 2012).
The social–ecological perspective offers a potentially useful framework for the exploration of bullying behaviors as bullying is constructed and enacted via a complex interplay between individuals and their immediate and distant ecologies (Hong & Espelage, 2012). Evidence demonstrates the influence of the environment on young people’s behavior, with bullying involvement as either perpetrator, victim, or bystander varying across time, space, and context (Swearer & Hymel, 2015). Perceiving bullying as a result of complex interactions between young people and the different environments allows for identification of elements that foster a vulnerability to either bullying victimization or perpetration. Risk factors have been identified from across the ecologies, but most notably, at the individual level including gender, poor health status, and anti-social personality traits (Swearer & Hymel, 2015), and at the micro level including negative family interactions (Lee, 2011), peer influence (Hong & Espelage, 2012), and an unsupportive school environment (Barboza et al., 2009). Furthermore, a number of studies have implicitly identified risk factors from different ecologies of the adolescent world without explicitly framing the research in social–ecological theory; for example, poor teacher support and class management have been associated with an increased risk of bullying (Azeredo, Rinaldi, de Moraes, Levy, & Menezes, 2015).
The social–ecological framework can also be used to identify assets that are protective against experiencing bullying. An assets model not only considers how protective health assets are located as internal to the individual but also how resources located around the young person and in their environment work to protect young people’s health and well-being and enhance capacities and capabilities (Morgan & Ziglio, 2007). An assets model suggests that there is a fundamental dynamic interaction between ecological factors in the environment of the young person and internal positive attributes. The identification of assets that protects against bullying has seen less attention than the mapping of risk factors; however, recent research has highlighted the protective role parents and the family environment can play in preventing bullying (Boel-Studt & Renner, 2013; Sapouna & Wolke, 2013).
The social–ecological theory has proven to be invaluable to the study of bullying and helped lead the development of interventions that extend across domains of the adolescent world (Barboza et al., 2009), yet little research has examined cyberbullying from this context. By its very nature, cyberbullying has the potential to extend beyond a victims immediate peer group, with bystanders not confined to the same class, grade, school, or country; emphasizing the importance of considering the influence of the environmental domains of the adolescent. Furthermore, Cross et al. (2015) proposed the social–ecological framework is broadened to acknowledge the online environment as an additional context that young people are interacting with and thus influenced by.
The Present Study
In the last decade, cyberbullying has become a burgeoning field of inquiry. Many articles have addressed prevalence rates and definitions, made comparisons between traditional and cyberbullying, as well as exploring the psychosocial outcomes associated with cyberbullying. Yet, despite the wealth of articles, there remain notable gaps in terms of understanding the factors that might operate protectively against being cyberbullied. Overall, relatively few studies have focused on identifying ways to address cyberbullying, highlighting the need for empirically driven interventions at the level of community, school, and family (Aboujaoude et al., 2015). The present article will examine cyberbullying utilizing a social–ecological framework, seeking to identify assets from across ecological systems that help protect young people from experiencing cyberbullying. Through consideration of what factors may be protective or mitigate against being cyberbullied we can draw practical conclusions about cyberbullying prevention among young people.
The present article draws on the English data from the World Health Organization (WHO) Health Behaviour in School-aged Children (HBSC). The HBSC study is a unique cross-sectional survey that asks young people about their social environment, providing a detailed picture of the context in which young people live (Brooks et al., 2015; Currie et al., 2012). The breadth of the HBSC study is appropriate for consideration of factors across the different levels of Bronfenbrenner’s (1977) ecological model of development. Individual traits including gender, age, and ethnicity will be considered. The scope of the HBSC data allows for careful consideration of the four microsystems surrounding adolescents previously proposed by Lee (2011): interaction with family, peer relationships, interaction with teachers and school climate. In addition, the present article will examine the lesser researched neighborhood environment; fewer studies have examined the influence of the exosystem on bullying behaviors (Hong & Espelage, 2012). Previous analysis of HBSC data identified three asset domains integral to the health and well-being of young people: sense of belonging, autonomy, and social support (Brooks et al., 2012). The present study will examine the association between these asset domains and cyberbullying across the ecological systems with which young people engage.
Although there are undoubtedly overlaps between cyberbullying and traditional forms of bullying, cyberbullying alone is the main focus of the analysis and findings presented here. Data suggest cyberbullying, and traditional bullying may differ in relation to psychosocial outcomes (Wang et al., 2011), the social demographic picture of victims is unclear (Tokunaga, 2010), and the qualitative differences between the two types are widely acknowledged (Kowalski et al., 2014). Consequently, it is feasible that protective assets may differ across the different types of bullying behaviors.
Method
Participants and Procedure
HBSC is an international WHO collaborative study that examines young people’s health and well-being, health behaviors, and their social context. The study collects information from school students aged 11, 13, and 15 years through anonymous self-completed questionnaires administered during class time. HBSC is conducted every 4 years in more than 40 countries and regions across Europe and North America, carried out by national research teams following an international protocol (Currie et al., 2014).
The present study utilized data collected from the 2014 HBSC survey carried out in England (Brooks et al., 2015). A random sample of all secondary schools in England (state and independent) stratified by region and school type was drawn to ensure representative participation. Sampling was done by replacement so that if one school declined to participate, a second matched school was contacted. In total, 48 schools were recruited, resulting in 5,335 students from 261 classes. The final sample was representative of regions and school type. The response rate at the student level was 92%. Prior to participation, students and parents received information letters and an opt-out form if they did not wish to participate. Questionnaires were administered by either a member of the research team or teachers, and students were asked to seal their completed questionnaire in an envelope to ensure confidentiality. The study gained ethics approval via the University of Hertfordshire Ethics Committee for Health and Human Sciences (HSK/SF/UH/00007).
Measures
Cyberbullying
Cyberbullying was measured via two items that asked young people how often in the past 2 months (a) “someone sent mean instant messages, wall postings, emails and text messages, or created a website that made fun of me” and (b) “someone took unflattering or inappropriate pictures of me without permission and posted them online.” Response options include “haven’t been bullied in that way,” “once or twice,” “2-3 times per month,” “once a week,” or “several times a week.” From these two variables, a single binomial variable was created indicating whether or not respondents had ever been a victim of cyberbullying (i.e., had replied “once or twice” or more often to either question). A categorical measure of cyberbullying was adopted following recent discussion highlighting the difficulty of measuring cyberbullying severity (Smith, 2012; Waasdorp & Bradshaw, 2015); for example, an online post may be seen, shared (both publically and privately) and commented on multiple times. Other forms of bullying were not included in the model as these would serve to confound the effect of the variables below on the existence of cyberbullying. However, the proportions of young people experiencing both cyber and traditional forms of bullying are reported.
Variables relating to family, school, peer, and neighborhood assets were created from related survey items. For those assets where items are being combined, Cronbach’s alpha coefficients are shown in Table 1 and in all cases are (practically) at or above the .7 rule of thumb. For those asset variables that are not created from established measurement instruments, it was not felt appropriate to use them as simple scales, and they have thus been categorized into “Low,” “Medium,” and “High” as detailed in Table 1.
Creation and Categorization of Asset Variables.
Reverse coding.
Family assets
Family communication with mother (FCM) and father (FCF) were assessed by the question “How easy it is for you to talk about things that really bother you?” measured on a 4-point scale from very easy to very difficult. Responses were collapsed into easy versus difficult. Personal autonomy in relation to family (PAF) was measured by the question “How much say do you have when you and your parents are deciding how you should spend your free time outside school?” Responses were categorized into high, medium, and low PAF (see Brooks et al., 2012, for full details). Family sense of belonging (FSB) was categorized into low, medium, and high FSB (see Table 1 for details). The Multidimensional Scale of Perceived Social Support (Zimet, Dahlem, Zimet, & Farley, 1988) measured family social support (FSS); responses to the four items concerning were averaged to provide an overall score of FSS.
School assets
School sense of belonging (SSB) and teacher social support (TSS) were both measured via three items and respondents were categorized into low, medium, and high (see Table 1 for details).
Peer assets
Peer social support (PSS) was measured with the Multidimensional Scale of Perceived Social Support (Zimet et al., 1988); responses were averaged to provide an overall score of PSS (see Table 1 for details).
Neighborhood assets
Neighborhood sense of belonging (NSB) was categorized as low, medium, and high based on seven items (see Table 1 for details).
Demographics
Gender, age, ethnicity, and socioeconomic status, as measured by the Family Affluence Scale (FAS), were all included in the present analysis. FAS is based on a set of six questions concerning material conditions of the family home (Currie et al., 2014). Responses are summed to produce a score between 0 and 13 and categorized into low (0-6), medium (7-10), and high (11-13) family affluence.
Statistical Methods
Due to the hierarchical nature of the data, multilevel modelling was undertaken using the package MLwiN (Version 2.34) via the R2MLwiN package (Version 0.8-1) in R (Version 3.2.1).
A single model was built using forward selection of main effects, enabling the demographic variables to act as controls for the school, family, neighborhood, and peer assets. Wald tests were used to judge significance. The 1% level of significance was used opposed to 5% so as to allow for the fact that multiple hypothesis tests were being conducted. The inclusion of random slopes and then interactions between main effects were then considered using the 0.1% level of significance so as to avoid the inclusion of spurious effects/interactions. At each stage, removal of terms from the model was considered.
Results
In all, 893 of 4,985 (17.9%) respondents reported being a victim of cyberbullying in the previous 2 months. Across all ages, girls were more likely to report being a victim of cyberbullying, and for both boys and girls, being a victim of cyberbullying increased with age (Table 2). Just above half (57.8%) of young people who had been cyberbullied also reported being bullied traditionally in the past 2 months.
Prevalence of Reported Cyberbullying, by Gender and Age.
A total of eight variables were retained in the final model. No random slopes or interactions entered the model. Results are given in Table 3 as odds ratios (ORs) with 95% confidence intervals (CI) and p values. Due to the number of comparisons that are being conducted, results are only discussed where statistical significance reaches the 1% level. Those comparisons with a p value of less than .01 have been highlighted in bold. The main effects contained in the model were as follows.
Odds of Being a Victim of Cyberbullying for Different Explanatory Variables.
Note. OR = odds ratio; CI = confidence interval; FAS = Family Affluence Scale; PAF = personal autonomy in relation to family; FCF = family communication with father; SSB = school sense of belonging; TSS = teacher social support; NSB = neighborhood sense of belonging.
Gender
Boys are estimated to have 44% of the odds of being a victim of cyberbullying experienced by girls.
Age
Eleven-year-olds are estimated to have approximately 68% of the odds of being a victim of cyberbullying experienced by 15-year-olds.
PAF
Those with low PAF are estimated to have approximately 68% of the odds of being a victim of cyberbullying experienced by those with high PAF. We found insufficient evidence to claim that this effect varied with age.
FCF
Those rating FCF “easy” are estimated to have 66% of the odds of being a victim of cyberbullying experienced by those who rate their communication as “difficult.”
SSB
Those with high SSB have 32% and those with medium SSB have 42% of the odds of being a victim of cyberbullying experienced by those with low SSB.
TSS
Those with high TSS have 42% and those with medium TSS have 59% of the odds of being a victim of cyberbullying experienced by those with low TSS. Those with high TSS have 71% of the odds of being a victim of cyberbullying experienced by those with medium TSS.
NSB
Those with high NSB are estimated to have 51% of the odds of being a victim of cyberbullying experienced by those with low NSB.
FAS
Those with low FAS have 54%, and those with medium FAS have 72% of the odds of being a victim of cyberbullying experienced by those with high FAS.
There was insufficient evidence to claim that schools or classes differ in the odds of pupils reporting being victims of cyberbullying (having taken into account the variables in the model). An initial model before the introduction of explanatory variables suggested that such clustering effects might exist but once these were included, the effects diminished.
Discussion
The findings of this study identify a range of potential protective health assets that may operate in protecting young people against being cyberbullied, including assets from the multiple ecologies of the adolescent world, notably, family, school, and neighborhood. The present study adds to the limited theoretical discussions surrounding cyberbullying (Dooley et al., 2009); providing support for the extension of the social–ecological framework beyond traditional bullying behaviors to encompass cyberbullying. While the current study was unable to incorporate the online world as an additional ecology (Cross et al., 2015), it is unique in examining cyberbullying among multiple ecologies simultaneously. Furthermore, the present study goes beyond prior research that has used the social–ecological framework to identify risk factors for bullying, to identify assets and protective factors from across the ecologies of young people’s lives.
Echoing earlier research (Boel-Studt & Renner, 2013; Fanti, Demetriou, & Hawa, 2012; Perren et al., 2012; Sapouna & Wolke, 2013; Wang, Iannotti, & Nansel, 2009), the findings from this study identify the family microsystem is integral to young people’s cyberbullying involvement. Although there are a number of studies that have identified parental support (Perren et al., 2012) and parental communication, including interest and knowledge regarding young people’s online activities (Cerna, Machackova, & Dedkova, 2016; Mesch, 2009), as protective against cyberbullying, there remains a paucity of studies focusing particularly on the father’s role and contribution in the protection of young people from this form of bullying.
Recent evidence has highlighted the significance of a father’s involvement in an adolescents’ life, with a strong impact on young people’s well-being, happiness, life satisfaction, and self-esteem (Allgood, Beckert, & Peterson, 2012; Cava, Buelga, & Musitu, 2014; Clair, 2012; Fenton et al., 2010; Jafari, Baharudin, & Archer, 2013). Our findings support the idea that family communication and support, particularly communication with the father, can work to protect against cyberbullying. Very few studies have considered the importance of communicating with a father for young people’s well-being, especially in relation to bullying, and our article adds weight to the significant contribution of a father figure to young people’s well-being. The current findings highlight the importance of continued investigations that focus on parental communication, specifically including fathers, as a protective health asset. Traditional perspectives on adolescent development tend to emphasize a transition from the central influence of parents to peers as young people move from early to mid-adolescence; consideration of the role of the family via assets-based analysis is challenging this position (Brooks et al., 2012; Fenton et al., 2010). The findings in this article support the increasing challenge to rather simplistic notions of peer/parent displacement and further the understanding of the significant contribution of parenting during adolescence.
In line with other research (e.g., Wang et al., 2009) our findings suggest lower family affluence is associated with less cyberbullying; this may be partially due to the more limited availability and access to electronic devices among poorer families, which reduces the potential of young people to be exposed to bullying online. Prior research has identified lower socioeconomic status as putting young people at risk of experiencing traditional forms of bullying (Elgar et al., 2013), so in essence, the results relating to family affluence are likely to be preventive in nature rather than protective.
As with earlier investigations (Dehue, Bolman, Vollink, & Pouwelse, 2012; Perren et al., 2012) the findings from our article suggest that parental supervision and control can be protective against cyberbullying: Young people whose parents were involved in decision making about leisure time and thus had lower levels of personal autonomy (PAF) were less likely to become a victim of cyberbullying than those who showed high PAF. Our article highlights the important role parents can play in monitoring and addressing cyberbullying. But does this mean that low levels of autonomy can by itself be a protective asset against cyberbullying? This contradicts an earlier study which identified that having parents who are highly protective and allow limited independence increases the risk of becoming a victim of the more traditional forms of bullying because autonomy and assertion skills are underdeveloped (Lereya, Samara, & Wolke, 2013). It could be suggested that high levels of parental involvement in decision making may result in monitoring of electronic media use through which cyberbullying is conducted, but a similar means of control is not available for traditional forms of bullying. This needs further investigation, in particular, examining whether the role of parental supervision differs across the virtual and real world.
Although it is difficult to indicate conclusions from null results, it is worth noting assets from the peer microsystem were not retained in the final model. Wang et al. (2009) found numbers of friends was protective against traditional bullying but not cyberbullying, suggesting the physical separation from friends can diminish their protective impact. Moreover, victims of cyberbullying have reported that in most cases, it was friends who were perpetrating the bullying (Waasdorp & Bradshaw, 2015). This suggests that peers are potentially less likely than others in the adolescent’s microsystem to be operating as protective assets which ameliorate the impact of cyberbullying. Furthermore, it opposes the traditional assumption that peers become more influential on the lives of young people and supports recent empirical and theoretical work that has identified the family as continuing to play a pivotal role in adolescent life (United Nations Children’s Fund, 2010).
Although traditional bullying is often confined to the school grounds and constricted by school hours, cyberbullying extends beyond the school environment (Sabella, Patchin, & Hinduja, 2013). Despite this, the present article highlights the important role that feeling connected to school and having a sense of belonging in the school community can play in protecting young people against cyberbullying. School belonging has been found to be higher in schools where pupils feel safe and where the school has taken steps to create lower levels of bullying overall (Goldweber, Waasdorp, & Bradshaw, 2013), suggesting that schools that develop a positive supportive culture and ethos may also be providing a protective function against the perpetration of cyberbullying, even if the bullying behaviors occur online and outside of the school environment. Cross-national analysis exploring the association between SSB and bullying demonstrates the relationship as consistent across countries (Freeman et al., 2009).
Teachers have been shown to play an important role in adolescent health and well-being and can potentially fulfill a compensatory role for lower family support (Brooks et al., 2012; Fenton et al., 2010; Garcia-Moya, Brooks, Morgan, & Moreno, 2014). The present findings emphasize the important role teachers can have in protecting young people from being victims of cyberbullying, with increasing levels of teacher support associated with lower chances of victimization. Positive TSS has not only been linked to students reporting that they are experiencing bullying but also seeking help for other peers who are being victimized (Eliot, Cornell, Gregory, & Fan, 2010); feasibly one of the underlying mechanisms for how TSS functions as protective. Moreover, poor teacher support has been identified as a significant predictor of the perpetration of cyberbullying (Wei, Williams, Chen, & Chang, 2010; Williams & Guerra, 2007).
Of interest, much research suggests that young people’s perception of the school environment is influenced by demographic factors (O’Brennan & Furlong, 2010), with decreases in teacher support with age (Garcia-Moya et al., 2014) and variations by gender noted (Griffith, 2000). However, the present analysis did not identify significant interactions with demographic variables which suggests the potential for these factors of being a protective health asset for boys and girls of all ages, stressing the relevance of the school microsystem for the health and well-being of young people.
The current study did not consider the location of bullying activity (e.g., whether it occurred at school or at home), and it is possible that the importance of teacher versus parental support is context specific. However, what differentiates cyberbullying from traditional forms of bullying is that context is changeable, fluid, and potentially ever present. It is often difficult to ascertain whether a bullying episode was instigated inside or outside of school, and because of the enduring nature victimization may move from the school setting, to home, and back again. This means that young people who have supportive networks across different life domains are likely to be most protected against the adverse effects of cyberbullying. As previously discussed, strong support in one domain (e.g., from teachers) may compensate for low support in another domain (Brooks et al., 2012; Fenton et al., 2010; Garcia-Moya et al., 2014).
The present article was able to contribute to the currently limited discussion of ecologies beyond the microsystems, through examination of the neighborhood. The current analysis supports previous research that identified the protective function of neighborhood and community on the health and well-being of young people (Brooks et al., 2012; Morrow, 1999). Having a strong sense of neighborhood belonging may be indicative of being part of a collective (as opposed to individualistic) community, something that has been found to correlate with lower incidence of bullying behavior (Lee, 2011). Living in a supportive, welcoming community may also therefore have an effect of reducing the incidence or drivers toward participating in, and exposure to, cyberbullying. Moreover, cyberbullying is associated with increased time spent online (Wade & Beran, 2011) and young people who spend disproportionally large amounts of time on social media and other electronic platforms, aside from being at increased risk of exposure to cyberbullying, may also feel less engaged in their communities.
Identifying individual traits of cyberbullying victims can aid the prevention and detection of cyberbullying by highlighting groups of potentially vulnerable young people. A review by Tokunaga (2010) did not draw any definitive conclusions concerning gender differences in relation to cyberbullying. However, recent research has identified that girls are more likely than boys to experience cyberbullying (Livingstone et al., 2014; Olenik-Shemesh et al., 2012; Schneider et al., 2012; Tsitsika et al., 2015; Waasdorp & Bradshaw, 2015), with the present study offering additional support for girls being most at risk. This finding is in stark contrast to the current understanding of gender and traditional forms of bullying, where boys have consistently demonstrated a greater risk of being involved as either perpetrator or victim (Craig et al., 2009; Nansel et al., 2001). A higher prevalence of cyberbullying among girls may be explained in part by existing research which demonstrates girls are more likely than boys to use electronic forms of communication (Brooks et al., 2011; Lenhart, 2015). It has also been suggested that the anonymity of the Internet enables people to act in ways that are outside of regular social norms, and one consequence of this may be to allow females to display more aggression than they otherwise would (Ybarra & Mitchell, 2004).
Young people of all ages can be cyberbullied, but age appears to be a significant individual trait with older adolescents more at risk of being victimized in this way. Other studies that have looked at age and cyberbullying have found a trend toward a “peak” of bullying perpetration that occurs roughly at age 13 to 15 depending on the study (Aboujaoude et al., 2015; Wade & Beran, 2011), and our findings appear to support this in an English population. There may be many reasons why the incidence of cyberbullying peaks at a later age than traditional bullying, but it has been hypothesized that different forms of bullying necessitates different levels of cognitive ability (Peeters, Cillessen, & Scholte, 2010). For example, Sutton, Smith, and Swettenham (1999) had argued that certain forms of bullying relies on sophisticated manipulation and are grounded in theory of mind, which would require a level of cognitive ability and social intelligence that may not develop until mid-adolescence. The need for a certain level of social intelligence in relational aggression has been further supported by Björkqvist, Österman, and Kaukiainen (2000). Thus, the ability to understand how to use social media and other online settings for cyberbullying may require a sophisticated level of development that is not yet evident in younger adolescents.
Limitations and Future Research
As is the nature of cross-sectional research, the results cannot imply causality; the multilevel analysis identified associations between assets from varying social environments and cyberbullying, but the direction of these relationships cannot be concluded. For example, a positive school environment may foster lower levels of cyberbullying, but equally it could be that lower levels of cyberbullying create a more positive perception of the school environment. The study of cyberbullying from an assets-based perspective is a novel approach, and the findings reported here provide an initial snapshot of protective ecological assets from different domains of the adolescent world. Future longitudinal research would be able to confirm the direction of causality among cyberbullying and the protective assets.
We acknowledge that bullying response rates have been shown to vary across measurement approaches (Modecki et al., 2014). The breadth of the HBSC England survey prevented cyberbullying from being examined in detail, and we appreciate the behavioral checklist utilized in the present study may have omitted other cyberbullying behaviors. However, text messages and social media have been identified as the most common forms of cyberbullying (Whittaker & Kowalski, 2015), both of which are addressed in the present study and thus increase confidence that the current measure is capturing the vast majority of cyberbullying experiences. Future research exploring protective assets would benefit from a more comprehensive measure of cyberbullying.
The current article did not control for other forms of bullying, and as such, a proportion of young people who were cyberbullied were also victimized in other ways. The purpose of the study was to identify factors associated with the existence of cyberbullying, and inclusion of other forms of bullying behaviors would likely confound the effect of these variables. However, it warrants further research to examine whether protective assets differ across types of bullying experiences.
Conclusion
Online activity has become an integral aspect of young people’s lives, and as such, should be examined in its social context (Chapman & Buchanan, 2012). The social–ecological theory provides a useful framework for examining the interplay between environmental factors and adolescent cyberbullying. Moreover, it is important that empirical work seeks to examine what protective health assets work for who and in what context. Utilizing an assets-based approach to the study of cyberbullying highlights the importance and the protective nature of young people feeling connected to and having a sense of belonging in any of the multiple environments of school, family, and community.
Moving beyond a traditional risk perspective and utilizing the social–ecological model to identify protective ecological health assets enable the development of interventions that span environments of the adolescent. The present article emphasizes that the importance of engagement between the ecological systems, namely, the school, family, and neighborhood, may be most effective in reducing cyberbullying. Older adolescents and girls were identified as experiencing higher levels of exposure to cyberbullying, indicating that potential value of targeted interventions. However, it is important to note that no significant interactions with age or gender were retained in the model, suggesting that the protective assets identified in the present article provide a useful overall set of protective factors that are equally beneficial to both boys and girls of all ages.
Footnotes
Acknowledgements
The authors acknowledge the schools, teachers, and especially the young people who took part in the study. We are grateful for their participation in this project.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the Department of Health, England.