Are rates of bullying and cyberbullying getting better or worse?

The Health Behaviour in School-aged Children (HBSC) surveys have provided data on incidence rates of being bullied and bullying others, at four-year intervals since the 1990s. Molcho et al. (2009) presented trends over time, between 1993/1994 and 2005/2006. Decreases were observed in the majority of the 27 countries for which there were data over all the time points. The most striking finding was for chronic bullying of others; the country average fell from 19.3% in 1993/1994, to 16.1% in 1997/1998, 11.1% in 2001/2002, and 10.6% in 2005/2006; on a larger set of countries, this figure fell further from 10.7% in 2005/2006 to 10.1% in 2009/2010. It does appear to have risen slightly in 2013/14, this being the first survey to include questions on cvberbullying.

Apart from the HBSC surveys, Rigby and Smith (2011) reviewed findings from independent studies in England, Wales, Finland, and Australia. These too generally supported evidence for a decline. Another example comes from Waasdorp, Pas, Zablotsky and Bradshaw (2017), who reported ten-year trends from 109 schools in Maryland, USA. Between 2005 and 2014, many bullying indicators showed a fairly steady decline over this period. For example, reports of being victimized showed a drop from 28.5% in 2005 to 13.4% in 2014. Reports of having witnessed bullying showed a monotonic decline every year from 66.4% in 2005 to 42.7% in 2014.

Given that we know that antibullying interventions generally have some, albeit partial, success (Ttofi & Farrington, 2011), it is plausible that increased awareness and implementation of antibullying interventions has helped produce this decline. However as of yet, the evidence mainly concerns traditional bullying. There is much less evidence for a decline in cyberbullying, and some evidence of an increase.

Obviously cyberbullying increased as the relevant technology has become more and more accessible in the early years of this century. Jones, Mitchell and Finkelhor (2012) reported data from Youth Internet Safety Surveys in the USA conducted in 1999–2000, 2005 and 2010. The data closest to cyberbullying are that on internet harassment; the surveys showed an increase in this from 6% to 9% and then 11%, this being more marked for girls. A composite measure of amount of internet use increased from .24 to .41 and then .49 over the three time points.

Olweus (2012) reported data from a very large US sample from 2007 to 2010; no systematic time trends were found, although if anything the trend is slightly upward. The Maryland study by Waasdorp et al. (2017) included data specifically on cyberbullying. Whereas most bullying measures showed a nearly or perfectly monotonic yearly decline over the period 2005–14, the data on being victimized by cyberbullying appear to show a more curvilinear trend, with an increase in the first half of the period (up to 2010) before a decrease in the second half.

Olweus (2012) also analyzed data from 41 Norwegian schools for 2006–10, and again found little systematic trend, although if anything the direction is slightly downward. In Europe more widely, the EU Kids online study (Livingstone et al., 2011) carried out surveys in 2010 of children aged 9–16 years, in 25 European countries; a follow-up in seven countries from 2010 to 2013/14 has suggested some rise in cyberbullying, especially for girls, from around 7% to 12% (Hasebrink, 2014).

In summary, there is good evidence that in many countries rates of involvement in traditional bullying have declined over the last 10 or 20 years. There is much less evidence about cyberbullying involvement; but it clearly increased in the early years of this century, and has proliferated into different forms as the proportion of students having access to mobile phones and the internet increased. This penetration may now be approaching saturation in many communities; there is mixed evidence about whether rates of cyberbullying are continuing to rise, are rather stable, or may even be declining in some studies.

A fundamental aim of cyberbullying research should be to help design effective ways of reducing cyberbullying and the harm it causes. The apparent partial success in tackling traditional bullying, and the mixed picture regarding cyberbullying, points to the need to develop (or continue developing) resources and intervention strategies further for cyberbullying, as has been done for traditional bullying. This book brings together an important and diverse set of such attempts.

Interventions can vary along various dimensions. One dimension is technological versus relationship oriented. A second is general (bullying) versus specific (cyberbullying). A third is where interventions focus on an ecological dimension from individual to small group to institution to society.

Technological versus relationship oriented interventions

As posed very clearly in Chapter 3, technological interventions aim to make bullying or cyberbullying more difficult to do, or the costs more obvious. They do not directly tackle the relationship issues thought to underlie bullying, in a way that many other interventions (for example all those described for traditional bullying in Chapter 2) do. Chapter 3 gives an excellent overview of such interventions for cyberbullying, while also pointing out the paucity of evaluation of such measures. However, there is no reason to suppose that they could not be implemented together with relationship-based measures. Also, as Chapter 6 makes clear, technological approaches such as online social marketing can tackle relationship issues such as empathic attitudes. Another example is the text-messaging program BullyDown for bullying generally (Ybarra, Prescott, & Espelage, 2016).

Legal measures are not technological, but can also be seen as not concerned with relationships per se, but as signaling the wrongness of cyberbullying and making some form of negative outcome for perpetrators more likely. Chapter 4 sets out some important issues regarding legal approaches, especially in Australia, the UK and the USA. Again, we know little of how effective legal approaches really are, but there are a couple of studies from the USA. Hatzenbuehler et al. (2015) compared 25 states, and (after partialing out some confounding factors) found that having some legislative component to antibullying laws was associated with reduced rates of being both bullied and cyberbullied. This was a cross-sectional study, but Ramirez et al. (2016) using a more longitudinal design, examined rates of being bullied in the US state of Iowa in 2005 (just before an antibullying law was introduced), 2008, and 2010. They found an increase in rates of being bullied in 2008, possibly due to increased reporting, but then a decrease by 2010. However, being cyberbullied (one of four aspects of being bullied, and the least frequent) did not decrease by 2010. While both these studies have drawbacks, they are a beginning to evaluating the impact of legal measures.

Specialized versus generalized interventions

Should resources and interventions be targeted very specifically for cyberbullying, or can we rely on those devised for bullying generally, or even specifically for offline bullying (see Chapter 2), to generalize? The chapters in this book present a range of actions along this dimension. Purely technical actions, as exemplified in Chapter 3, can be for bullying or cyberbullying; and similarly many countries have laws about bullying or harassment generally, but there is also discussion of laws specifically on cyberbullying, as demonstrated in Chapter 4. Again there are many curricular resources for bullying, but specific curricula for cyberbullying, for example e-safety teaching, can be utilized, for example Chapter 11. This can include specific coping strategies for dealing with cyberbullying. All these resources may be rather specific to cyberbullying, or part of a more general curricula approach, such as social and emotional learning and general anti-bullying prevention work.

MARC (Chapter 8) has material on bullying generally, but also (especially for older students) more specialized curricula material on cyberbullying. NoTrap! (Chapter 10) has both general bullying and specific cyberbullying elements. Cyber-Friendly Schools (Chapter 7) has many components specifically on cyberbullying, but also material on bullying generally.

ConRed (Chapter 15) focuses mainly on cyberbullying, including e-safety and relationship education (convivencia). MediaHeroes (Chapter 11) mainly focuses on cyberbullying, as does the IMB model (Chapter 17) and EPPM model (Chapter 18) interventions, with curricula material on cyberbullying.

Stop Online Bullies (Chapter 13) specifically argues for the advantages of targeting both traditional and cyberbullying. In fact, we know that there is considerable overlap between pupil involvement in traditional and cyberbullying (Modecki et al., 2014; and Chapter 1). This means we can expect interventions primarily focused on traditional bullying to have some positive effects in reducing cyberbullying as well. Just such carryover effects have been reported for the KiVa program in Finland (Williford et al., 2013; and Chapter 7), and for the ViSC Social Competence Program in Austria (Gradinger et al., 2016; and Chapter 14). For example, both general empathy training and modifying beliefs supportive of aggression could be helpful in reducing all kinds of bullying including cyberbullying (Ang, 2015). However, resources and interventions tailored more specifically for cyberbullying are very likely to be important too. It is worth noting that the KiVa program does include some specific cyberbullying elements in its curricula work (Chapter 9). Interestingly, MediaHeroes (Chapter 11) is much more specifically on cyberbullying, but has carryover effects for traditional bullying.

An ecological perspective

Another dimension to consider comes from the ecological perspective (Cross et al., 2015; Hong & Espelage, 2012). Moving outward from the center of Bronfenbrenner’s model, we can consider the individual, his/her immediate social groups (parents, siblings, teachers, peers), the school, the community, and the wider society. Cross et al. (2015) extend this model by introducing online influences between peer and community level. A whole-school approach (e.g., Chapter 7, Chapter 8) can embrace much of this. Several aspects are covered for example in NoTrap! (Chapter 10; whole school, peers, and individuals) and in Media Heroes (Chapter 11; individual, class and parents).

Interventions at the level of individual factors would include attempts to enhance social skills and empathy, and reduce moral disengagement, as is aimed for in many curriculum interventions; and also to improve coping strategies for victims, as for example in NoTrap! (Chapter 10) and + Fort (Chapter 12).

Considering the immediate social group, family (parents and siblings) are clearly important. Parent-child and sibling relationships affect the likelihood of being involved in bully or victim roles (Lereya et al., 2013; Wolke et al., 2015); and parents also have an important role in working with schools, supporting antibullying initiatives, and liaising with schools if they have concerns about a child’s behavior (Axford et al., 2015). Given that much cyberbullying is initiated outside school, they also have a particular, but difficult, role in discussing e-safety and cyberbullying issues with their children, giving guidance on acceptable internet practice, and being involved without being overly restrictive (Sasson & Mesch, 2014). Chapter 5 lays out important issues around parental roles regarding cyberbullying.

Chapter 7 provides an example of a parent component as part of a whole-school approach. Interestingly it suggests that the often-touted digital divide may be beginning to lessen, in some respects at least, as incoming generations of parents are more internet aware. Nevertheless, the authors point to important misconceptions that parents may have about the risks their child may be taking and how best to deal with them. In this respect, more parental guidance continues to have an important role. Intergenerational collaboration is an important aspect in the Smartphone Summit approach in Japan (Chapter 16) (with adults sometimes learning from children). There are parent components in Media Heroes (Chapter 11) and the EPPM model intervention (Chapter 18).

An under-researched area appears to be the role of siblings in cyberbullying. We know something of the importance of siblings in bullying generally (Wolke et al., 2015), but not in cyberbullying specifically. Do siblings share experiences on the internet? Do they cyberbully each other? Or do they support each other if one is experiencing cyberbullying? And what variables (such as age gap, gender) affect these?

Within the school teachers have an influential role, especially at primary ages. So far as general bullying is concerned, we know that teacher aspects such as openness in communication, and perceived importance of bullying, are important predictors of success in the Olweus Bullying Prevention Program (Olweus, 2004); and in KiVa, the general pupil perception of homeroom teacher’s attitude to bullying was a significant predictor of levels of victimization (Saarento et al., 2013). The MARC (Chapter 8), KiVa (Chapter 9), and ViSC (Chapter 14) programs emphasize teacher training components.

Perhaps even more important than family and teachers by secondary school age is the peer group. We know the importance of class norms (descriptive and injunctive) about bullying (Saarento et al., 2013); the protective effect of a friend who is able to protect a victim (Fox & Boulton, 2006), and that bystander intervention can be increased (Polanin, Espelage, & Pigott, 2012) and can be effective especially if defenders have high status in the peer group (Caravita et al., 2009). This knowledge is built on in KiVa (Chapter 9), and bystanders’ responsibility is also emphasized in NoTrap! (Chapter 10). Peers can also act as mentors and this is an aspect of MARC (Chapter 8).

Many interventions at the individual and peer group level may have positive effects at the school level, as assessed for example by school climate measures. For example, peer support schemes may often have limited effects on victim rates (Flygare, Gill & Johansson, 2013; Ttofi & Farrington, 2011) but do appear to improve school climate (Cowie & Smith, 2010). Between school studies have found links of school climate to bullying rates (e.g., Elgar et al., 2009). There is also an obvious need to ensure that cyberbullying is explicitly included in school antibullying policies (Smith et al., 2012).

Although interventions are mainly school based and focus on individual, family, peer group, and whole-school aspects, wider aspects in the ecological model are important. Considering online level influences (Cross et al., 2015), social media companies and internet service providers are under pressure to do more to enhance internet safety and reporting of abuse (Chapter 3; and Coyne & Gountsidou, 2013).

The media generally often show violence, and some studies have established links between violent media exposure, and involvement in bullying or cyberbullying. For example, Calvete, Orue, Estévez, Villardón, and Padilla (2010) in Spain, Fanti, Demetriou and Hawa (2012) in Cyprus, and den Hamer, Konijn and Keijer (2014) in the Netherlands all found links from violent media exposure (on television, internet, movies, video games), to both cyberbullying and cybervictimization. den Hamer et al. (2014) supported a “Cyclic Process” model that adolescent victims of (mainly traditional) bullying feel anger and frustration, they may then tend to use violent media more, as a way of coping with or finding outlets for their anger feelings, and that this exposure in itself may lead to cyberbullying behaviors, for example through desensitization, imitation, or modeling of action scripts. The model is “cyclic” because the cyberbullying might then increase the chances of being a victim again.

How cyberbullying is portrayed in the media (such as newspapers) may also be important. Vandebosch et al. (2013) analyzed 1599 newspaper articles from 8 countries, over the period 2004–11, which covered cyberbullying. While raising awareness of the topic is vitally important, these authors also pointed out the dangers of creating a “moral panic,” and the difficulties in achieving balanced reporting on the issue.

At the community level, there is evidence that levels of neighborhood violence may change normative beliefs about aggression, as shown in a US study by Schwartz and Proctor (2000). Children who had witnessed community violence were more likely to be aggressive to classmates; and those who had been victims of community violence were more likely to be victims at school as well. A study in Colombia by Chaux, Molano and Podlesky (2009) also related violence at the municipality level, as well as poverty and inequality, to bullying involvement. These studies did not explicitly study cyberbullying, but given the high overlap between online and offline bullying, may well be relevant.

At the society level, issues regarding legal provisions have already been discussed (Chapter 4). However, we also know that factors such as country wealth and distribution of wealth may be important (at least for traditional bullying). Research comparing countries (using for example HBSC data) has found that income inequality, as a measure of power differentials between those who have access to resources and those who do not, is a strong predictor of country differences in bullying involvement (Chaux et al., 2009; Due et al., 2009; Elgar et al., 2009). Elgar et al. (2009) argued that “it is possible that redistributing wealth and creating more egalitarian societies would do more for reducing bullying than school-level policies or individual-level intervention” (p. 358). However, such political change may be even more difficult to achieve than changes at lower levels in the ecological model.

Sample characteristics

Should interventions be tailored to individual characteristics, or to sample characteristics such as age, gender, disability, or culture? The most effective might be tailoring to the individual circumstances of the young person, but this is obviously difficult for school or class based interventions. However, it was done in the Stop Online Bullies’ (Chapter 13) project, which was adult-free, with a computer program adjusted to the pupil’s level of knowledge and skills. This was also for low academically achieving pupils, but unfortunately perhaps because of this, the program lacked sustainability through repeat visits. Some element of tailoring was also present in ConRed (Chapter 15), and at the workshop level, in the IMB model intervention (Chapter 17).

Age is an important factor in bullying intervention research, with most programs showing less effectiveness in secondary compared to primary school ages (Yeager et al., 2015). For example, KiVa has been found to be more effective with younger pupils (Chapter 9). Nevertheless, and appropriately given a somewhat shifted age curve for cyber compared to face-to-face bullying (Kowalski et al., 2014), many of the interventions in this volume target secondary school students; NoTrap! (Chapter 10), Media Heroes (Chapter 11), + Fort (Chapter 12), Stop Online Bullies (Chapter 13), ViSC (Chapter 14), and the IMB model intervention (Chapter 17). Interestingly, the IMB model intervention found more effect on risk perception at higher grades. In contrast to age, there has been less discussion of gender as an issue. Tailoring by gender would anyway be difficult in mixed-sex schools (it would be less difficult in, for example, India, where many classes are sex segregated; Schäfer et al., 2018). By and large, different rates of effectiveness were not expected by gender. This was explicitly found in NoTrap! (Chapter 10), and was also the case in the Cyberprogram 2.0 intervention reported by Garaigordobil and Martinez-Valderrey (2016).

Interventions may also work differently, in different cultures. We know there are considerable differences in bullying rates between different countries, even if these are difficult to measure reliably (Smith, Robinson, & Marchi, 2016). Following the EU Kids online model (Livingstone & Helsper, 2013), these might be explained by factors such as cultural values, education system, technological infrastructure, regulatory framework, and socioeconomic stratification. Besides differences in prevalence, there may be differences in what is considered to be bullying, the types of bullying that occur, attitudes to bullying and seeking help (e.g., Smith, Kwak, & Toda, 2016), and even issues affecting assessments such as social desirability and reference group effects (Guillaume & Funder, 2016).

The interventions described in this book are mostly from western countries, although the Smartphone Summit was in Japan (Chapter 16) and the IMB model intervention in South Africa (Chapter 17). Interventions are also developed within one cultural setting. Translating an intervention from one country to another can be problematic; for example, the OBPP has been very successful in Norway but less so in the USA (Smith, 2014). Nevertheless, some interventions have been tried elsewhere with some success, for example KiVa in Italy (Nocentini & Menesini, 2016) and several other countries, ViSC (Chapter 14) in three other countries (though with more mixed success than in Austria), and MediaHeroes (Chapter 11) is being tested elsewhere. The kinds of issues (such as cultural values) that affect differences in prevalence and attitudes may have implications for certain components of interventions. For example, it is interesting that getting pupils to take part in a sequence of voluntary activities was achieved successfully in a collectivist country such as Japan (Smartphone Summit. Chapter 16) but much less so in a more individualist society, Belgium (Stop Online Bullies, Chapter 13).

Some issues in intervention research

A couple of distinctive issues come from reading the chapters in this collection. One is the idea of “pupil voice” or pupil participation in the intervention. The other is the role of theory in designing interventions.

Pupil voice: “Don’t do it about us, without us” is a slogan heard in relation to involving pupils in meaningful ways in designing and implementing interventions. There are various ways of doing this, and a number are exemplified in this collection. In Chapter 6 we see young people involved in design of online social marketing. In Cyber-Friendly Schools (Chapter 7), student cyberleaders develop activities. In NoTrap! (Chapter 10), peer educators play an important role. Children decide the topic initiative in Smartphone Summit (Chapter 16); and pupils advise one another in the EPPM model intervention (Chapter 17). Another example of pupil participation is in the Cross-Age teaching Zone (CATZ) model, in which 11-year-old students acted as e-safety tutors for 9 year olds (Boulton et al., 2016).

Role of theory: some interventions have relatively unarticulated theoretical backgrounds. As noted, a background philosophy is often to intervene at various levels on the ecological model, and to increase awareness and empathy, change attitudes, and change behavior. One theoretical background to bullying intervention was proposed by Hawley and Williford (2015). They used Ajzen’s Theory of Planned Behavior to advocate changes in perceptions (for example of what bullying is), attitudes (about bullying behavior and toward victims; reporting bullying and intervening), subjective norms (how do others think or expect I should behave?), and efficacy beliefs (feeling confident that actions such as reporting or defending will be successful and not result in negative consequences) in order to achieve desired changes in actual behaviors. In addition, they took ideas from organizational science to argue the necessity of targeting these with pupils, teachers, and other staff in the school or relevant to the school, in order to change school culture and school climate.

This is an ambitious blueprint, but some of the interventions in this book go some way toward this, using the same or similar theoretical models. The Ajzen Theory of Reasoned Action/Planned Behavior is invoked in MediaHeroes (Chapter 11). A theory of normative social behavior is used in ConRed (Chapter 15). An Information-Motivation-Behavioral (IMB) Skills’ model is adopted in Chapter 17, and an Extended Parallel Process model (EPPM) in Chapter 18. These all target some aspects of attitudes, norms, intentions, self-efficacy, and behavior. From a somewhat different perspective, REBT is used in Stop Online Bullies (Chapter 13).

Implementation and evaluation

There are a range of issues to be considered in the actual implementation of interventions, and their evaluation (see for example Ryan & Smith, 2009). Here I consider aspects under four main headings: Feasibility (can you do it?); Fidelity (is it done?); Follow-up or sustainability (will it continue to be done?); and Efficacy (does it work?).

Feasibility: the ecological model, and the theoretical rationale proposed by Hawley and Williford (2015), suggest that a successful intervention will be multifaceted in terms of various target groups and outcome objectives. This could involve a major investment of time and resources. Even if feasible in an experimental study, can this be subsequently generalized to other schools without intensive input from researchers? Perhaps the best example of this so far is the KiVa program, which has been taken up by most schools in Finland. This has however been enabled by substantial financial support from the Finnish Ministry of Education and Culture (Chapter 9).

Some examples are much more modest, and consequently less costly to implement, but generally have more limited targets or outcome objectives. For example, the IMB model intervention (Chapter 17) has one 50-min workshop, with a main outcome of changing pupil risk perception. Similarly, the EPPM model intervention (Chapter 18) has one 45-min session. Stop Online Bullies (Chapter 13) has low cost as a standalone procedure (not needing adult guidance), but this in itself may have contributed to a large dropout rate over sessions (especially for low academically achieving young people). + Fort (Chapter 12) uses a mobile app (focused on coping strategies), which can be downloaded for free.

Another way to increase feasibility is to employ some kind of “train the trainer” or cascade model of introducing the intervention. MARC (Chapter 8) and MediaHeroes (Chapter 11) used train the trainer in this way, and ViSC (Chapter 14) used a cascade model for teacher training.

Smartphone Summit (Chapter 16) used a cascade model for dissemination by children.

Fidelity: How well are program interventions followed? For bullying generally, this issue was highlighted by Flygare, Gill and Johansson (2013). They found that in Sweden, bully intervention programs were implemented such that components were mixed from more than one program. Rather than evaluating programs they switched to evaluating components. This is an endemic problem. In Cyber-Friendly Schools (Chapter 7), only about one-third of the intervention was implemented by classroom teachers, In Kiva, schools varied greatly in implementation; in primary schools, teachers delivered on average 8.7 out of 10 lessons in the initial RCT trial, but this fell to 7.8 during the first year of the national rollout, and to 7.2 in the second year (Salmivalli & Poskiparta, 2012).

Fidelity does appear to be important. In Kiva, it correlated with victim and bully reductions (Salmivalli & Poskiparta, 2012). In the OBPP, it was an important predictor of success (Olweus, 2004). The ViSC project (Chapter 14) found more effects on teachers who implemented the program with greater fidelity.

Follow-up, Sustainability: Do programs continue effectively once the researchers have left?

Schober et al. (2016) emphasized the importance of implementation research to connect with intervention research and to work with policy makers and practitioners from the start; ViSC being an example of this. Indeed ViSC (Chapter 14) reported continuing positive effects of the intervention at follow-up.

Efficacy: How well do interventions actually work? In order to ascertain this, the Randomized Control Trial (RCT) has been seen as the gold standard, in terms of ruling out most alternative explanations for findings. Many reports in this book have used RCT at some level: Cyber-Friendly Schools (Chapter 7), KiVa (Chapter 9) in their initial study, and ViSC (Chapter 14) used RCTs by random assignment of schools; MediaHeroes (Chapter 11) and the IMB model intervention (Chapter 17) used RCT by random assignment of classes; and Stop Online Bullies (Chapter 13) planned RCT by random assignment at the individual level but was unable to complete this due to heavy dropout. RCTs are indeed likely to be the most convincing, but they are not without problems, including how natural (and ethical) it is to require schools to follow an experimental or control procedure that they might not have followed if given free choice.

Sometimes RCTs are not feasible for such practical reasons. It is then possible to use an experimental/control pre/post-test design, which is still very effective at ruling out alternative explanations of effects if initial differences between experimental and control groups are taken account of. Such nonrandomized designs were employed by NoTrap! (Chapter 10), + Fort (Chapter 12), and ConRed (Chapter 15). The IMB model intervention (Chapter 17) also compared experimental and control classes.

Even if there are no control schools, evaluation of an intervention can be made using an Extended Selection Cohorts’ design (comparing same-age children who have not or have experienced the intervention, in successive years). This controls for age effects, but not history effects (Smith, 2014). This has been used in OBPP evaluations, and for the KiVa nationwide rollout of the program (when control schools were not so feasible). In addition, given imperfect fidelity of implementation, the correlation of fidelity with outcome, or the dosage-response relationship, can also inform on the effects of intervention.

Finally, “softer” evaluation procedures can be employed and still give useful information, albeit not so convincing as those given earlier. These can include more qualitative assessments, for example of how the implementation was perceived by stakeholders. The Smartphone Summit (Chapter 16) is an example of this. However, qualitative methods can be combined powerfully with quantitative methods to give more insight into process (Ryan & Smith, 2009). MARC (Chapter 8) mixed qualitative and quantitative methods in a quasiexperimental design, and + Fort (Chapter 12) and Stop Online Bullies (Chapter 13) used both qualitative and quantitative evaluations.

Clearly the most desired for outcomes are changes in behaviors, but changes in knowledge and attitudes, empathy and feelings of efficacy, may be way stations toward this (Hawley & Williford, 2015). Positive effects on attitudes as well as behavior were reported in KiVa (Chapter 9), on awareness, empathy and behavior in NoTrap! (Chapter 10), and on empathy, and cyberbulling perpetration (but not cybervictimization) in Media Heroes (Chapter 11). MARC (Chapter 8) reports more on knowledge and attitudes than on behaviors.

A related issue is who are the informants for the outcomes. Ryan and Smith (2009) advocated a multi-informant assessment; for example, self, peers, and teachers. Cyber-Friendly Schools (Chapter 7) provides an example of multiple informants (students, including cyberleaders, and school staff).

Generally, outcomes reported are positive, or else nonsignificant. However, the possibility of iatrogenic effects (negative effects resulting from the intervention) have to be faced. Some interventions could be counterproductive. An example is some kinds of peer support (Cowie & Smith, 2010; Flygare et al., 2013) which may be stigmatizing either of those who seek support (e.g., buddy bus stops) or of peer supporters themselves (e.g., uniforms that can be made fun of). Other examples might be certain kinds of social skills’ training, which might simply enhance the skills put to negative ends by bullies; or encouraging victims to tell about bullying without effective follow-up.

Cost effectiveness: Even if an intervention is demonstrated to have effectiveness in terms of desired outcomes, whether it will be more widely adopted is likely to depend on how cost effective it is. This links to issues of feasibility discussed earlier, but also takes account of quantifying the benefits of the intervention. If bullying is reduced, then fewer pupils are likely to have mental health problems, and more are likely to live productive lives. Persson and Svensson (2013), and Beckman and Svensson (2015), considered the costs of two Swedish antibullying programs, Friends, and the OBPP; and weighed them against actual costs of court settlements brought by individuals against schools or education authorities for being bullied in school; and to a survey of what taxpayers were prepared to pay to reduce school bullying. On these measures, the programs were deemed to be cost effective.

Summary

Antibullying interventions have a history of some 30 years, but those targeted at or to include cyberbullying have a history of only a few years. This volume illustrates a range of attempts to tackle it. The diversity of approaches is, at this stage, to be welcomed. The interventions described have varying levels of ambition, and varying levels of success. Interventions that are wider in scope are likely to be more successful, but may also be more costly. Among a range of issues brought forward for consideration in this chapter, the cost effectiveness of interventions deserves more scrutiny if we hope to have a wide take-up as well as a useful impact. Also, the issue of pupil voice, or pupil participation in designing interventions, may be particularly relevant for cyberbullying. Our challenge is to further develop and refine a range of approaches and resources so that young people can be safer in cyberspace.