Introduction

The prevalence of bullying victimisation among Irish students in second-level education is high. Earlier studies of Irish adolescents reveal that in the 1990s, almost 15.6 % of 12–18-year-old students reported having been bullied at some point (O’Moore et al. 1998). By the mid-2000s, the number of 12–16-year-old students in second level who reported being bullied in the previous 3 months was 36.5 % (Minton 2010). However, it is difficult to estimate if the prevalence of bullying is rising, given that studies use contrasting measurements of prevalence and frequency, and the ways in which adolescents are bullied are changing with the involvement of new media (Rigby and Smith 2011). Internationally, it is also challenging to compare prevalence rates of bullying; reports of bullying vary from 13 to 75 % across studies and methodology is a key factor contributing to these discrepancies (Jimerson et al. 2010).

Researcher definitions of bullying

The single largest methodological issue affecting the comparability and consistency of bullying research findings is the lack of a standard definition of bullying among researchers. A 2014 report by the Centers for Disease Control and Prevention in conjunction with the US Department of Education claimed that without a uniform research definition of bullying, our ability to understand the true magnitude, scope, and impact of bullying is severely impeded (Gladden et al. 2014).

Accordingly, they consulted with bullying experts and practitioners and developed a uniform research definition of bullying for children and adolescents: “Bullying is any unwanted aggressive behavior(s) by another youth or group of youths who are not siblings or current dating partners that involves an observed or perceived power imbalance and is repeated multiple times or is highly likely to be repeated. Bullying may inflict harm or distress on the targeted youth including physical, psychological, social, or educational harm” [(Gladden et al. 2014), p. 7].

While researchers had not previously reached a unanimous agreement on a definition of bullying, many bullying researchers generally agreed that to differentiate bullying from other forms of aggressive behaviour, bullying must meet three criteria: intention to cause harm, a power imbalance in favour of the bully, and repetition of bullying over time (Farrington 1993; Olweus 1999). These elements are reflected in the Gladden et al. (2014) definition.

Adolescent definitions of bullying

Although there may be general agreement among researchers about what constitutes bullying, this definition is not representative of adolescent conceptualisations of bullying (Cuadrado-Gordillo 2012; Frisén et al. 2008; Guerin and Hennessy 2002; Madsen 1996; Naylor et al. 2006; Vaillancourt et al. 2008). While elements of the “research definition” are evident in adolescent definitions in some studies, it is generally accepted by a minority of students. For example, Vaillancourt et al. (2008) found that children and adolescents rarely include references to intentionality (1.7 %), repetition (6 %), or power imbalance (26 %) in their spontaneous definitions of bullying. Similarly, Naylor et al. (2006) reported that 11–14-year-olds seldom include these criteria when asked to define bullying (3.9 % for intentionality, 7.9 % for repetition, 40.5 % for power imbalance). Likewise, in a sample of Swedish 13-year-olds, Frisén et al. (2008) found that repetition (30 %) and power imbalance (19 %) did not appear in the majority of definitions. Qualitative work with 10–13-year-olds shows a similar disjoint between researcher and young persons’ bullying conceptualisations (Guerin and Hennessy 2002). However, Cheng et al. (2011) note that Taiwanese adolescents frequently report examples reflecting intentionality and power imbalance (but not repetition) when asked to define bullying. The authors concluded that Asian adolescents may differ in their conceptualisation of bullying compared to their Western counterparts due to culture and collectivism.

In brief, it is evident that research on bullying typically uses a definition provided by researchers, which is not always congruent with the definition provided by young people. The above examples demonstrate that young people in Western cultures may not view intentionality, repetition, or power imbalances as central to their classification of bullying, despite these aspects being integral to researcher-generated definitions. This discrepancy is important as adolescents are less likely to report an instance of bullying when they are provided with a researcher’s definition compared to when they engineer the definition themselves (Madsen 1996; Vaillancourt et al. 2008). In addition, adolescent descriptions of bullying differ from adult and teacher descriptions (which tend to be more in line with researchers’ definitions), and this is a concern for how adults respond to adolescent bullying (Menesini et al. 2002; Mishna et al. 2005; Naylor et al. 2006). A person’s definition of bullying can shape how they respond to bullying in everyday life (Madsen 1996), and therefore examining adolescent definitions is paramount to guiding intervention design.

Age, gender, and experience of bullying

Monks and Smith (2006) found clear age-related differences in children’s and adolescents’ (14-year-olds’) understanding of bullying and suggest cognitive development as a possible driver of the change in definition, given that more advanced cognitive processes allow adolescents to conceptualise bullying along a number of dimensions. However, almost every study that has considered adolescent definitions of bullying has not included adolescents over the age of 13 or 14 (Frisén et al. 2008; Guerin and Hennessy 2002; Naylor et al. 2006; Menesini et al. 2002; Monks and Smith 2006; Smith et al. 2002) which leaves a significant gap in the literature regarding how older adolescents conceptualise bullying. It also hinders our developmental understanding of how conceptualisations of bullying may change from early adolescence to late adolescence.

The evidence for gender differences in conceptualisations of bullying is equivocal. Some research suggests that females are more likely to emphasise the effects upon the victim (Frisén et al. 2008). However, Smith et al. (2002) found few gender differences among definitions of bullying, despite there being a difference in the bullying behaviours exhibited by males and females. In addition, Guerin and Hennessy (2002) reported no gender differences in the definitions of bullying provided by their sample of Irish 10- to 13-year-olds.

Adolescents’ definitions of bullying also change as they observe or experience bullying in school (Monks and Smith 2006; Monks et al. 2003). However, few studies have directly addressed whether there is a difference in the self-generated definitions provided by non-bullied versus bullied adolescents (Naylor et al. 2001), and this requires further attention.

The present study

The aim of this study was to examine the themes that emerge from adolescents’ self-generated definitions of bullying. Given that previous research indicates that definitions of bullying may differ due to previous which leaves a significant, age, or gender (Monks and Smith 2006; Frisén et al. 2008), this study will consider how the definitions provided by adolescents differ along these dimensions. As previous literature has already shown that adolescents’ definitions do not frequently contain key elements of research definitions of bullying (Cuadrado-Gordillo 2012; Frisén et al. 2008; Guerin and Hennessy 2002; Naylor et al. 2006; Vaillancourt et al. 2008), this study will not attempt to examine themes of repetition, intentionality, or power imbalance. Rather, this study seeks to identify explicit patterns within the data, and themes will be identified if they capture an important element of young people’s definitions of bullying (Braun and Clarke 2006). This study uses an inductive approach to look at previously researched definitions of bullying from a different perspective. In addition, some previous studies investigating adolescent definitions of bullying have provided adolescents with a list of agree/disagree statements regarding what bullying is (Cuadrado-Gordillo 2012) or asked participants to watch a cartoon scenario and rate whether it constituted an episode of bullying or not (Monks and Smith 2006). To avoid the possible demand characteristics associated with prompted responses, this study will employ a free-response method in line with the methodology of Vaillancourt et al. (2008) and Naylor et al. (2006) to ascertain unprompted definitions. This is the first nationally representative study in a European country to investigate adolescents’ self-generated definitions of bullying across a wide age range (12–19 years), unlike previous studies that typically only include responses from adolescents up to the age of 13 or 14 (Frisén et al. 2008; Guerin and Hennessy 2002; Naylor et al. 2006; Menesini et al. 2002; Monks and Smith 2006; Smith et al. 2002). This will allow for the examination of age-related differences in conceptualisations of bullying.

Method

Procedure

This study used previously collected, cross-sectional data from the My World Survey-Second Level (MWS-SL; Dooley and Fitzgerald 2012). A multi-stage, random sampling strategy was used to ensure the sample was representative of second-level students in Ireland. At least one second-level school in every county took part. Male only, female only, and mixed-gender schools were included in the sample. The sample reflected the national distribution of non-disadvantaged and designated disadvantaged schools. A total of 171 schools were included in the sampling frame, 72 of which agreed to take part (42 %). Of the students within these schools, on average, 45 % of students agreed to take part. Response rates of students in different schools varied between 8 and 89 %. Full ethical approval was granted for the study by the authors’ institution. Parent and student written consent was obtained for each participant.

Participants

The second-level sample comprised 6085 students, 4358 of whom provided a self-generated definition of bullying. Of the sample, 54 % were female. The age range of the sample was 12–19 years old (M = 14.99, SD = 1.63). The age of the students was balanced across gender. Of the sample, 43 % were in the 12–14-year-old age range, with the remainder aged between 15 and 19 years old.

Within the Irish school system, students typically complete five to six school years. The Junior Cycle (JC) consists of first, second, and third year. The Senior Cycle (SC) consists of fourth, fifth, and sixth year. Fourth year, or “transition year”, is not mandatory in all schools. During this year, students undertake a programme of vocational and social development (Jeffers 2011). Kenny et al. (2013) stated that as the age range of students within the Senior Cycle can vary considerably, due to the impact of some students electing to complete the transition year, grouping students according to academic cycle may be more appropriate than classifying according to biological age. Therefore, in the present study, school cycle was used as a proxy for age.

Measures

The MWS-SL is a self-report instrument assessing risk and protective factors associated with adolescents’ psychological functioning (see Dooley and Fitzgerald 2013). The survey was deemed to be reliable and valid following a pilot study in a sample of Irish adolescents (Tobin 2009). The present study utilised the measure of bullying from the survey.

Adolescent definitions of and experience of bullying

Participants were asked to complete the sentence “bullying can be described as” in order to identify how adolescents’ defined bullying. Using this definition, adolescents were asked if they had been bullied. This is in contrast to widely used bullying scales, for example, the Olweus Bully/Victim Questionnaire, which provides participants with a definition of bullying (Olweus 1996).

Analysis plan

A content-analysis methodology was considered to be suitable for use with this data, given that it allows for an inductive examination of the emergence of themes from the definitions which adolescents’ provided (Guerin and Hennessy 2002). Themes which emerge from this process can then be deductively applied across the dataset. Stage one of Guerin and Hennessy’s (2002) method specifies that key areas of interest in the data are identified. Namey, Guest, Thiaru, and Johnson (2008) propose that with very large qualitative data sets, such as the one used in this study, frequency analysis of words can be used to identify key areas of interest within the data, e.g. through using data management tools such as NVivo (QSR International 2012). While data management packages can aid qualitative data analysis through providing a transparent view of the content of the data, they cannot replace the ability of the researcher to notice patterns and trends within the data (Hilal and Alabri 2013); thus, a combination of computer-aided data management and analysis by hand was used in the present study (Welsh 2002). The second stage of Guerin and Hennessy’s (2002) method involves inductively investigating the key responses to each area of interest identified in stage one. The third stage involves grouping these responses into thematic patterns and subthemes. These themes and subthemes are then organised in a coding frame, and the coding frame is then used to review the data again, in order to deduct where a particular theme is located within the data.

Results

Data preparation

The units of analysis in the present study were definitions of bullying provided by 4358 second-level students. The total number of words in these definitions was 52,808. In order to prepare these data for analysis, the definitions were spell-checked. This was done to ensure each unit of data could be accurately detected by NVivo (QSR International 2012). As the definitions provided were self-generated, the researchers did not exclude any of the definitions provided from the body of data to be analysed. The data were initially split into two groups, consisting of all the responses generated by Junior Cycle (JC) and Senior Cycle (SC) participants. Within these two groups, the data were further delineated into responses by males (M) and females (F) and responses by bullied (B) non-bullied (NB) males and bullied and non-bullied females (see Fig. 1). As the dataset was large, pertinent definitions were chosen on the basis of containing a word to describe bullying, which was frequently used in the dataset (Namey et al. 2008).

Fig. 1
figure 1

Representation of groups which participants were streamed into and most frequent word(s) within these groups. JC Junior Cycle, JCM Junior Cycle males, JCF Junior Cycle females, JCB Junior Cycle bullied, JCNB Junior Cycle not bullied, SC Senior Cycle, SCM Senior Cycle males, SCF Senior Cycle females, SCB Senior Cycle bullied, SCNB Senior Cycle not bullied

Full size image

Data analysis

Data analysis of the JC and SC groups consisted of two phases: the identification of the word most frequently used to describe bullying and a content analysis of the definitions in which the frequent words were used. To examine the frequently occurring terms used in students’ definitions to describe bullying, a word frequency query was conducted using NVivo (QSR International 2012). Any words consisting of less than two characters, and stop words (conjunctions, prepositions), were excluded from the frequency counts in order to ensure that only meaningful words were included. The process was repeated for definitions provided by males, females, those who had been bullied, and those who had not been bullied, across both JC and SC groups. Identifying the most frequent word used in each of these groups allowed for the identification of key areas of interest, as per stage one of Guerin and Hennessy’s (2002) content analysis methodology. This content analysis methodology was then used to examine the definitions which contained the most frequently used word for each group. This stage of the analysis was conducted by hand. Responses containing the most frequently used word were reviewed through reading and re-reading the definitions provided. The responses of interest were grouped into mutually exclusive thematic categories, themes, and subthemes. A coding frame was then developed and the data were reviewed using this frame.

Inter-rater reliability

In order to maintain rigour and increase confidence within qualitative research, it is suggested that inter-rater reliability checks should be employed (Barbour 2007; Elliott et al. 1999). Therefore, an independent inter-rater was asked to indicate their agreement with the application of a particular code. Percentage agreement was 92 %. A kappa statistic (which takes into account that two raters may disagree or agree by chance; Viera and Garrett 2005) was also calculated. The Kappa statistic for the thematic categories ranged between 0.79 and 0.82. Values in excess of 0.75 are considered acceptable, which suggests that in the present study, a good level of inter-rater agreement was observed (Cicchetti 1994).

Key thematic categories

The definitions provided by JC and SC students, males (M) and females (F), those who had been bullied (B), and those who had not (NB), were compared. The results of the analyses of frequent words within these definitions highlighted that JC students most frequently used the word mean (cruel) to describe bullying. This was observed for JCM and JCF. However, victims of bullying (JCB) differed in terms of how they described bullying compared to those who were not bullied (JCNB). The most frequent word among students in the JC who had been bullied was feel, and the definitions using this word discussed the feelings which being bullied engendered within them, while those who had not been bullied most frequently described how bullying was mean, which was in line with the broader trend among JC students.

As with JC students who had been bullied, SC students most frequently referred to the feelings which bullying could bring about in the victim. This pattern was seen for SCF and those who had been bullied within the Senior Cycle (SCB). SCM were more likely to discuss types of bullying, as were those who had not been bullied among the Senior Cycle (SCNB). The most frequent word used in terms of types of bullying by SCNB and SCM was physical, and definitions containing this word usually named a number of other types of bullying behaviour.

In analysing the patterns in the above data, the results were organised into three broad thematic categories: “bullying is mean”, “bullying affects feelings”, and “types of bullying” (see Table 1).

Table 1 Thematic categories and the groups which these themes emerged in
Full size table

Thematic category one: bullying is mean

Among JC students, the most frequently used word to describe bullying was “mean”. Over a quarter of the definitions provided by JC students (26 %) and JCNB (32 %) made reference to the concept of the mean nature of bullying. Within this thematic category, three themes emerged from the data: “reasons for mean behaviour”, “certain people are at risk of being a victim of mean behaviour”, and “there are certain types of mean behaviour”. These themes were further broken down into subthemes which are listed alongside illustrative quotes in Table 2.

Table 2 Themes and subthemes for thematic category one: bullying is mean
Full size table

In terms of the first theme, “reasons for mean behaviour”, some JC students highlighted that there was no reasonable explanation for mean behaviour. Others emphasised that bullying may on occasion be carried out on purpose to provide a source of humour to the bully, because the bully may be “unsatisfied with life” (JCM) in some way or in order that the bully may “feel good about themselves” (JCM).

A second theme which was identified in definitions containing the word mean was that certain individuals were at risk of being treated in a mean way. JC students highlighted that individuals who may differ from expected in terms of their behaviour, appearance, or ability may be at risk of experiencing mean behaviour towards them. For example, one JCF participant stated that mean behaviour was carried out towards others because “they look or sound different, or because they don’t like the way that person acts”. Additionally, being of a different race, or being academically able, was also highlighted as a characteristic that participants highlighted as potentially putting individuals at risk of victimisation.

A third theme to be identified among definitions containing the word mean was that there were certain types of mean behaviour. Physical bullying (hitting, kicking, and punching) was described as mean behaviour, while “sending mean texts” (JCM) and “saying mean things” (JCNV) were frequently mentioned. Others cited “ignoring” (JCF) and isolating others as the types of mean acts which bullies might carry out.

Thematic category two: bullying affects feelings

SC students, SCF students, and those who had previously experienced bullying (both JC and SC) referred to the word “feel”, and its derivative “feeling”, when asked to share what they thought about bullying (see Table 3). Specifically, feelings were referenced in 43 % of SC students’ definitions, 34 % of SCF students’ definitions, and 56 % of the definitions provided by victims of bullying. Two broad themes were isolated: “victims of bullying feel negative emotions” and “bullies feel positive and negative emotions”. These broad themes were further delineated into subthemes which are presented in Table 3 alongside illustrative quotes.

Table 3 Themes and subthemes for thematic category two: bullying affects feelings
Full size table

The first theme in this category comprised of the range of negative feelings that were described; participants cited that being bullied could make someone “feel worthless”. Participants also described how bullying could make victims feel “sad”. In some cases, this sadness could be felt regardless of whether the bullying was of a verbal, relational, or physical nature as they could “all hurt you in the same way” (SCF). Many participants also referred to the feelings of upset which bullying victimisation could create. According to some participants, this feeling of upset could arise when a bully victim’s sense of self was disrupted. For example, one female participant stated that bullying could make you “feel upset about you as a person” (SCF). Furthermore, SC students also stated that bully victims could feel anger not only at themselves but also at the bully. Participants also referred to feelings of isolation and how the nature of bullying could make an individual feel “like they don’t belong or fit in” (SCF) or “not accepted” (SCBV).

A second theme which was identified centred on how SC students also used the word feel in the context of the feelings the bully may have. Frequently, participants described how bullying may make the bully feel “good about themselves by feeling higher than someone else” (SCF). According to participants, bullies may also carry out bullying behaviour as they “have issues” (SCV) or were “sad inside” (SC). One participant proposed that for the bully, witnessing the victims’ “pain and fear makes it all better” (SCF).

Thematic category three: types of bullying

Among SCM and SCNB students, participants most frequently described types of bullying when asked to describe bullying. Specifically, 45 % of these participants referenced physical bullying, 34 % referenced mental/psychological bullying, 20 % referenced verbal bullying, and 1 % referenced cyberbullying. Illustrative quotes are provided below.

Physical bullying

Participants stated that physical bullying could involve “hitting people” (SCNB) and “pushing, kicking, punching” (SCM). Physical bullying was considered to be a form of intimidation so bullies could “get what they want” (SCM).

Mental and psychological bullying

“Mental bullying” was described as “extremely damaging” (SCNB) and could make an individual “feel worse about themselves” (SCNB). Mental bullying was also described as being “mentally hard on the victim” (SCM), as they were being subjected to behaviour which was “psychologically degrading” (SCM). One participant described mental bullying as “messing with your head” (SCM).

Verbal bullying

Participants described verbal bullying as “calling someone names” (SCM), “slagging them a lot” (SCM), or “insulting someone and disrespecting them” (SCNB).

Cyberbullying

The least discussed form of bullying was cyberbullying. Some participants included texting in their definition, e.g. “even cyberbullying can hurt in the form of texts or [on] social websites” (SCM). “Facebook” (SCM) was mentioned as one platform through which cyberbullying could occur, as was text messaging.

Results summary

To summarise, participants differed in terms of their definitions of bullying due to previous experience of bullying, age, and gender. Those within the JC focused on how bullying was mean. In contrast to this SC students (SC females in particular), those who had previously been victimised emphasised the negative consequences of victimisation on the psychological wellbeing of those who experienced it, alongside an understanding of the emotional mindset of bullies. SC students who had never been victimised and SC male students frequently described types of bullying (physical, verbal, psychological, electronic forms).

Discussion

The aim of this study was to examine the themes that emerge from adolescents’ unprompted definitions of bullying and evaluate if these definitions differ based on age, gender, and previous experience of bullying. As previous literature has already shown that adolescents’ definitions do not frequently contain key elements of research definitions of bullying (Cuadrado-Gordillo 2012; Frisén et al. 2008; Guerin and Hennessy 2002; Naylor et al. 2006; Vaillancourt et al. 2008), our study, using an inductive methodological approach, also failed to find evidence supporting the defined themes of repetition, intentionality, or power imbalance, as conceptualised by researchers in the field (Farrington 1993; Gladden et al. 2014; Olweus 1999).

Key findings

Three thematic categories emerged from the data: (a) bullying is mean, (b) bullying affects feelings, and (c) types of bullying. Within the bullying is mean thematic category, definitions reflected reasons for mean behaviour (e.g. the behaviour is humorous to the bully, a bully hurts someone because they are hurt themselves, there is no reason for mean behaviour), those at risk for being treated in a mean way (e.g. those who look and sound different, those who are weak, those of a different race, physical appearance, gender, those who are smart), and types of mean behaviour (e.g. ignoring, isolating, physically hurting others). Within the bullying affects feelings thematic category, definitions reflected victims of bullying feeling negative emotions (e.g. worthlessness, sadness and upset, anger, isolation), and bullies feeling both positive and negative emotions (e.g. bullying to feel good about themselves or because they are sad inside). Finally, within the types of bullying thematic category, participants referred to physical bullying, mental/psychological bullying, verbal bullying, and cyberbullying.

However, the frequency at which elements of these thematic categories were referenced differed according to age, gender, and experience of bullying, in line with previous research (Monks and Smith 2006; Frisén et al. 2008; Vaillancourt et al. 2008).

Regarding age, the definitions which Junior Cycle (JC) students provided usually involved listing a number of adjectives to describe bullying, the most common of these being mean (26 %), reflecting the thematic category bullying is mean. In comparison, Senior Cycle (SC) students tended to discuss the feelings associated with bullying, reflecting the thematic category of bullying affects feelings—feelings were referenced in almost a quarter of SC adolescents’ definitions (24 %). They also highlighted the positive and negative feelings which a bully could experience when engaging in victimisation. In comparing these groups, we can see that the definitions provided qualitatively differed, moving from a focus on describing what bullying is conceptualised as in the JC, to focusing on the effects of bullying in the SC. Monks and Smith (2006) state that these contrasts could be attributed to cognitive differences, whereby older, more cognitively advanced adolescents can conceptualise bullying along a number of dimensions compared to younger students. For example, many older adolescents were able to incorporate a description of the emotional effects of bullying into their definitions, while this was less evident for younger adolescents. In the present study, we could attribute the differences between JC and SC adolescents’ definitions to cognitive advancements.

However, it may also be a function of the education about bullying victimisation which students received, given that SC students may have had more opportunity to engage with education programmes compared to their younger counterparts. Therefore, definitively saying that cognitive development can account for differences in adolescents’ definition of bullying is problematic (Scheithauer et al. 2012). However, the results of the present study suggest that it is important for researchers to consider that older and younger students may differ in terms of their definition and therefore their understanding of bullying.

In exploring gender differences, male and female definitions differed, with SC females describing the effects of bullying, and SC males describing the types of bullying which could be experienced. This partly supports previous research which states that females may focus on the effects upon the victim (Frisén et al. 2008). In contrast, SC males were more likely to describe types of bullying, which could be attributed to the finding that males are more likely than females to minimise the expression of emotions (Perry-Parrish and Zeman 2011), and therefore they provided more descriptive definitions of bullying.

In comparing adolescents who had experienced bullying and those who had not, victims from both JC and SC focused on the emotional aspect of victimisation. This suggests that those who have personal experience of victimisation may be more aware of the serious negative emotional consequences associated with being bullied. Although Monks and Smith (2006) state that the definition of bullying may change with increased experience of observing bullying behaviour, few studies have explicitly addressed whether there is a difference in the definitions provided by non-bullied versus bullied adolescents (Naylor et al. 2001).

In sum, our findings highlight that the influence of age, gender, and experience of bullying should be considered by researchers in the measurement of bullying. Currently, this practice is not regularly used within bullying research.

Strengths and limitations

In terms of strengths, this study used a large, nationally representative dataset in Ireland to examine adolescents’ definitions of bullying. Unlike previous studies that only include responses from adolescents up to the age of 13 or 14 (Frisén et al. 2008; Guerin and Hennessy 2002; Naylor et al. 2006; Menesini et al. 2002; Monks and Smith 2006; Smith et al. 2002), we examine definitions of adolescents aged 12–19 to examine age-related, gender-related, and experience-related differences in conceptualisations of bullying. In terms of analyses, we did not impose thematic categories relating to researcher definitions of bullying, but rather allowed themes to emerge inductively from the data. Regarding methodology, a free-response method was employed to ascertain unprompted definitions of bullying as opposed to asking participants to agree or disagree with a list of statements (e.g. Cuadrado-Gordillo 2012). Research with adolescents is increasingly moving towards incorporating the voice of the young person (McDonagh and Bateman 2012), and allowing the participants in the present study to contribute their own meaningful definition of bullying is one way in which the voice of the young person can be captured in research.

In examining the study’s limitations, data were cross-sectional in nature. Longitudinal data, comparing the same participants’ definitions of bullying over time (e.g. Frisén et al. 2008), would allow for more confirmatory conclusions to be reached regarding developmental differences in younger versus older adolescents’ conceptualisations of bullying. Future research may examine how definitions change over time based on age, gender, and experience of bullying. The aim of the present study was to examine explicit meanings within adolescent definitions of bullying. As a result, implicit meanings were not identified. Future research may be required in order to examine implicit meanings within adolescent definitions.

Implications and conclusions

The findings of this study have implications for future research and future intervention design; they raise significant questions concerning the assessment of bullying in psychological studies. While researchers’ definitions of bullying typically emphasise power imbalance, repetition, and intention in their definitions, these concepts were not explicitly identified in the present study as adolescents’ definitions most frequently referred to the mean nature of bullying, feelings associated with bullying, and types of bullying. Our findings add support to the conclusions of Vaillancourt et al. (2008) that “it is no longer tenable to assume that students’ spontaneous definition is in keeping with that purported by the research field” (p. 494). However, one could argue that while the explicit concepts outlined in researchers’ definitions may not be explicitly present in adolescent definitions, it is possible that they may be implicitly present in adolescent definitions. In-depth qualitative research would be needed to explore this. While we recognise that consistent terminology with standardised definitions is necessary to improve public health surveillance of bullying and inform efforts to address bullying (Gladden et al. 2014), we suggest, in line with other researchers, that future studies examining the prevalence of bullying in school should consider the perspective of the adolescent in addition to the researcher (Guerin and Hennessy 2002; Naylor et al. 2006). It may also be hypothesised that researcher and adolescent definitions may not always be at odds with each other and may perhaps be complimentary, with adolescent definitions explicitly describing bullying behaviours, and researcher definitions implicitly examining the nature of these behaviours.

These inconsistencies are important to consider regarding education, intervention, and prevention programmes for adolescent bullying so that these programmes are informed by the experiences of adolescents and their conceptualisations of bullying behaviour. As aforementioned, adult and teacher definitions of bullying tend to align with researcher definitions more than adolescent definitions (Menesini et al. 2002; Mishna et al. 2005; Naylor et al. 2006). This is important to highlight as such discrepancies may prevent adults from intervening in situations considered by adolescents as bullying (Frisén et al. 2008; Mishna et al. 2005). Additionally, lack of intervention from the school staff for situations regarded by adolescents as bullying (e.g. “mean” behaviour) may be seen by adolescents as an acceptance of such behaviour (Madsen 1996). In line with Naylor et al. (2006), we recommend that teachers are made aware that adolescents, depending on gender, age, and experience of bullying, may have different conceptualisations of bullying compared to teacher perspectives.

Our results indicate that bullying programmes should be tailored depending on whether they target younger or older pupils and that the content of such programmes should reflect the differences in how these groups conceptualise bullying. Also, given the observed difference between bullied and non-bullied adolescent definitions (particularly regarding the emotional impact on the victim), it may be useful for educational programmes to involve the use of vignettes, personal experiences, and role-playing to increase awareness of the negative effects of bullying (Pepler et al. 1994). This may allow students who have not experienced bullying to appreciate the emotional impact of victimisation. However, it is worth noting that the effectiveness of certain anti-bullying prevention and intervention programmes has been questioned (Ferguson et al. 2007). Therefore, educators should carefully consider the type of behaviour they are trying to target and the aims of a programme before implementing it in their school, being mindful that “one size” does not fit all, and putting young people at the centre of an intervention.