Bullying is a serious problem for millions of adolescents worldwide (Volk et al., 2006). This makes its study and prevention an important priority for the social sciences (O’Higgins & Hinduja, 2019). Unfortunately, while some interventions have had a modest degree of success in reducing adolescent bullying prevalence, bullying has proven to be a behavior that is strongly resistant to interventions (Gaffney et al., 2019). One of the challenges in understanding bullying has been a fundamental challenge in universally defining bullying (Jia & Mikami, 2018; Volk et al., 2017). Researchers have proposed different definitions (e.g., Olweus, 2013; Thomas et al., 2015), while adolescents have applied their own views to defining bullying that often do not take the key elements of research definitions into consideration (Jeffrey & Stuart, 2019; Patton et al., 2017; Vaillancourt et al., 2008). These different definitions hamper both research and interventions as they blur the lines between bullying and other kinds of aggression (Volk et al., 2017). Furthermore, various quantitative methods of measuring bullying, including self-report, peer-nomination, teacher-rating, and parent-rating measures have relatively small inter-correlations, which illustrates the need to better understand differing perspectives (Volk et al., 2017). Efforts to adequately study the key elements in the definition of bullying have been hindered by the fact that some of its underlying concepts are quite difficult to measure using common quantitative methods. A qualitative approach allows for researchers to address and understand the emotions, experiences, and motivations of the participants (Lurhmann, 2006). This depth helps researchers to translate and unify the different perspectives provided by different agents (Mishna, 2004). Furthermore, quantitative research requires a priori knowledge of concepts that will be measured, whereas qualitative allows researchers to explore phenomenological aspects that might be unknown (Cypress, 2015).

For all of these reasons, we argue that the field of bullying research would benefit from a greater utilization of qualitative methods that supplement our quantitative understanding of what bullying is (and is not). To illustrate how qualitative methods can fill gaps in the bullying literature not addressed by quantitative methods, we will focus on the differential application of quantitative and qualitative methods to the measurement of bullying in relation to a relatively recent definition of bullying that is rooted in evolutionary theory (Volk et al., 2014). The definition states that “bullying is aggressive goal-directed behavior that causes harm to another individual within the context of a power imbalance.” Rather than conducting a thorough review of all of the evidence for that definition (e.g., see Kaufman et al., 2020 for a quantitative analysis), our goal is to use that definition as an example to highlight when, where, and how qualitative methods can be fruitfully applied towards a better understanding of bullying. Qualitative research has shown that concepts such as intentionality, goals, power imbalances, and harm are all challenging to adequately capture using quantitative methods (Hellström et al., 2015; Jeffrey & Stuart, 2019; Spadafora et al., 2020). While acknowledging that qualitatively based works from the last two decades have proposed other definitions and forms of revisions, we will focus on how existing and future qualitative research can improve our understanding and quantitative measurement of three separate aspects of that definition: power imbalance, goal directedness, and harmful impact (Volk et al., 2014).

Balance of Power

As noted by Vaillancourt and colleagues (2003), “bullying is power.” Power allows the bully to engage in conflict with a victim that they are confident in prevailing against (Veenstra et al., 2010; Volk et al., 2012). This allows the bully to gain the benefits (i.e., goals; see the following section) of the behavior while experiencing a reduced cost due to retaliation or failed aggressive attempts. In general, quantitative measures of bullying ask respondents to rate the frequency of aggressive behavior in which the perpetrator has more power than the victim, generally in the form of physical strength, popularity, or strength in numbers (Olweus, 2013; Thomas et al., 2015). Power is a challenging concept to define, given that it can vary in its form, intensity, and context (Wrong, 2017). Indeed, many researchers have argued that understanding and wielding social power is what led to the rapid expansion of human intelligence over evolution (Sternberg & Kaufmann, 2013); thus, it is not surprising that it is a complex construct. Quantitative approaches to measuring power balances in bullying take two general forms: definition-based and behavior-based approaches (Thomas et al., 2015). Definition-based approaches, as the name suggests, first provide respondents with a definition of bullying, including the imbalance of power favoring the bully. The most widely used definition-based measure, the Olweus Bully/Victim Questionnaire, refers to the imbalance of power by stating that “it is difficult for the student being bullied to defend himself or herself” and “it is not bullying when two students of about the same strength or power argue or fight” (Solberg & Olweus, 2003). Olweus adopted a deliberately qualitative approach to power that depends on the subjective beliefs and contextual capacities of the victim relative to the bully rather than relying on a purely quantitative approach (e.g., who was bigger, who had more friends, etc.; Olweus, 2013). This qualitative nature of bullying led Olweus to emphasize the importance of self-report measures of bullying (Olweus, 2013). Keeping this definition in mind, respondents who are intimately involved with the behavior self-report the frequency of the bullying behavior, often in terms of specific subtypes of bullying such as: physical (e.g., hitting, and kicking), verbal (e.g., insults and threats), social (e.g., social exclusion and spreading malicious rumours), and cyber (e.g., posting hurtful pictures, videos, comments) forms (Solberg & Olweus, 2003).

Behavior-based approaches adopt a more quantitative approach by asking about a series of specific behaviors related to bullying or victimization that ideally also reference the imbalance of power in the behavioral description for each item, with relatively elementary descriptions to indicate specific aspects of power such that the perpetrator was “more powerful,” or “stronger or more popular” (e.g., Book et al., 2012). That said, many behavior-based measures of “bullying” contain no references to the imbalance of power, nor any other aspect of the definition of bullying, but merely solicit frequency ratings for a range of aggressive behaviors (e.g., Land, 2003; Thomas et al., 2015; see Hawley et al., 2011; Salmivalli & Peets, 2009).

Quantitative approaches permit a variety of statistical analyses (e.g., correlations, regression analyses, structural equation modeling, social network analyses) that can be used to examine cross-sectional and longitudinal relations with a wide spectrum of psychosocial or biological variables, to test hypotheses about the correlates, antecedents, consequences, and social contexts of bullying and victimization (e.g., Reinjtjes et al., 2018; Ttofi et al., 2012). However, as noted by Olweus (2013), many nuances about the balance of power in bullying are difficult to consider using these methodologies. For example, beyond strength and popularity, are there other sources of power that contribute to power imbalances in different forms of bullying? Does popularity matter more in the context of social or relational than physical bullying? How do you operationalize balance of power in the context of cyberbullying; does this vary depending on whether the perpetrator is known or anonymous? Clearly, balance of power is an aspect of bullying that has proven challenging to adequately capture with current quantitative methods (Hellström & Lundberg, 2020).

Qualitative methods may help to answer some of these questions. Educators and parents have expressed in interviews that they have often found it difficult to determine which behaviors were bullying, as they reported difficulty in determining if an incident included a power imbalance (Mishna, 2004). Nevertheless, qualitative interviews with these external observers may yield important insights into behavior that adolescents are not aware of or willing to report (Sawyer et al., 2011). Indeed, adolescents may deliberately withhold information from adults (Mishna et al., 2008), particularly for bullying behavior that is subject to adult sanctions (Volk et al., 2014) or reports of victimization that victims may fear cause them to appear weak and unable to solve their problems (DeLara, 2012). A common example of mismatches in adult and adolescent reports is when adults (teachers and/or parents) interpret a victim’s lack of retaliation as due to a lack of motivation rather than an inability to do so because of a power imbalance (DeLara, 2012; Sawyer et al., 2011). The discrepancy between these two sources (adults versus adolescents) would be difficult to parse using purely quantitative data, whereas qualitative data can allow us to determine potential causes of different views.

The qualitative use of focus groups enables adolescents to provide richer accounts of the individual and social dynamics of bullying (Guerrea et al., 2011) that can allow researchers to differentiate bullying from other forms of aggression (Shute et al., 2008). This approach would be particularly beneficial for a further investigation of power operating at both individual and social levels (Wrong, 2017). Researchers could not only ask participants about who has power and what aspects or abilities they consider to be associated with power (e.g., strength, popularity, likability), but also ask why adolescents felt certain peers had this power, and how they used it within the peer context. For example, Guerra and colleagues (2011) noted that younger focus groups tended to focus on individual and physical power differences, whereas older adolescent groups focused more on social power as being related to bullying. These differences likely relate to the developmental shift away from physical and towards social forms of bullying that occur over the course of adolescence (Volk et al., 2006). Thus, focus groups can not only shed light on descriptive and developmental aspects of social power, but also illustrate how individuals respond to those perceptions of power.

It is particularly challenging to conceptualize power imbalances in the context of cyber bullying, and there have been questions raised about how this type of bullying fits with the characteristics in definitions of traditional bullying (Dennehy et al., 2020; Englander et al., 2017). Definitions have recently been modified to ensure the components apply equally to traditional and cyberbullying (e.g., Moreno et al., 2021; Volk et al., 2014). However, given the novelty and complexity of the latter form of bullying, researchers have advocated that it is essential to engage with young people who have typically used digital communication and social media all of their lives and can therefore provide a novel perspective on their experiences with cyberbullying that might elude adult researchers who are less familiar with the online environments (Cross et al., 2015). Regrettably, traditional definitions of bullying have nevertheless often been applied in the measurement of cyberbullying without consulting youth about their views, resulting in a confusing lack of consensus about the definition and measurement of this cyberbullying (Dennehy et al., 2020). For example, focus groups with youth suggest that the potential anonymity of the perpetrator subverts traditional notions of power imbalances in bullying so that relatively powerless youth in face-to-face contexts can engage in bullying in a cyber context with less fear of retaliation (e.g., Jacobs et al., 2015). They further revealed that anonymity makes victims powerless to defend themselves and susceptible to considerable harm (e.g., Baas et al., 2013). To further complicate matters, quantitative (e.g., Hamm et al., 2015) and qualitative research (e.g., Baas et al., 2013; Jacobs et al., 2015; Mishna et al., 2009) agree that cyberbullying is usually perpetrated by known rather than anonymous assailants, so that the relative offline power of the perpetrator and victim could be assessed. Taken together, this research suggests that power imbalances remain a definitional feature of cyberbullying, but that further qualitative research is needed to reconcile differing views about power in this context, especially to inform measurement of cyberbullying with both anonymous and known perpetrators.

In individual interviews with adolescents, a similar picture emerges of bullies being concerned with gaining and/or maintaining their high social status and power (Thornberg & Derby, 2019). Again though, there have been discussions about the permanency and/or generalizability of a bully’s power advantage that relate to both the individuals involved as well as bystanders’ willingness to become involved (e.g., should the victim’s friends arrive to support them; Spadafora et al., 2020; Thornberg et al., 2018). This is further complicated by the heterogeneity of the types of power differences reported. In retrospective interviews, Oblath and colleagues (2020) report several factors that have influenced a power imbalance: victims’ low self-esteem, fear of making a situation worse, a lack of external supports, the characteristics of the bully (e.g., size, popularity), a previous or current close relationship with the bully that made retaliation difficult or undesirable, and the inability to control one’s emotions under pressure. These kinds of qualitative data are invaluable, not only for identifying different risk factors that signal vulnerability to victimization, but for revealing the challenge of designing victim-oriented interventions such as social skills programs intended to mitigate risk factors that promote vulnerability (e.g., Da Silva et al., 2016).

Goal Directedness

The second definitional criterion that we will focus on is goal-directedness or intentionality. Given that goals can reflect both conscious and unconscious motives (Dijksterhuis & Aarts, 2010), a focus on goals allows one to avoid ascertaining intentionality that individuals or observers may, or may not, be aware of (Volk et al., 2014). For example, one can lash out at an unprovoked victim without recognizing that one is in fact displacing anger better directed at someone else. Similarly, engaging in conscious aggression (e.g., verbal insults) for “fun” might be the result of actual, but unconscious, goals related to sexual competition or jealousy (Pellegrini, 2002). Relying solely on conscious intent can therefore distort or miss actual motives (Vaillancourt et al., 2003). Further, different goals can help to distinguish between bullying and other forms of aggression (Volk et al., 2014). According to Olweus (2013), the intention to harm another person is a critical feature differentiating bullying from accidental behavior that causes harm. For example, an individual may unintentionally behave in a rude or uncivil fashion (e.g., eating loudly during class) without intending to cause harm to another person (Spadafora & Volk, 2021). Most bullying research to date has focused on differentiating between an intent to hurt/be mean versus accidental aggression (Jia & Mikami, 2018). However, along with issues of harm (that will be discussed below), this sort of intentionality is difficult to capture.

To circumvent these issues surrounding intentionality, Volk and colleagues (2014) assumed intentionality by specifying that bullying was goal-directed aggression, as goals necessarily require intentionality (Muñoz et al., 2014), but do not necessarily require explicit conscious awareness (Dijksterhuis & Aarts, 2010). They suggest three specific goals of bullying from an evolutionary perspective (the Three R’s): Resources, Reproduction, and Reputation. According to this definition, bullying is predominantly proactive aggression that is used in a premeditated and planned fashion to pursue a range of evolutionarily relevant goals, including the procurement of resources, status seeking, negotiation of dominance hierarchies, intersexual selection (displaying aggression to attract mates), and aggression deterrence (Volk et al., 2012, 2014). Although those who bully also aggress impulsively, emotionally, and reactively (Book et al., 2012; Runions et al., 2018), reactive aggression is generally not an act of bullying (Salmivalli & Peets, 2009). Therefore, a key element of the goal-directedness aspect of bullying is shifting the focus beyond an intent to harm to consider a wide spectrum of goals that are intentionally and strategically pursued by bullies.

As the goal-directedness aspect of bullying is not included in several definitions of bullying (Gladden et al., 2014; Solberg & Olweus, 2003), this facet is not directly assessed in definition-based and behavior-based quantitative measures of bullying per se. Instead, research on the goals of bullying has proceeded along two lines. First, researchers have employed measures to assess proactive (e.g., status seeking, procurement of resources) and reactive (provoked, emotional, impulsive) functions of aggression using behavior-based self-report, peer-nomination, and teacher-report measures (e.g., Marsee et al., 2011; Raine et al., 2006) that they have used to investigate concurrent and longitudinal associations with quantitative measures of bullying, generally finding that bullying is more consistently and strongly associated with proactive aggression, though often correlated with both functions (Runions et al., 2018; Volk et al., 2014). Another approach has been to examine concurrent and longitudinal relations between quantitative definition-based or behavior-based measures of bullying and measures related to evolutionary relevant goals, including peer-nominations of popularity and resource control, and self-reported dating and sexual relationships (Dane et al., 2017; Pronk et al., 2017; Reinjtjes et al., 2013, 2018; Van der Ploeg et al., 2020; Volk et al., 2012, 2015).

Although these studies have provided empirical evidence that bullying is goal-directed aggression that may be used to pursue several evolutionarily relevant goals, they do not offer in-depth information about adolescents’ perceptions of the utility of bullying. Qualitative research may also be used to examine key questions that remain unaddressed. A deeper qualitative investigation to identify a broad range of goals would also enhance our understanding of important aspects of the context and social ecology of adolescent bullying and allow for the study of motives that might not be readily available in response to brief, constrained, quantitative responses. For example, bullying for resources has not been well-studied by quantitative methods despite the popular stereotype of bullying for lunch money. As an illustration, Hawley’s Resource Control Theory mentions both physical and social resources, but generally fails to measure the former or discuss how those different kinds of resources may involve different adolescent behavior or strategies (Hawley, 2003, 2015). Bullying for private physical resources (e.g., lunch money) may not require or benefit from the audience (Veenstra et al., 2010) that is typically associated with bullying for social status (Volk et al., 2014). Qualitative analyses have revealed competition over resources as varied as food (Strindberg et al., 2020; Turnbull, 1987) and academic recognition (Flanagan, 2007; Maestripieri, 2012), but these concepts have not been widely measured (if at all) by experts using quantitative methods. Thus, quantitative research into the goals of bullying have left important gaps both between existing goals as well as the possibility of overlooked goals.

Do adolescents use bullying for purposes not considered by researchers? Recently, research has argued that bullying is used sadistically, for no explicit purpose other than fun and recreation (Runions et al., 2018). Some investigators have suggested sadistic aggression may serve several ultimate purposes beyond the immediate experience of enjoyment, including status or dominance seeking and revenge (e.g., Pinker, 2011). Indeed, qualitative interviews have revealed that revenge motives appear to be an important factor for bullying (Thornberg & Delby, 2019). Their presence calls for revisiting how the three goals of bullying laid out by Volk and colleagues (2014). Their “Three R’s” were motivated by similar categories (the Four F’s) of behavior common to all animals, including Feeding (Resources), Fighting (Reputation), and Fornicating (Reproduction; Pribram, 1960). However, a fourth aspect of behavior, Fleeing, was left out as it was viewed as more reactive/defensive (Volk et al., 2014; see discussion above). Perhaps the definition of goal-directed bullying should be broadened to include Revenge as the “Fourth R”, whereby bullies use aggression defensively to maintain their social status (Thornberg & Delby, 2019). Focus groups on cyberbullying provide additional evidence to this effect, noting that it is often motivated by revenge, including retaliation for in-person bullying (e.g., Jacobs et al., 2015; Mishna et al., 2009). The potential expansion of the goals of bullying to include a fourth goal is a reminder of the value of bottom-up/grounded qualitative research in revealing patterns that might be missed using a purely theory-driven approach.

One exception may be adolescents who are in the role of “bully-victims,” that is, adolescents who are bullied and experience harm and then engage in bullying a means of coping with their own victimization. Unlike adaptive bullying that is the focus of the current manuscript, bully-victim status is not associated with social dominance and power and these adolescents tend to be unpopular and rejected (e.g., Marini et al., 2006). Instead of engaging in goal-directed behavior, bully-victims are more likely to respond using reactive aggression and their actions are usually defensive or due to an emotional response (e.g., Salmavalli, 2010). Regardless of motive, qualitative methods can be a way for researchers to better contextualize youths’ experiences through understanding their environment or relationships from their own personal perspectives.

There are several qualitative methods that might be particularly useful in furthering our understanding of adolescent goals related to bullying. The use of hermeneutics (i.e., a focus on textual/content meaning; Chan et al., 2020), particularly in combination with individual diaries or autoethnographies (i.e., self-reflections of the researcher; Berry, 2016), could yield a greater depth of information about bullying used to pursue goals related to reputation, resources, reproduction, and revenge. Hermeneutics’ focus on interpretations of symbols, language, or cultural artifacts could lend considerable weight to our understanding of how bullying goals are defined and interpreted at both the individual and cultural level. The use of diaries and autoethnographies would provide the necessary material to help ensure that the broadest range of experiences are reported. Previously, diaries have typically been used to measure the frequency of bullying, but not its goals (Nishina & Juvonen, 2005; Pellegrini & Long, 2002). These underutilized methods could allow adolescents to not only report their daily experiences with bullying, but also why they chose to engage in various behavior. This should be of value in adolescence, particularly in older adolescents who are better able to accurately record and use diary entries to promote deliberate introspective reflections on their thoughts and actions (Hunt et al., 2015), thereby revealing how closely adolescents’ behavior actually relate to the four goals mentioned above.

Younger adolescents’ elaboration of goals may be scaffolded by the use of individual or group structured (tight control over questions order and content) or semi-structured (looser control) interviews as researchers have shown that it may be more challenging for younger children or adolescents to correctly identify and understand their (and others’) motives with respect to aggressive behavior (Vaillancourt et al., 2008). While not focusing on goal directedness, these studies may nevertheless shed light on motives and goals such as a focus on popularity or social dominance. This goal appears as prominently in qualitative literature (Cuadrado-Gordillo, 2012; Forsberg & Thornberg, 2016; Gamliel et al., 2003; Gumpel et al., 2014; Jeffrey & Stuart, 2019; Thornberg & Delby, 2019) as it does in quantitative literature, suggesting convergent validity for its importance. Semi-structured group (Strindberg et al., 2020) and individual (Gamliel et al., 2003) interviews have shed light on how these popularity-related goals play out in social groups to influence the role (e.g., bully, reinforcer, victim) students take on during bullying incidents, providing motivational details that are generally lacking in quantitative data. Further, Søndergaard (2012) discusses adolescents being motivated by a fear of social exclusion, and a need to belong to the group when it comes to bullying behavior. Adolescent interviews also reveal the importance of social contagion (Gumpel et al., 2014) and reinforcement (Strindberg et al., 202) on the adoption and pursuit of social dominance goals. These same interviews demonstrate the importance of local (e.g., teacher or classroom level) and broader cultural norms in determining how acceptable bullying behavior is and what roles individuals tend to adopt (Gumpel et al., 2014; Wójcik & Mondry, 2020).

Whereas quantitative methods allow participants to select or rate motives that explain why they chose to engage in bullying, qualitative methods allow for the participant to more fully explain why they chose to engage in certain behavior, and what they felt they could potentially gain or lose as a result. This freedom of elaboration may aid the recollection of motives that are more unconscious in nature (Kahneman, 2011). Adolescents seem to be cognizant of their motives of engaging in bullying or bystander behavior, and able to articulate them quite clearly when asked open-ended survey questions (Spadafora et al., 2020; Strindberg et al., 2020; Thornberg et al., 2018). Further, the use of these questions on an anonymous survey may encourage adolescents to be honest in their motives of engaging in bullying behavior, compared to other qualitative methods (e.g., interviews) that occur face to face. Open-ended survey questions can be a good first step in a study in order to determine what is important to study in future studies that may use interviews or focus groups (Adams & Cox, 2008).

Focus group interviews have highlighted the importance of social dynamics and norms in promoting an emphasis of “insider versus outsider” as motives for engaging in bullying (Lyng, 2018; Strindberg et al., 2020). In combination with qualitative interview data, these focus groups show that “othering” of victims can be a way of moderating the impact of the bullying to preserve the social status of the bully, making it acceptable to signal a propensity towards using bullying to intimidate outsiders rather than members of the bullies’ group (Forsberg et al., 2014; Strindberg et al., 2020; Thornberg & Delby, 2019; Wójcik & Mondry, 2020). This victim isolation would also enable bullies to achieve benefits such as status or aggression deterrence while reducing the risk of costs to the loss of peer affection (Veenstra et al., 2010; Volk et al., 2012). Along similar lines, structured interviews of adolescents acknowledge that bullying need not be intentionally harmful, thereby implicitly recognizing the importance of instrumental goals other than an intent to harm (Jeffrey & Stuart, 2019). Instead, qualitative methodologies have uniquely revealed that when they are allowed time and space to fully explain their behavior, adolescents often acknowledge explicit cost–benefit calculations regarding engaging in bullying-related behavior that are often independent of victim harm, including: school sanctions, suspensions, peer retaliation, peer reputational loss, parental punishment, and even legal consequences (Jeffrey & Stuart, 2019; Søndergaard, 2012; Spadafora et al., 2020; Strindberg et al., 2020). When harm is mentioned in the context of motivation, it is typically used as a way of denying culpability by bullies who express a lack of desire to cause harm (Gamliel et al., 2003).

Research showing perpetrators to engage in “othering” victims and denying harm doing supports Volk and colleagues (2014) contention that bullying is often an act of social signalling (Van der Ploeg et al., 2020), where the victim’s experience is less important to the bully than the signal that the bullying sends to bystanders. Particularly if directed at children who are viewed as outsiders, it suggests that bullies are trying to send a signal about their capacity for aggressive dominance while simultaneously reassuring in-group peers that they are not incapable of cooperating with in-group members and not inclined to harm them. Thus, qualitative research has helped to differentiate harmful impact on the victim, which we will consider in the section below, from the goals pursued by bullies, which both quantitative and qualitative research has shown to be complex, numerous, and potentially related to adaptive outcomes. (Volk et al., 2014).

With regard to assessing goals, the greater ability of qualitative methods to rely on grounded theory (i.e., theory emergent from the data; Forsberg et al., 2014) has allowed for the more open discovery of new goals associated with bullying, particularly if it can be then be compared to our a priori understanding (Thornberg, 2011) of how bullying is a cost–benefit driven behavior (Volk et al., 2014). Qualitative methods clearly afford important opportunities for better understanding the goal-directed nature of bullying. What do they offer for understanding the harmful impact of bullying?

Harmful Impact

The third component of the Volk et al. (2014) definition of bullying focuses on whether it has had a harmful impact on the victim. In his original definition, Olweus (1993) states that “a student is being bullied or victimized when he or she is exposed, repeatedly and over time, to negative actions on the part of one or more other students”. With regards to repetitiveness, Olweus (2013) said that he “never thought of this as an absolutely necessary criterion” (p.757) as he included it to avoid classifying trivial, relatively unharmful incidents as bullying. The emphasis on harmful impact in the definition by Volk et al. (2014) is a more direct way of determining the severity of bullying, an aspect that is implied by Olweus’ reference to “negative actions” by the perpetrators of bullying, and which was the central reason that Olweus differentiated repeated from single acts of bullying (Solberg & Olweus, 2003). Specifically, Volk and colleagues (2014) contended that aggressive acts do not necessarily have to be repeated to qualify as bullying as individual acts can have serious (e.g., lethal; QMI Agency, 2013) consequences and it may in fact be in a bully’s interest to bully less often to avoid being caught (e.g., Flanagan, 2007). Furthermore, there is no universal standard as to how often a behavior needs to be repeated to be considered bullying and it is a difficult standard to measure with cyberbullying (Solberg & Olweus, 2003). Qualitative interviews of adolescents suggest that repetition is not necessarily a core feature of bullying (Land, 2003), though both qualitative and quantitative studies (Kaufman et al., 2020; Ybarra et al., 2014) show that repetition can increase the harmfulness of bullying. While Olweus’ original definition in 1993 did include repetition as a core definitional feature of bullying, his more recent work has suggested that this component may need to be reconsidered and additional research is certainly required (Olweus, 2013). Some definition-based quantitative measures of bullying refer to intent to harm rather than harmful impact in their descriptions, noting that mean-spirited bullying is different from playful acts or joking statements that are friendly challenges (Felix et al., 2011; Malecki et al., 2015). Given these complications, we strongly recommend examining harmful impact directly, to avoid conflating bullying with trivial incidents of aggression while circumventing the challenges of establishing the perpetrator’s intent to harm (see above) and/or determining what specific threshold of repetition qualifies as bullying.

Longitudinal, quantitative research has further documented harmfulness, demonstrating prospective links between quantitative measures of victimization by bullying and outcome measures such as anxiety, depression, suicidal ideation and behavior, poorer physical health, and poorer academic functioning (e.g., Copeland et al., 2014; Wolke & Lereya, 2015). Thus, there is clear evidence that bullying causes harm to victims (including bully-victims). Quantitative data suggest that the costs to bullies appear to be primarily related to risky/antisocial behavior (Wolke & Lereya, 2015) alongside reductions in likeability, but not popularity (Pronk et al., 2017). Clearly then, quantitative methods have provided abundantly clear evidence that bullying is indeed harmful in a variety of ways.

However, key questions remain unanswered by quantitative methods, including the costs of bullying noted above (e.g., school and parental discipline and/or legal charges). Furthermore, qualitative methods can demonstrate how children and adolescents can differentiate between a bona fide act of bullying that is harmful and playful teasing or rough and tumble play that is relatively harmless and perhaps even enjoyable to the participants. This is particularly important given qualitative data showing a tendency (noted above) for bullies to frequently minimize or overlook the harm caused by their behavior (Gamliel et al., 2003; Gumpel et al., 2014; Søndergaard, 2012; Strindberg et al., 2020). Addressing the question of bullying versus playful teasing generally appears to require more in-depth, nuanced information than quantitative methods can yield. For example, it would be useful to know whether the perceived harmfulness of an act that could be viewed as bullying depends upon the content or nature of the act itself, the identify or status of the perpetrator (e.g., friend vs. enemy; popular vs. unpopular), the presence and nature of bystanders (e.g., private exchange vs, public humiliation), the presence of bullying reinforcers, the interventions of adults, or the power balance between the perpetrator or target (e.g., Gumpel et al., 2014).

Qualitative methods may help to fill in the gaps in our knowledge about the harmfulness of bullying, and they may provide an opportunity for victims of bullying to share their stories and perspectives so that researchers, educators, and practitioners can offer more effective assistance and mount better preventative interventions. A variety of qualitative evidence shows that adolescents appear to place a high value on victim harm as a proxy for intentionality in order to differentiate bullying from playful or accidental acts of teasing or rough housing (e.g., Hellström et al., 2015; Jeffrey & Stuart, 2019). Qualitative research in the cyber context confirms that youth focus more on harmful impact than the intent of the perpetrator to differentiate cyberbullying from joking, in large part because the potential ambiguity of online posts and statements (Baas et al., 2013; Dennehy et al., 2020; Jacobs et al., 2015; Mishna et al., 2009). Furthermore, youth in focus groups note that aggressive (versus playful) intent can increase the harm victims perceive from ambiguous online content (Pelfrey Jr. & Weber, 2014). On the other hand, while adolescents understand that one-time online posts can have long-lasting harmful effects, several qualitative researchers question whether youths fully appreciate the long-term consequences of hurtful material that circulates online (Baas et al., 2013; Dennehy et al., 2020; Pelfrey Jr. & Weber, 2014). The ability of adults to appreciate the long-term impact of pejorative online materials on future jobs, relationships, or wellbeing may represent one of the few areas where adults have a better concept of harm than adolescents.

Generally though, qualitative methods, including the use of interviews, can contribute to the assessment of harmful impact, by giving researchers access to the perspectives and experiences of victims of bullying (Mishna, 2004). Semi-structured interviews with adolescents would be a useful method for understanding victim perceptions of harm done by bullying. In the context of qualitative methods, a one-on-one environment may encourage honest sharing of experiences and feelings more than interviews with a group of participants (e.g., in a focus group). A key consideration for conducting these types of interviews, particularly with victims (for whom the interviews necessarily raise unpleasant thoughts and emotions), is that the researcher fosters an open environment and develops rapport with the participant (Mack et al., 2009). Ethnographic research may also be a useful mechanism with which to develop rapport. Ethnographies provide rich social and contextual data, as the researcher is able to build a rapport with the adolescents and gain insight into their relations through a combination of prolonged observations and interviews (Gumpel et al., 2014).

Another important aspect of harmful impact is capturing harmful outcomes beyond the mental and physical health or scholastic outcomes that are typically considered in quantitative research (Volk et al., 2006). For example, Nishina and Bellmore (2010) found that being victimized was associated with physical symptoms, maladjustment, and poorer school functioning. However, this type of quantitative research sheds little light on how harm might impact the daily lives of adolescents. Quantitative surveys can illuminate associations between victimization and negative outcomes; however, they are generally retrospective in nature and ask the participant to report how often various behavior has been done to them. Qualitative methods may allow researchers to gain a more comprehensive understanding of how victimization affects adolescents on a daily basis. Daily diaries may be particularly useful in this regard, as adolescents report their daily experiences with, and feelings about, peer victimization, so their perceptions are captured in a timely fashion rather than well after the fact, perhaps creating a more accurate record. In addition to providing a richer picture in general, information from diaries offers a nuanced view of how negative thoughts and feelings align with fluctuations in victimization, which would not be captured by quantitative measures (e.g., Livingston et al., 2019; Nishina & Juvonen, 2005).

Qualitative data have captured negative outcomes of victimization seldom considered in quantitative research, such as having to change schools (Shute et al., 2008), losing out on scholarships or prizes (Flanagan, 2007), performing worse or withdrawing from sports, having fewer opportunities for friendships (Wójcik & Mondry, 2020), or even losing the food necessary for survival (Turnbull, 1987). While all of these outcomes could themselves be associated with mental/physical health or school outcomes, the unique nature of each harmful impact may be important for determining the best ways to support victims. We know from clinical data that bullying is associated with different kinds of anxiety disorders that in turn require different therapeutic approaches (McCabe et al., 2003). Qualitative data on harmful impact can therefore serve as an important guide towards best practices for designing interventions that adequately support the specific needs of victims.

Limitations of Qualitative Methods

Although they offer important benefits to understanding key features of adolescent bullying, qualitative methods are not without their limitations. To begin with, the most salient limitation is how costly they are in terms of time and analyses. Time in particular is a major constraint as they tend to involve a much greater investment of participant time, or in the case of interviews and focus groups, can require a substantial time commitment of the researcher to transcribe and code the data (e.g., Queirós et al., 2017). This increases the difficulty of obtaining large sample sizes using qualitative research and that in turn impacts the reliability and generalizability of qualitative methods (e.g., Leung, 2015). However, while often not generalizable, qualitative studies can provide insight into theories and hypotheses to be explored in future qualitative and quantitative studies (Patton et al., 2017).

Qualitative methods also tend to be susceptible to both researcher and participant bias. The latter is most commonly seen when issues of social desirability emerge (Schonfeld & Mazzola, 2013), and this can be a problem for bullying research. Bullies may deny being aggressive while victims may exaggerate power imbalances to gain sympathy (DeLara, 2012; Patton et al., 2017). While these issues exist for quantitative data as well, the greater use of anonymity in quantitative data offers some protection from these biases. Similar methods can, and should, be employed with qualitative methods where possible.

A more challenging problem for bias is that many qualitative methods rely on the judgement, experience, perceptions, and/or knowledge of the researcher coding the data (e.g., Schonfeld & Mazzola, 2013). While the same may be generally said about quantitative data, qualitative data may present a superficially less imposing challenge that can lead to a false sense of grasping the necessary nuances that require significant training, practice, and knowledge to properly employ. To this end, numerous protocols have been developed to address potential biases and methodological pitfalls. First, researchers should be well trained in creating effective open-ended questions for surveys or (group) interviews (Sofaer, 2002, p. 334). Conducting interviews and focus groups requires substantial training and consideration should be given to each individual sample and the research question at hand (Chenail, 2011). To be blunt, while we recognize that quantitative methodologies are increasingly sophisticated and require dedicated training to properly deploy and interpret, the same degree of dedicated methodological training and awareness must be recognized and employed for qualitative methods that have progressed vastly beyond simple verbal interviews or casual impressions. For example, qualitative research may have nuanced ethical issues due to the frequently intimate nature of the data precluding anonymity and/or revealing sensitive information with the potential for legal liability. For instance, youths may report serious harm such as suicidal ideation. These situations are generally precluded by quantitative methods that specify and limit the nature of responses, whereas more open-ended qualitative methods create known and novel ethical considerations for qualitative researchers who may need to make extra judgements regarding reporting, labeling, offering supports, and intervening to prevent future harm (e.g., Sercombe & Donnelly, 2013). It is also worth noting that qualitative methods are generally more effective at capturing rare or minority opinions (Patton, 1999) that get averaged out in quantitative data.

While sample size and bias can largely be mitigated by careful research design, perhaps the most fundamental limit to qualitative data is their subjective nature. If an adolescent says they feel really anxious, how anxious is that? If they believe bullying is a big problem at their school, how big is big? Quantitative data allow for the use of a universal mathematical language that is both precise and without bias (Queirós et al., 2017). The downside is that the “grammar” of math is not intuitive and thus attempts to translate known or lived experiences into math can result in clumsy and/or inaccurate communication (e.g., Pimm, 1987). The need for these sorts of a priori translations (e.g., creating a numerical scale for power) necessarily involves the greater use of top-down theory in quantitative research, offering a potential advantage to qualitative research in areas where clarity is still lacking (e.g., many aspects of bullying).

Theoretical and Practical Implications

In summary, our recommendation is similar to what has been said previously about studying bulling (Volk et al., 2017). The best practices are likely those that employ a mixture of data sources (e.g., peers and parents), a mix of analytical methods (e.g., peer networks of relationships and diary entries about relationships), and a healthy balance between firmly adopting and testing theoretical models while still maintaining an open mind towards novel bottom-up trends in data. Triangulation in qualitative research is the combining of methodologies to comprehensively study a phenomenon (Patton, 1999). For example, researchers might engage in a direct qualitative method such as an interview, as well as an indirect, non-reactive method such as analyzing physical materials such as photos or diary entries (Flick, 2004). This multi-pronged approach can easily adapt to mixed-methods. What is more, qualitative research might also adopt designs typically used in quantitative research, such as using longitudinal qualitative designs that can uncover qualitative changes in experiences and relationships over time (Patton et al., 2017). Whereas in this paper, we have discussed qualitative studies or the need for further qualitative research with regards to the three definitional criteria of bullying (Volk et al., 2014), future work should conduct a thorough thematic review of existing qualitative research relevant to power imbalances, goal directedness, and harmful impact, to continue to further our understanding of the complexities of bullying.

Bullying is a complex problem that has proven challenging to prevent within and beyond academia. The use of qualitative research affords vital and novel information for both research and applied uses. Having a strong definition is necessary to both unify research and to create effective interventions that both reduce bullying behavior and address the needs of victims. Given the current imbalance favoring quantitative methods in the study of bullying, we suggest that it may be a good time for researchers to consider adding qualitative data to their research programs and intervention paradigms to better answer the surprisingly challenging and complex questions of what bullying actually is and how we should best address it.